A Subtilisin-like Protein in Secretory Organelles of Plasmodium falciparum Merozoites*

In the vertebrate host, the malaria parasite invades and replicates asexually within circulating erythrocytes. Parasite proteolytic enzymes play an essential but poorly understood role in erythrocyte invasion. We have identified a Plasmodium falciparum gene, denoted pfsub-1, encoding a member of the subtilisin-like serine protease family (subtilases). Thepfsub-1 gene is expressed in asexual blood stages ofP. falciparum, and the primary gene product (PfSUB-1) undergoes post-translational processing during secretory transport in a manner consistent with its being converted to a mature, enzymatically active form, as documented for other subtilases. In the invasive merozoite, the putative mature protease (p47) is concentrated in dense granules, which are secretory organelles located toward the apical end of the merozoite. At some point following merozoite release and completion of erythrocyte invasion, p47 is secreted from the parasite in a truncated, soluble form. The subcellular location and timing of secretion of p47 suggest that it is likely to play a role in erythrocyte invasion. PfSUB-1 is a new potential target for antimalarial drug development.

Plasmodium falciparum, the causative agent of the most severe form of human malaria, is an obligate intracellular apicomplexan parasite. The life cycle of the organism includes a number of specialized invasive (zoite) stages. In the vertebrate host, replication of the parasite in circulating erythrocytes is initiated when the cells are invaded by merozoites. The parasite replicates asexually within the infected erythrocyte to produce a number of progeny merozoites. Upon rupture of the host cell, these are released to invade fresh erythrocytes and perpetuate the blood stage cycle. Erythrocyte invasion by the malaria merozoite has been the subject of intensive study, since intervention strategies that prevent invasion would effectively block both replication of the parasite and the associated clinical disease.
Electron microscopic studies have shown that erythrocyte invasion by the malaria merozoite takes place in a number of discrete stages. Initial reversible attachment of the parasite to the red cell surface is rapidly followed by reorientation, the formation of an irreversible junction between the apical prominence of the merozoite and the host cell surface, and finally entry of the parasite into the cell by a mechanism resembling a form of induced endocytosis (1)(2)(3)(4). The process is facilitated by the controlled release of the contents of three types of secretory organelles, called rhoptries, micronemes, and dense granules, situated at or toward the apical domain of the merozoite (2,5,6). There is extensive evidence indicating an essential role for parasite-derived proteases in invasion. Invasion by P. falciparum merozoites is blocked in the presence of the serine protease inhibitor phenylmethylsulfonyl fluoride (PMSF) 1 (7), and invasion by merozoites of a number of Plasmodium species is prevented by chymostatin (8 -13). The inhibitory effect of chymostatin on invasion can be reversed by pretreatment of target erythrocytes with chymotrypsin (10), suggesting that the chymostatin-sensitive step in invasion involves an essential, parasite-induced proteolytic modification of the red cell surface (10 -12). A glycosylphosphatidylinositol (GPI)-anchored malarial serine protease activity has been described that may be involved in this modification (14). Treatment of isolated, invasive merozoites of the simian malaria P. knowlesi with N-tosyl phenylalanylchloromethyl ketone or N-tosyl lysylchloromethyl ketone prevents primary attachment of the parasites to host cells, whereas chymostatin blocks a later stage in the invasion pathway, indicating that more than one distinct protease activity may be involved (15). Consistent with this, a P. falciparum serine protease activity that mediates an essential processing and shedding of a major merozoite surface protein (merozoite surface protein-1; MSP-1) at invasion is highly sensitive to inhibition by PMSF but not by chymostatin (16,17). Parasite proteases involved in invasion are attractive potential targets for new rational approaches to antimalarial chemotherapy.
Here we report the identification of a novel, single copy P. falciparum gene (denoted pfsub-1) encoding a member of the subtilisin-like serine protease superfamily (subtilases). The primary gene product is expressed in the latter stages of intracellular merozoite maturation, and is post-translationally modified during secretory transport in a manner consistent with it being processed to form a mature, enzymatically active product. The putative mature protease is concentrated in a subset of dense granules within the apical domain of free merozoites and then is released in a soluble form during erythrocyte invasion, suggesting that it may play a role in invasion. This is the first molecular characterization of a putative apicomplexan serine protease.

EXPERIMENTAL PROCEDURES
Parasites, Materials, and Nucleic Acid Manipulations-Blood stage cultures of the T9/96 and FCB-1 lines of P. falciparum were maintained in vitro in human Aϩ erythrocytes and synchronized when required, as described previously (18,19). Protease inhibitors and fluorogenic or chromogenic peptide and protein substrates were obtained from Sigma and Boehringer Mannheim. Restriction endonucleases and other DNA modifying enzymes were obtained from Boehringer, Amersham Pharmacia Biotech, and New England Biolabs. The majority of the methods used for DNA manipulations were based on those of Sambrook et al. (20). Preparation of malaria parasite genomic DNA and analysis by Southern blotting was by standard procedures (20,21). DNA probes for Southern analysis and cDNA library screening were radiolabeled with [␣-32 P]dATP (220 TBq mmol Ϫ1 , Amersham Pharmacia Biotech) by the random primer method (Prime-It TM II random primer labeling kit, Stratagene). DNase-treated total parasite RNA from asynchronous blood stage cultures was prepared from saponin-treated parasites (18) using a S.N.A.P. TM total RNA isolation kit (Invitrogen), following the manufacturers' recommendations. Oligo(dT)-primed single-stranded cDNA synthesis from total parasite RNA was performed using a cDNA Cycle ® kit (Invitrogen) as directed by the manufacturers; "mock cDNA" preparations were prepared by subjecting similar RNA samples to identical processing except that no reverse transcriptase was present. All DNA sequencing was by the dideoxynucleotide (Sequenase) method using custom synthesized primers. Homology searches and alignments were performed with the Basic Local Alignment Search Tool (BLAST) program of the National Center for Biotechnology Information, National Institutes for Health (22), using the European Bioinformatics Institute server. 2 Polymerase Chain Reaction (PCR)-Most PCR was carried out using AmpliTaq ® DNA polymerase (Perkin-Elmer). Degenerate oligonucleotide primers for amplification by PCR and reverse transcription-PCR of malarial subtilase genes were SUB-1 (sense; 5Ј-CAYGGIACICAY-GTIGCIGG-3Ј) and SUB-2 (antisense; 5Ј-CCIGCIACRTGIGGIGTIGC-CAT-3Ј), based on the amino acid sequences HGTHVAG and MATPH-VAG, respectively. Primers for amplification by inverse PCR (23) of genomic sequences flanking the T9/96 pfsub-1 gene were SUB-SEQ11 (sense; 5Ј-TGCATGGGCAGGATATGCAG-3Ј) and SUB-SEQ12 (antisense; 5Ј-ATGGGAACCTCAGTATCCGG-3Ј). For inverse PCR, 5 g of T9/96 genomic DNA in a total volume of 20 l was digested for 5 h with 20 units of Sau3A. The restricted DNA was ethanol-precipitated, redissolved at a concentration of 2.5 g ml Ϫ1 , and ligated overnight with T4 DNA ligase. The circularized DNA was then used as template for PCR amplification with SUB-SEQ11 and SUB-SEQ12. A major, ϳ1.4-kilobase pair PCR product was gel-purified and cloned into the T-cloning vector pMOSBlue (Amersham Pharmacia Biotech) to create plasmid PM494-B. This and all other cloned AmpliTaq ® -generated PCR products were maintained in the MOSBlue Escherichia coli strain (Amersham Pharmacia Biotech). PCR amplification of regions of the pfsub-1 gene for subcloning into expression vectors was carried out with Pfu polymerase (Stratagene). Primers used for this purpose were SUB-EX1 (sense; 5Ј-GAGCTCGGATCCGAAGGAAGTAAGGTCTGAAGAA-3Ј), SUB-EX2 (sense; 5Ј-GAGCTCGGATCCGAGTAGACCAGGTAAATAT-CATTTC-3Ј), SUB-EX5 (sense; 5Ј-GAGCTCGAATTCAGTAGACCAGG-TAAATATCATTTC-3Ј) and SUB-EX3 (antisense; 5Ј-AGCCTCGAATT-CCTAGTTAATGCAAATATCTAC-3Ј). DNA fragments amplified with these primers were verified by sequencing on both strands.
Production and Screening of a P. falciparum Blood Stage cDNA Library-Poly(A ϩ ) RNA was isolated from total T9/96 parasite RNA using oligo(dT) 12-18 -cellulose (Amersham Pharmacia Biotech) column chromatography, using two cycles of binding and elution according to the manufacturer's instructions. Two g of poly(Aϩ) RNA was reverse transcribed into cDNA using 0.5 g of a random unidirectional XhoI linker primer (Stratagene) according to the method of Gubler and Hoffman (24). First and second strand synthesis was monitored by incorporation of [␣-32 P]dCTP (220 Tbq mmol Ϫ1 ; Amersham Pharmacia Biotech) and subsequent agarose-formaldehyde electrophoresis, followed by autoradiography (20). The cDNA was purified by spun column chromatography (Sephacryl S-300, Amersham Pharmacia Biotech), ligated to EcoRI/NotI cohesive end adaptors (Amersham Pharmacia Biotech), and treated with T4 polynucleotide kinase according to standard procedures (20). Following phenol extraction and ethanol precipitation, the cDNA was digested with 100 units of XhoI and purified using Sepharose CL-4B spun column chromatography (SizeSep 400 TM , Amersham Pharmacia Biotech). The cDNA was ligated into EcoRI-XhoIdigested -ZAP II vector (Stratagene) and then packaged (Gigapack II Gold, Stratagene), and the packaged phage was used to infect E. coli SURE TM host cells (Stratagene). Screening, selection, and in vivo rescue of pBluescript ® phagemid from plaque-purified phage by co-infection with Exassist TM helper phage was according to the manufacturer's instructions (Stratagene).
RNase Protection Assay-Recombinant pMOSBlue plasmid PM494-B, containing the ϳ1.4-kilobase pair DNA fragment obtained by inverse PCR amplification from genomic T9/96 DNA with oligonucleotides SUB-SEQ11 and SUB-SEQ12, was linearized with AvrII and used as a template for production of an [␣-32 P]UTP-radiolabeled, 422-nucleotide antisense RNA probe using the T7 promoter region of the plasmid (20). Protection of this probe from RNase activity following hybridization to blood stage parasite mRNA was assessed using an RPA II TM Ribonuclease Protection assay kit (Ambion), following the manufacturer's recommendations. Briefly, aliquots of 2 ϫ 10 4 cpm of gel-purified probe were hybridized overnight at 44°C with varying amounts (up to 12 g) of DNase-treated total parasite or yeast control RNA. The reactions were then treated with an empirically determined dilution of a mixture of RNase A and RNase T1 for 30 min at 30°C, followed by RNase inactivation and precipitation of the remaining RNA. Samples were subjected to electrophoresis on 6% acrylamide gels in the presence of 8 M urea and analyzed by autoradiography. Markers for electrophoresis were end-radiolabeled HinfI-digested dephosphorylated ØX174 DNA markers (Promega).
Expression, Metabolic Radiolabeling, and Purification of PfSUB-1 Recombinant Fusion Proteins and Antibody Production-DNA fragments amplified by PCR with primers SUB-EX1 and SUB-EX3, SUB-EX2 and SUB-EX3, or SUB-EX5 and SUB-EX3, were digested with BamHI and/or EcoRI and then ligated into either BamHI-and EcoRIdigested pTrcHisB (Invitrogen), or EcoRI-digested pGEX-1T (Amersham Pharmacia Biotech). E. coli strain DH5␣ (Stratagene) was transformed to ampicillin resistance, and recombinant clones were selected. In each case, the identity and orientation of the insert in recombinant clones was confirmed by complete sequence analysis on both strands.
Recombinant E. coli clone PM500 -7A, containing the SUB-EX2/ SUB-EX3 PCR product inserted into the expression vector pTrcHisB, was induced with isopropyl-1-thio-␤-D-galactopyranoside, and the cells were extracted by sonication in 20 mM sodium phosphate, pH 7.4, 0.5 M NaCl, 10 mM imidazole (starting buffer). The clarified extract was run on a HiTrap ® chelating column (Amersham Pharmacia Biotech), and the bound fusion protein (His 6 -PfSUB-1 m ) was eluted using the starting buffer supplemented with 8 M urea and 400 mM imidazole. For metabolic radiolabeling of His 6 -PfSUB-1 m , bacteria were grown to an A 600 of 0.7 in M9 minimal medium and then induced for 2 h with 1 mM isopropyl-1-thio-␤-D-galactopyranoside in the presence of 20 Ci ml Ϫ1 [ 35 S]methionine/cysteine (Pro-mix TM , Amersham Pharmacia Biotech).
Recombinant E. coli clone PM511-3A, harboring the SUB-EX5/SUB-EX3 PCR product inserted into the expression vector pGEX-1T, was induced with isopropyl-1-thio-␤-D-galactopyranoside. Cells were resuspended in 25 mM Tris-HCl, pH 8.2, 1 mM EDTA, 0.2% (v/v) Triton X-100 containing 1 mM PMSF and then sonicated. The inclusion bodies were pelleted by centrifugation at 12,000 ϫ g for 20 min and then washed twice more in the same buffer. Purification of the fusion protein (denoted GST-PfSUB-1 m ) by affinity chromatography on a column of glutathione-agarose then proceeded following the sarkosyl method of Frangioni and Neel (25). Fusion protein was eluted from the glutathioneagarose with 8 M urea in 50 mM Tris-HCl, pH 8.2, and then further purified by gel filtration in the same buffer on a 2.6 ϫ 77-cm column of Sephacryl S-200 HR (Amersham Pharmacia Biotech). The protein was finally precipitated by dialysis against PBS and used to immunize Balb/c mice using conventional procedures (26).
riched to a parasitemia of 85-90%, and naturally released merozoites were purified by filtration through 3-m and 1.2-m pore size acrylic membrane filters as described previously (19). Some merozoites were resuspended in fetal calf serum and used to prepare thin films, which were air-dried and stored desiccated at Ϫ70°C. Merozoite preparations were consistently free of schizont contamination, as determined by microscopic analysis of Giemsa-stained samples. For labeling with tritiated diisopropyl fluorophosphate ([ 3 H]DFP), 10 l (10 Ci) of [ 3 H]DFP (NEN Life Science Products) in propylene glycol was added with mixing to approximately 5 ϫ 10 9 merozoites in 200 l of PBS. Following incubation for 45 min at 37°C, the merozoites were pelleted, washed three times in PBS, and analyzed directly by SDS-PAGE or by immunoprecipitation with anti-PfSUB-1 m antibodies.
When required, mature schizonts enriched from highly synchronous cultures were metabolically radiolabeled for 2 h with 100 Ci ml Ϫ1 [ 3 S]methionine/cysteine as described previously (16). For analyzing the effects of brefeldin A (BFA; Boehringer Mannheim) on PfSUB-1 processing during secretory transport, the drug was prepared as a stock solution in methanol at 1 mg ml Ϫ1 and then added to schizont cultures to a final concentration of 5 g ml Ϫ1 15 min before the addition of [ 35 S]methionine/cysteine. Control cultures were similarly pretreated with methanol only. Following a 2-h incubation at 37°C, labeled parasites were washed twice in PBS and frozen at -70°C.
In some instances, radiolabeled schizonts were washed, placed back into culture in complete medium, and allowed to undergo merozoite release in the presence of fresh erythrocytes as described previously (16). After a reinvasion period of 4 -7 h, the culture supernatants were retained, and the cells were centrifuged over a cushion of 67.5% (v/v) isotonic Percoll as described previously (18). The cell pellet was resuspended in serum-free medium and layered onto a fresh Percoll cushion to repeat the treatment. The resulting pellet, containing only uninfected red cells and young ring-stage parasites, was analyzed by immunoprecipitation in parallel with samples of the radiolabeled starting schizonts and the reinvasion culture supernatants.
Pulse-Chase Experiments-Mature T9/96 schizonts enriched from highly synchronous cultures were washed twice in warm methionine/ cysteine-free medium, then cultured for 30 min in the same medium. The culture was then pelleted, resuspended to a hematocrit of 20% in 900 l of fresh warm methionine/cysteine-free medium, and labeling was initiated by adding [ 35 S]methionine/cysteine to 300 Ci ml Ϫ1 . After a 9-min incubation at 37°C, the chase was initiated by adding 100 l of a nonradioactive methionine and cysteine stock solution in RPMI to a final concentration of 3 and 1 mM, respectively. The culture was then immediately pelleted and resuspended in fresh complete medium at 37°C supplemented with methionine and cysteine to 3 and 1 mM, respectively, and two 1-ml "zero time" samples were removed to chilled 1.5-ml Eppendorf tubes containing 10 l of 10% (w/v) sodium azide. The sampled parasites were rapidly pelleted by centrifugation, washed once in ice-cold PBS containing 0.1% sodium azide, and then snap-frozen in a dry ice-ethanol bath. Samples taken from the bulk culture at 5, 15, 30, and 60 min following commencement of the chase were similarly processed.
Western Blotting, Immunoprecipitation, and Peptide Mapping-Analysis of proteins by SDS-PAGE and Western blotting was as described previously (19,27) except that in most cases antigen detection following incubation of blots with a horseradish peroxidase-conjugated second antibody was by ECL using a commercially obtained substrate (SuperSignal ® Substrate, Pierce). For immunoprecipitation analysis, radiolabeled parasite or bacterial pellets were solubilized into five volumes of a denaturing solubilization buffer (1% (w/w) SDS in 50 mM Tris-HCl, pH 8.0, 5 mM EDTA, 1 mM PMSF) with sonication. Samples were immediately boiled for 5 min and clarified by centrifugation at 12,000 ϫ g for 30 min, and then the clarified supernatants were diluted 10-fold into immunoprecipitation buffer (1% (v/v) Triton X-100 in 50 mM Tris-HCl, pH 8.2, 5 mM EGTA, 5 mM EDTA, 150 mM NaCl, supplemented with leupeptin, antipain, pepstatin A, and aprotinin, all at 10 g ml Ϫ1 , and N-tosyl lysylchloromethyl ketone at 10 M). Aliquots of the extract were then incubated overnight on ice with 2-8 l of anti-PfSUB-1 m antiserum. Immune complexes were bound to protein G-Sepharose 4 Fast Flow (Amersham Pharmacia Biotech) and then washed twice with wash buffer I (0.5% (v/v) Triton X-100 in 50 mM Tris-HCl, pH 8.2, 5 mM EDTA, 1 mg ml Ϫ1 bovine serum albumin, 0.5 M NaCl), and four times in wash buffer II (wash buffer I minus bovine serum albumin or NaCl), before elution into SDS sample buffer. Immunoprecipitates were subjected to SDS-PAGE in the presence or absence of 100 mM dithiothreitol as a reducing agent, and radiolabeled proteins were detected by autoradiography or fluorography.
For two-dimensional peptide mapping of metabolically radiolabeled, immunoprecipitated proteins, autoradiography was first used to determine the position of migration of polypeptides of interest. The bands were excised from the dried gels, and the gel fragments were rehydrated by incubation for 30 min in 1 ml of 0.2 M NH 4 HCO 3 . Following a further two 30-min incubations in 0.1 M NH 4 HCO 3 , 50% acetonitrile, the gel slices were rinsed briefly in 50% acetonitrile and lyophilized. The gel was then rehydrated by the addition of 100 l of a 0.1 mg ml Ϫ1 solution of salt-free trypsin (DPCC-treated, Sigma) or chymotrypsin (Boehringer Mannheim) in 25 mM NH 4 HCO 3 . The gel was allowed to swell on ice for 15 min, and then a further 100 l of 25 mM NH 4 HCO 3 was added and digestion was allowed to proceed overnight at 37°C. Supernatants were harvested, and digestion products were extracted from the gel by two successive incubations in 60% acetonitrile. Extracts were pooled, lyophilized, taken up in 20% aqueous pyridine, and spotted onto 0.1-mmthick cellulose TLC plates (Merck). The peptides were then separated by electrophoresis at pH 4.4 in pyridine/acetic acid/acetone/water (2:4: 15:79 by volume) (28), followed by ascending chromatography in the second dimension in butanol/acetic acid/water/pyridine (15:3:12:10 by volume). Dried plates were sprayed with EN 3 HANCE (DuPont), and radiolabeled peptides were detected by fluorography. Indirect Immunofluorescence (IFA), Confocal Microscopy, and Immunoelectron Microscopy-Analysis of acetone-fixed thin smears of malaria parasite cultures or purified merozoites by IFA and confocal microscopy was essentially as described previously (29). Briefly, samples were fixed in cold acetone and then probed with a primary mouse anti-PfSUB-1 m serum or a mouse antiserum against Schistosoma japonicum glutathione S-transferase (mouse anti-GST), typically at a dilution of 1:100, followed by a fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG (Seralab) or a Texas Red-conjugated goat anti-mouse IgG as second antibody, also diluted 1:100. Evans blue (0.01% (w/v) in PBS) was used as a counterstain, and 4,6-diamidino-2phenylindole was used at 1 g ml Ϫ1 in PBS as a nuclear stain. Direct conjugation of FITC to MSP-1-specific monoclonal antibody 2F10 (30) was as described previously (29). Slides were mounted in Citifluor (Citifluor UKC Chemical Laboratories, Canterbury, UK) and examined using a Zeiss III Photomicroscope fitted with an epifluorescence III RS condenser. For confocal analysis, samples were examined on a Bio-Rad MRC 600 confocal microscope using an argon-krypton laser for twochannel recording. Images were taken at 400-nm intervals along the z axis. For immunoelectron microscopy, preparations of mature segmented schizonts containing some free merozoites were fixed for 30 min at 4°C with 1% formaldehyde, 0.1% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4. Fixed samples were washed, dehydrated, and embedded in LR White resin as described by Aikawa and Atkinson (31). Thin sections were blocked in PBS containing 5% (w/v) nonfat milk powder and 0.01% (v/v) Tween 20. Grids were then incubated for 2 h at room temperature with primary antibodies (mouse anti-PfSUB-1 m serum, mouse anti-GST serum, or normal mouse serum) diluted in PBS containing 1% (w/v) bovine serum albumin and 0.01% (v/v) Tween 20 (PBT). After washing, grids were incubated for 1.5 h in 15-nm goldconjugated goat anti-mouse IgG (Amersham Pharmacia Biotech) diluted 1:20 in PBT, rinsed with PBT, and fixed with glutaraldehyde to stabilize the gold particles. Samples were finally stained with uranyl acetate and lead citrate and examined using a Zeiss CEM902 electron microscope.

PCR Amplification and Molecular Cloning of the pfsub-1
Gene-Oligonucleotide primers were designed based on conserved motifs flanking the catalytic histidine and serine residues of the subtilisin-like class of serine proteases. PCR primers SUB-1 (sense), and SUB-2 (antisense) were based on sequences His 170 to Gly 176 and Met 328 to Gly 335 of the Bacillus subtilis subtilisin E gene product (Genbank TM /EMBL accession no. K01988). Primers were completely redundant, no account being taken of the bias toward A ϩ T-rich codon usage in the P. falciparum genome. Inosine was used in positions of more than 2-fold degeneracy in order to minimize overall primer degeneracy and maximize promiscuity. Reverse primer SUB-2 was designed to take advantage of the nonredundant methionine codon at the 3Ј-end of the oligonucleotide. In PCR reactions using as template single-stranded cDNA prepared from DNAfree total blood stage parasite RNA, a single DNA fragment of approximately 560 bp was amplified. No product was obtained from mock cDNA preparations produced in the absence of re-verse transcriptase. A product of identical size was also obtained by PCR amplification from T9/96 or FCB-1 genomic DNA (not shown). Products from all three PCR reactions were cloned into the T-vector pMOSBlue, and 2-5 representative clones from each transformation were sequenced on both strands. All of the cloned sequences were identical. The cloned DNA was 561 bp long and contained a single open reading frame (ORF); a comparison of the deduced amino acid sequence with the protein data bases indicated strong homology with a number of bacterial subtilisins. The 561-bp fragment was then used to probe a T9/96 -ZAP asexual blood stage cDNA library, and a total of four strongly hybridizing clones were obtained (Fig. 1, top). Complete sequence analysis of the cDNA clones showed that they represented overlapping fragments of a sin-gle contiguous sequence containing an uninterrupted ORF of 2073 bp, encoding a protein of 690 amino acids with an estimated molecular mass of 77,874 daltons (Fig. 1, bottom). Consistent with the A ϩ T-rich P. falciparum genome, the A ϩ T content of the ORF is 72%.
The pfsub-1 Gene Encodes a Subtilase-The deduced pfsub-1 gene product (designated PfSUB-1) clearly belongs to the subtilisin-like (subtilase, S8) protease clan or superfamily as defined by Rawlings  protein (see below), the PfSUB-1 sequence exhibits a number of residues that are highly or completely conserved among all known subtilases (33,34); these include Asp 374 , His 430 , and Ser 608 (the catalytic site residues); the oxyanion hole residue Asn 522 ; and residues Gly 606 , Thr 607 , and Pro 612 . Furthermore, PfSUB-1 clearly does not belong to the proteinase K or lantibiotic peptidase families (subtilase families C and D; Ref. 34), both of which lack the 6-residue-long Ca1 Ca 2ϩ -binding loop sequence (Asn 443 -Val 448 in the deduced PfSUB-1 sequence) just C-terminal to the active site histidine residue. Additionally, PfSUB-1 clearly does not belong to the kexin family (family E; Ref. 34) of prohormone-processing proteases of yeasts and higher eukaryotes. One characteristic of this group, for example, is the DDG motif present at the active site Asp residue; PfSUB-1 possesses the more common DSG motif at this position (Asp 374 -Gly 376 of the PfSUB-1 sequence). It is thus unlikely that PfSUB-1, if expressed in the parasite as a catalytically active protease, has a dibasic cleavage specificity (35).
By homology with other subtilases, PfSUB-1 is likely to be synthesized as a pre-pro-protease; a putative signal peptide 25 residues long (the predicted prepeptide) is present at the amino terminus of the deduced protein sequence (Fig. 1, bottom), and the alignment referred to above would suggest that the mature protease sequence (denoted PfSUB-1 m ) probably extends from Ser 330 to the C-terminal His 690 residue, having a predicted M r of 40,432. If this is so, the pro-domain extends from Lys 26 to Arg 329 and has an estimated M r of 34,652. Consistent with this proposal, a hydropathy profile of the complete deduced sequence exhibits noticeable asymmetry, with a clear bias toward hydrophilic residues in the predicted pro-peptide sequence (not shown). This is a commonly observed characteristic of subtilase pro-domains (36). Other than the N-terminal signal peptide, no hydrophobic domain typical of a transmembrane or GPI signal sequence is present in the deduced PfSUB-1 sequence.
The pfsub-1 Gene Is Single Copy, Contains No Introns, and Is Transcribed in Blood Stages of the Parasite-Southern blot analysis of T9/96 genomic DNA was performed using as a probe the cloned 561-bp PCR product. Fig. 2A shows that following digestion of genomic DNA with any of six different restriction enzymes (none of which cleaves within the fragment used as a probe), the probe hybridized strongly to only one DNA fragment in each case. However, even under highly stringent washing conditions, weak hybridization was also reproducibly evident with at least one other DNA fragment in some digests. This result was interpreted as indicating that pfsub-1 is a single copy gene, but that the genome also contains at least one other locus with significant homology to the probe. The 561-bp pfsub-1 fragment was also used to probe blots of P. falciparum chromosomes separated by pulse-field gel electrophoresis (a kind gift of D. Williamson and P. Moore, Division of Parasitology, National Institute for Medical Research); the probe hybridized strongly to a single chromosome, further supporting the existence of a single genomic copy of the pfsub-1 gene (not shown).
A composite cDNA clone containing the complete pfsub-1 ORF was constructed by utilizing the single HindIII restriction site within the gene sequence and the pBluescript-derived EcoRI site at the 3Ј-end of each cloned partial cDNA insert shown in Fig. 1 (top). Partial cDNA clones 8-1 and 10-1 were restricted with HindIII and EcoRI. The 330-bp fragment thus released from clone 8-1 was replaced with the 700-bp HindIII/ EcoRI fragment released from clone 10-1. This resulted in plasmid PM481-1, which contains the complete pfsub-1 ORF plus 121 and 81 bp of noncoding sequence, respectively, up- A radiolabeled 422-nucleotide antisense RNA probe was produced using T7 RNA polymerase and the bacteriophage T7 promoter region of AvrII-linearized plasmid PM494-B. Equal amounts (2 ϫ 10 4 cpm) of the probe were hybridized with varying amounts of DNase I-treated total parasite RNA (sample RNA) isolated from asynchronous asexual blood stage parasites (lanes 4 -8 correspond, respectively, to hybridization with 12, 6, 3, 1.5, and 0.75 g of sample RNA) before the addition of an empirically determined amount of RNase. A major protected RNA fragment (arrow) of about 366 nucleotides was observed, the amount of protection being dependent of the amount of sample RNA present in each hybridization reaction. The bands of approximately 170, 175, and 240 nucleotides in lanes 4 -8 probably represent partially protected RNA species derived from internal nicking of the major 366-nucleotide protected fragment. This is a frequently observed feature of RNase protection experiments, particularly in the analysis of AU-rich sequences due to "breathing" (local denaturation) of the hybrid formed between the RNA probe and the target mRNA. No protection from RNase digestion was observed following hybridization of probe with 12 g of torulla yeast RNA (lane 3). Lanes 1 and 2 were loaded with 1 ϫ 10 3 cpm and 2 ϫ 10 2 cpm, respectively, of a sample of probe hybridized with 12 g of yeast RNA only, without subsequent RNase digestion. No protection was observed when sample parasite RNA was RNase-treated prior to incubation with the probe (not shown). stream of the start ATG and downstream of the TAA stop codon (Fig. 1). PCR amplification of the complete pfsub-1 ORF from PM481-1 or total cDNA with primers SUB-EX1 and SUB-EX3 produced a single DNA fragment of 2027 bp. Amplification from T9/96 genomic DNA under the same conditions produced a fragment of the same size, suggesting that the pfsub-1 gene contains no introns. Complete sequencing of the genome derived product confirmed this.
The pfsub-1 cDNA contains no Sau3A site, and the Southern blot data therefore indicated that the whole locus could be isolated on a single Sau3A genomic fragment of about 3.2 kilobase pairs (Fig. 2A, lane 4). This information was utilized in an inverse PCR approach to confirm the beginning and end of the pfsub-1 ORF and to ensure that no rearrangements had occurred in the regions immediately flanking the ORF during cDNA cloning and phagemid excision. Genomic T9/96 DNA was digested with Sau3A and religated under conditions designed to preferentially obtain intramolecular ligation. Two further PCR primers were then used to amplify a DNA fragment of ϳ1.4 kilobase pairs from the religated DNA. This was cloned and sequenced on both strands. The fragment was found to correspond to the expected inverse PCR DNA product, consisting of sequence including and flanking the ends of the pfsub-1 ORF (Fig. 1, top). The first Sau3A site upstream of the pfsub-1 gene lies precisely 389 bp 5Ј to the start ATG of the pfsub-1 ORF, and the distance from the TAA stop codon of the ORF to the first downstream Sau3A site is precisely 721 bp. The total length of the Sau3A fragment containing the pfsub-1 gene is therefore 3183 bp, corresponding well to the Southern blot data. No open reading frame other than the pfsub-1 ORF was identified within this stretch of DNA.
No RNA species hybridizing to the 561-bp probe was detectable in Northern blot analysis of poly(A ϩ ) T9/96 asexual blood stage mRNA (not shown). RNase protection assays, however, allowed detection of the presence of pfsub-1 transcripts in total RNA of asexual blood stage parasites. Plasmid PM494-B, containing the inverse PCR product, was linearized with AvrII and used as a template for production of a 422-nucleotide antisense RNA probe using the T7 promoter region of the plasmid. The probe therefore contained 56 nucleotides of plasmid-derived sequence, and a total of 366 nucleotides of sequence derived from the inverse PCR product insert, extending from the position of the SUB-SEQ12 oligonucleotide to a position 121 bp upstream of the first ATG of the pfsub-1 ORF (the AvrII site; see Fig. 1, top). Following hybridization of this probe with total RNA isolated from asynchronous asexual blood stage parasites and digestion with RNase, a major protected fragment of about 366 bases was reproducibly detectable by gel electrophoresis and autoradiography (Fig. 2B). This result confirmed the presence of the 5Ј-noncoding region of the pfsub-1 cDNA in mRNA.
Expression of PfSUB-1 in E. coli and Production of Monospecific Antibodies-pfsub-1 sequences encoding Lys 26 to the final His 690 residue (the putative proprotease sequence, denoted Pf-SUB-1 pm ), or Ser 330 to His 690 (the predicted mature protease domain, PfSUB-1 m ), were amplified by PCR and cloned into the expression vector pTrcHisB to produce constructs encoding fusion proteins with an N-terminal hexahistidine domain, separated from the insert sequence by an enterokinase cleavage site (Fig. 3). The region encompassing the cleavage site also contains an 8-amino acid residue epitope recognized by a commercially available monoclonal antibody (Anti-Xpress TM ). When clones PM500-6E and PM500-7A, containing the proprotease and mature protease sequences, respectively, of PfSUB-1, were assessed for fusion protein expression following induction by Western blot, using the Anti-Xpress TM monoclonal antibody, fusion proteins of the expected size were detected (not shown, but see Fig. 4); the calculated size of the fusion partner sequence encoded by these constructs is 3659 daltons, so the predicted M r of the proprotease PfSUB-1 fusion protein (denoted His 6 -PfSUB-1 pm ) is 78,707, and that of the mature PfSUB-1 fusion protein (denoted His 6 -PfSUB-1 m ; Fig. 3) is 44,073. The latter fusion protein was partially purified by metal chelate chromatography for use in experiments described below. DNA encoding Ser 330 to His 690 of pfsub-1 was similarly amplified and cloned into the EcoRI site of pGEX-1T to produce clone PM511-3A. The resulting fusion protein, denoted GST-Pf-SUB-1 m (Fig. 3), was purified to homogeneity and then used to raise anti-PfSUB-1 m antisera in mice; the sera were used to identify and characterize the parasite pfsub-1 gene product.
A number of approaches were taken to look for protease activity in both crude and purified preparations of the fusion proteins, including analysis on gelatin and azocasein SDS-PAGE (37) and examination for hydrolytic activity in solution or in agarose gels against denatured casein, azocasein, gelatin, and hemoglobin (38,39). Activity against the fluorogenic synthetic peptide substrates benzyloxycarbonyl-Gly-Met-Leu-AMC, benzyloxycarbonyl-Gly-Gly-Leu-AMC, and benzyloxycarbonyl-Ala-Ala-Leu-AMC was also investigated, as was activity against FITC-casein and FITC-bovine serum albumin (39). No protease activity was detected (data not shown).
Identification of pfsub-1 Gene Products in the Malaria Parasite- Fig. 4 shows that in both Western blot and immunoprecipitation analysis of P. falciparum merozoite or schizont extracts, the anti-PfSUB-1 m antibodies recognized only two polypeptide species; these migrated at 47 kDa (the major species) and 54 kDa (minor species) on SDS-PAGE under reducing conditions, and the mobility of both proteins was reductionsensitive. The proteins were denoted p47 and p54, respectively. Efficient immunoprecipitation of these proteins was obtained only after SDS extraction of parasite preparations; solubilization with TX-100 or sodium deoxycholate alone resulted in poor yields (not shown). To investigate the relationship between these two proteins and to seek further evidence that one or both proteins were products of the pfsub-1 gene, p47 and p54 were immunoprecipitated from extracts of biosynthetically radiolabeled T9/96 schizonts, digested with trypsin or chymotrypsin, and then analyzed by two-dimensional peptide mapping in parallel with similar digests of biosynthetically labeled recombinant fusion protein His 6 -PfSUB-1 m . Peptide maps derived from the parasite-derived proteins and the recombinant His 6 -PfSUB-1 m (Fig. 5) showed high relatedness but were not identical. Given the known differences between these polypeptides (i.e. the presence of approximately 3 kDa of nonmalarial sequence at the N terminus of His 6 -PfSUB-1 m and the difference in size between the malarial and recombinant proteins), as well as the possibility of differential post-translational modification of the malarial and E. coli-derived proteins, identity was not expected. The maps are, however, sufficiently similar to strongly suggest that p47 and p54 are products of the pfsub-1 gene. Further peptide mapping of radiolabeled in vitro translated pfsub-1 gene products has confirmed this (not shown). Peptide maps of p47 and p54 were virtually indistinguishable (Fig. 5). It was concluded that both proteins are pfsub-1 gene products, probably derived from differential post-translational processing of a full-length precursor protein.
PfSUB-1 Undergoes Post-translational Processing during Secretory Transport-This indication was confirmed by further immunoprecipitation analysis of parasites biosynthetically radiolabeled in the presence of the drug BFA and by pulse-chase experiments. Fig. 6 shows that radiolabeling in the presence of BFA, which blocks secretory transport of proteins from the endoplasmic reticulum (ER) to the Golgi apparatus, resulted in a complete absence of detectable labeled p47 and a clear relative increase in levels of labeled p54. This is consistent with p47 being derived from p54 via a proteolytic processing step taking place during or following transport from the ER to the Golgi; blockade of this step with BFA would be expected to result in an accumulation of p54 in the ER. Pulse-chase experiments further confirmed the derivation of p47 (Fig. 7). Highly synchronous cultures of mature T9/96 schizonts were metabolically pulse-radiolabeled with [ 35 S]methionine/cysteine and  4). Immunoprecipitates were subjected to SDS-PAGE under reducing conditions on a 10% gel. Lane 1 contains a sample of the total parasite extract, and the positions of molecular weight markers are shown. The positions of the two proteins (denoted p54 and p47) specifically precipitated by the anti-PfSUB-1 m antibodies are also indicated. B, PfSUB-1 is expressed in merozoites, and its migration on SDS-PAGE is reduction-sensitive. Lanes 1-5, proteins immunoprecipitated from extracts of metabolically radiolabeled T9/96 schizonts were subjected to SDS-PAGE on a 10% gel in the presence (ϩ) or absence (Ϫ) of 100 mM dithiothreitol as a reducing agent. Lanes 6 -11, purified, naturally released T9/96 merozoites (lanes 6 -10) or a sample of fusion protein His 6 -PfSUB-1 m (lane 11) was SDSsolubilized and subjected directly to SDS-PAGE on a 10% gel in the presence (ϩ) or absence (Ϫ) of dithiothreitol. The proteins were transferred electrophoretically to nitrocellulose, and the blot was probed with the anti-PfSUB-1 m antibodies. The migration of both p54 and p47 is reduction-sensitive, indicating the presence of intramolecular disulfide bonds. In addition, the His 6 -PfSUB-1 m fusion protein was strongly recognized by the anti-PfSUB-1 m antibodies. The positions of migration of molecular mass markers bovine serum albumin (68 kDa) and ovalbumin (44 kDa) are also shown.
FIG. 5. Two-dimensional peptide mapping indicates structural similarity between p54/p47 and a recombinant pfsub-1 gene product. Biosynthetically radiolabeled E. coli-derived recombinant His 6 -PfSUB-1 m or parasite-derived p54 and p47 proteins were purified by immunoprecipitation and then subjected to digestion with trypsin (T) or chymotrypsin (C). Digests were analyzed by two-dimensional thin layer chromatography using electrophoresis at pH 4.4 in the first dimension (A), followed by ascending chromatography in the second dimension (B). Labeled peptides were detected by fluorography. then chased in the presence of excess nonradioactive methionine and cysteine. Samples of the culture taken at 0, 5, 15, 30, and 60 min following commencement of the chase were again analyzed by immunoprecipitation with the anti-PfSUB-1 m antibodies. Fig. 7 shows that at zero chase time following the pulse labeling (which was extended for a 9-min period in order to obtain detectable amounts of labeled protein) a number of immunoreactive proteins were evident; as well as p54, a clear doublet was evident at about 60 and 61 kDa, in addition to a larger band at 82 kDa. By 15 min of chase, the 82-kDa species and the 60/61-kDa doublet had virtually disappeared, concomitant with an increase in the intensity of p54 and a just detectable appearance of p47. By 30 min of chase, all bands other than p54 and p47 had disappeared, and between 30 and 60 min of chase, clear conversion of p54 to p47 was evident. These results are consistent with the following processing scheme. PfSUB-1 is initially detected as a large precursor protein with an approximate molecular mass of 82 kDa. Since signal peptide cleavage is generally cotranslational, it is likely that this largest species detectable in the pulse-chase experiments represents the complete PfSUB-1 sequence minus the predicted 25residue-long signal sequence. This product is then rapidly processed (estimated t1 ⁄2 Ͻ 15 min) in an early secretory compartment, probably the ER, possibly via the 60/61-kDa forms, to p54, which is subsequently converted to p47 upon transport to the Golgi. Note that since the antibodies used in these analyses were raised against a fusion protein containing only the C-terminal 361 residues of the PfSUB-1 sequence (PfSUB-1 m ), only those proteins corresponding to, or significantly overlapping, this putative catalytic domain sequence would be de-tectable in these experiments. Species derived from the N-terminal portion of PfSUB-1 (such as, for example, the putative pro-domain) would not be detected.
The predicted molecular mass of the full-length pfsub-1 gene product (the preproprotease) is 77,874 Da, and that of the proprotease (i.e. following signal peptide cleavage) is 75,066 Da. Both of these figures are significantly less than the calculated molecular mass of the short lived 82-kDa parasite protein detected in the pulse-chase experiments. However, the apparent size on SDS-PAGE of the product of clone PM500 -6E, which contains the putative proprotease domain of PfSUB-1 in the form of a C-terminal fusion protein with a 3659-Da fusion partner was 85 kDa (not shown); it therefore appears that both native and recombinant forms of the protein exhibit slightly aberrant migration on SDS-PAGE and that the mobility of the parasite 82-kDa species detected in the pulse-chase experiments is consistent with it being the PfSUB-1 proprotease.
Immunolocalization of PfSUB-1 to a Subset of Dense Granules of the P. falciparum Merozoite-The above experiments showed convincingly that antibodies raised against the GST-PfSUB-1 m fusion protein recognize exclusively products of the pfsub-1 gene in merozoite or schizont extracts. The same antibodies were therefore used to investigate the subcellular localization of these proteins. T9/96 schizonts, or naturally released T9/96 or FCB-1 merozoites were acetone permeabilized and analyzed by IFA. The anti-PfSUB-1 m antibodies reacted exclusively and strongly with a highly localized domain within the apical tip of free merozoites (Fig. 8, top). Similar IFA analysis of mature segmented schizonts (not shown) produced a weak punctate fluorescence similar to that previously observed with antibodies reactive with components of secretory apical organelles of the merozoite. Absolutely no reactivity was observed with intracellular trophozoite or young "ring-stage" forms of the parasite (not shown). To more stringently define the location of PfSUB-1, the antibodies were used in immunoelectron microscopy examination of thin sections of schizonts and free merozoites. Discrete antibody reactivity was observed with circular, electron-dense organelles with the morphological FIG. 7. Pulse-chase analysis shows that p54 and p47 are derived from higher molecular weight precursors. Highly synchronous cultures of mature T9/96 schizonts were metabolically pulse-radiolabeled for 9 min with [ 35 S]methionine/cysteine. A chase was then initiated, and cells were harvested at the indicated times and analyzed by immunoprecipitation. The positions of p54 and p47 are indicated, and positions of larger precursor proteins labeled with arrows. The bands at about 45 and 42 kDa present in all lanes are likely to be non-PfSUB-1-derived proteins co-precipitated or cross-reactive with the anti-PfSUB-1 m serum used.

FIG. 6. p47 is derived from p54 via a brefeldin A-sensitive step.
Cultures containing mature T9/96 schizonts were resuspended in medium and then divided into two identical 2-ml cultures. To one culture was added 10 l of a 1 mg ml Ϫ1 solution of BFA in methanol. To the other culture was added 10 l of methanol only. The cultures were incubated at 37°C for 15 min, and then both were metabolically radiolabeled with [ 35 S]methionine/cysteine for a period of 2 h. The labeled parasites were then analyzed by immunoprecipitation using the anti-PfSUB-1 m antibodies (lanes 1 and 2) or an anti-GST serum as a control (lanes 3 and 4). In BFA-treated parasites, the appearance of labeled p47 is completely blocked, with a concomitant increase in levels of p54. The fluorograph is deliberately overexposed, and the arrows indicate the positions of minor labeled species immunoreactive with the anti-Pf-SUB-1 m but not the anti-GST serum. Positions of molecular weight markers are shown. characteristics of merozoite dense granules (Fig. 8, bottom). Interestingly, although these organelles are dispersed throughout the merozoite cytoplasm (2), immunoreactivity was observed only with a subset of granules situated toward the apical end of the parasite. This result was entirely consistent with the IFA data showing a predominantly apical reactivity. Since p47 is the major pfsub-1 gene product detectable in merozoite extracts by Western blot and appears to be the terminal processing product in the parasite, it is likely that this is the species that concentrates in merozoite dense granules.
The presence of p47 within intact secretory organelles of naturally released free merozoites suggested that PfSUB-1 is unlikely to be involved in merozoite release from schizonts but that it might play a functional role in either erythrocyte invasion or postinvasion events. A number of reports have shown that dense granule release in apicomplexan parasites occurs predominantly subsequent to, rather than during, invasion (e.g. Refs. 2, 6, and 41-43), and a number of defined dense granule components have been localized to the parasitophorous vacuole of the invaded host cell (e.g. Refs. 44 and 45). Information on the fate of p47 following invasion might provide clues as to the function of the protein. To explore this, metabolically radiolabeled, mature T9/96 schizonts were washed extensively and then either immediately snap-frozen (zero time schizonts) or recultured for 4 -6 h, with or without the addition of a 10-fold excess of fresh red blood cells, to allow merozoite release and invasion to take place. Following this, culture supernatants were harvested. The cell pellets from cultures to which additional erythrocytes had been added were then processed by centrifugation on Percoll to remove residual schizonts. The final preparations contained only uninfected erythrocytes and young ring-stage parasites. These samples were then analyzed by immunoprecipitation with the anti-PfSUB-1 m antibodies, in parallel with the culture supernatants and zero time schizont samples. Fig. 9 shows that whereas the zero time schizont extracts contained p54 and p47 as expected, only one major immunoreactive species of approximately 43 kDa was detectable in culture supernatants following merozoite release. No immunoreactive proteins were precipitated from ring-stage parasites (not shown). This is consistent with the absence of reactivity with the antibodies in IFA analysis of rings. It is tentatively concluded that p47 can be further processed and shed from merozoites in a truncated, soluble form. It is unclear whether this occurs quantitatively at invasion, but our inability to detect any PfSUB-1-derived proteins in the newly invaded erythrocyte may suggest that the protease is not carried into the invaded cell or else that it is rapidly relocalized and/or degraded following invasion. If p47 plays a functional role in erythrocyte invasion, anti-PfSUB-1 m antibodies might interfere with this and inhibit merozoite entry into erythrocytes in vitro. Accordingly, the mouse anti-PfSUB-1 m sera were tested for their ability to inhibit erythrocyte invasion in cultures containing actively invading, highly synchronized mature T9/96 schizonts (18). No invasion-inhibitory activity was observed (data not shown).
PfSUB-1 Does Not React with DFP-In attempts to further characterize PfSUB-1, we explored its reactivity with DFP. This compound inhibits a wide range of serine proteases by forming a covalent adduct between the nucleophilic oxygen atom of the active site serine residue and the diisopropyl phosphoryl function of DFP. Treatment of purified merozoites with [ 3 H]DFP resulted in the radiolabeling of three major species of approximately 35, 40, and 55 kDa, as well as two minor species of about 70 and 85 kDa (Fig. 10). This labeling profile was completely reproducible in a number of experiments (n ϭ 8) using different batches of intact or sonicated merozoites (not shown). Since DFP is known to be able to react with certain residues other than reactive serine residues (46), cold competition experiments using other well characterized active site serine affinity reagents were performed. Pretreatment of the merozoites with nonradioactive PMSF abolished [ 3 H]DFP reactivity with the 40-and 70-kDa proteins. Similarly, reactivity of [ 3 H]DFP with the 40-kDa moiety was completely abolished by pretreatment with another active site serine affinity reagent, 3,4-dichloroisocoumarin (47). It is therefore highly likely that at least the 40-and 70-kDa species are active site serine enzymes. However, none of the [ 3 H]DFP-reactive merozoite proteins migrated on SDS-PAGE at the position of any of the major PfSUB-1 gene products. Immunoprecipitation from extracts of [ 3 H]DFP-labeled merozoites with anti-PfSUB-1 m sera similarly failed to precipitate radiolabeled p47 or p54 (not shown). Thus, at least in the form in which it is resident in the merozoite, PfSUB-1 does not detectably react with DFP. DISCUSSION This report describes the first gene to be identified of any apicomplexan parasite that encodes a serine protease-like protein. We have not directly demonstrated any proteolytic activity associated with any pfsub-1 gene product, but a number of indications lead us to propose that p47 or its truncated, se-creted product is likely to be an enzymatically active protease and that it may play a role in erythrocyte invasion.
First, at the primary sequence level PfSUB-1 shows significant homology to known subtilases, and possesses all of the features known to be required of an active subtilase. Only four amino acid residues are completely conserved among known subtilases (34); these are the catalytic triad residues (PfSUB-1 residues Asp 374 , His 430 , and Ser 608 ) and a glycine residue (Gly 606 in the PfSUB-1 sequence) two positions N-terminal to the active site serine. PfSUB-1 also possesses an asparagine (Asn 522 ) at the position of the oxyanion hole residue; the only substitution acceptable at this position is that of an aspartate, found in the PC2 subfamily of kexin-like convertases (34). While many bacterial subtilisins are devoid of cysteine residues, an increasingly large number of subtilases are known to possess up to two intramolecular disulfide bonds within the catalytic domain (34). There are seven cysteine residues within the putative catalytic domain of PfSUB-1, and the reductionsensitive mobility on SDS-PAGE of p47 is consistent with the presence of intramolecular disulfide bonds. Detailed homologybased molecular modeling of the putative PfSUB-1 catalytic domain 3 supports this postulate. Thus, at the structural level, PfSUB-1 resembles a subtilase.
Second, the proteolytic processing to which PfSUB-1 is subjected during secretory transport probably represents a process of enzyme activation common among subtilases. In a number of well studied examples, co-translational signal peptide cleavage allows folding of the subtilase catalytic domain, mediated by the intramolecular chaperone activity of the pro-domain (36,48). This is often (but not always) rapidly followed by autocatalytic cleavage of the pro-domain from the catalytic domain; this cleavage can take place in cis (intramolecular) or in trans (intermolecular) and in some cases simple pro-domain cleavage is sufficient to allow zymogen activation (49,50). However, there is a growing body of evidence that subtilase activation is often a substantially more complex process, which in eu- karyotes may be intimately linked to correct routing of the proenzyme through the secretory pathway and final compartmentalization of the active enzyme (51)(52)(53)(54). Here we have shown that the primary pfsub-1 gene product is subjected to at least two major post-translational processing steps. The first of these, conversion of the 82-kDa form, possibly via a 60/61-kDa intermediate, to the p54 form, takes place very rapidly following synthesis and therefore could represent an autocatalytic processing step triggered by signal peptide cleavage and cotranslational folding within the lumen of the ER. The p54 form is then quantitatively converted to p47 in a slower, BFA-sensitive process. Further work is required to establish whether either or both of these processing events represent enzyme activation, but certainly they are wholly consistent with a putative activation pathway. If so, the process is clearly more complex than a simple one-step pro-domain removal. BFA sensitivity is a common feature of transport to secretory organelles in apicomplexan parasites (55)(56)(57)(58)(59), but little is known about the structural requirements for correct sorting to these organelles (59); it is possible that part of the PfSUB-1 sequence could provide a targeting signal, as has been determined for other subtilases and yeast carboxypeptidase Y (52,60).
Third, the terminal intracellular PfSUB-1 processing product, p47, localizes to a subpopulation of dense granule-like organelles within the extreme apical domain of the merozoite and appears to be shed in a truncated, soluble form following merozoite release but before completion of erythrocyte invasion; no PfSUB-1-derived proteins were detectable in the newly invaded erythrocyte. These observations are in apparent conflict with the majority of the available data suggesting that dense granule secretion in apicomplexan parasites is a postinvasion event (2,6,(41)(42)(43)(44). However, dense granules in Plasmodium are currently defined essentially on morphological criteria; there is a paucity of specific markers for these organelles, and indeed only two other malarial proteins, RESA/ Pf155 and a 14-kDa protein denoted RIMA, have been previously localized to merozoite dense granules (41,61,63). It is therefore possible that distinct subpopulations of dense granules may exist in Plasmodium; there is evidence for this in the apicomplexan parasite Cryptosporidium parvum (64), and if so, these subpopulations may be functionally distinct and may undergo exocytosis at different stages of the invasion process. Our evidence that PfSUB-1 is released at invasion suggests that it could play a role in the process of erythrocyte entry. Anti-PfSUB-1 m antibodies had no inhibitory effect on erythrocyte invasion by released merozoites in vitro, but it is unclear whether these antibodies have access to PfSUB-1 in the intact merozoite or whether they can interfere with its function. The demonstration of enzymatic activity in native, merozoite-derived p47 or its truncated soluble product would be a step toward addressing this issue. Consistent with observations of other workers (65), we have not been able to detect activity corresponding to p47 on gelatin substrate SDS-PAGE. 4 However, the enzyme may be irreversibly denatured by exposure to SDS, and the insolubility of p47 in nonionic detergents precludes ready isolation and analysis of the merozoite-derived protein in a native state. These limitations could be overcome by recombinant expression of enzymatically active PfSUB-1, and this is a major priority of current work.
If PfSUB-1 is involved in erythrocyte invasion, what precise role might the enzyme perform? The best characterized proteolytic event associated with erythrocyte invasion is the processing and shedding of a merozoite surface protein complex derived from the precursor protein MSP-1. A single proteolytic cleavage at a Leu-Asn motif within the membrane-bound component of this complex, known as secondary processing, releases the complex quantitatively from the surface of the merozoite as it enters the host erythrocyte (16 -18, 29). MSP-1 secondary processing is thought to be essential for successful erythrocyte invasion, and it is mediated by a parasite-derived, calcium-dependent serine protease (16,17). Interestingly, this activity is exquisitely sensitive to inhibition by PMSF, but it is only poorly inhibited by even very high concentrations (up to 10 mM) of DFP (19). Given its apparent nonreactivity with DFP, its location in the merozoite, and the timing of its secretion, PfSUB-1 is a good candidate for this activity. This intriguing possibility clearly merits further investigation and is supported by molecular modeling of the proposed catalytic domain of PfSUB-1, which has indicated that a heptapeptide corresponding to the sequence flanking the MSP-1 secondary processing site (i.e. Gln-Gly-Met-Leu-Asn-Ile-Ser) fits well into the active site groove of the model, with the Leu-Asn scissile bond in the appropriate position for cleavage. 3 The only other serine protease activity previously localized to P. falciparum merozoites is a DFP-reactive enzyme thought to be activated following cleavage of a GPI anchor (14); it is unlikely that PfSUB-1 corresponds to this activity, since it shows no DFP reactivity and the sequence exhibits no C-terminal hydrophobic domain typical of GPI signal sequences (66).
Of the two major families of active site serine endoproteases, the chymotrypsin-like and the subtilisin-like families (32), chymotrypsin-like proteases are common in higher eukaryotes but appear to be rare in lower eukaryotes and prokaryotic organisms. Sakanari et al. (67) have reported evidence for the existence of a chymotrypsin-like gene in Trypanosoma cruzi, but there is no other previous genetic data on protozoal serine proteases. Extensive attempts in this laboratory to isolate chymotrypsin-like P. falciparum genes using a number of different PCR-based approaches have been unsuccessful. 4 It is conceivable that, like Saccharomyces cerevisiae (40), the malaria parasite (and perhaps other apicomplexan parasites) does not possess genes of this class. Proteases play a crucial role in the life cycle of the blood stage malaria parasite and are undoubtedly potential targets for the design of protease inhibitor-based drugs. Expression of PfSUB-1 in an enzymatically active form should allow the development and evaluation of such inhibitors.