Chitinases of the avian malaria parasite Plasmodium gallinaceum, a class of enzymes necessary for parasite invasion of the mosquito midgut.

The Plasmodium ookinete produces chitinolytic activity that allows the parasite to penetrate the chitin-containing peritrophic matrix surrounding the blood meal in the mosquito midgut. Since the peritrophic matrix is a physical barrier that the parasite must cross to invade the mosquito, and the presence of allosamidin, a chitinase inhibitor, in a blood meal prevents the parasite from invading the midgut epithelium, chitinases (3.2.1.14) are potential targets of malaria parasite transmission-blocking interventions. We have purified a chitinase of the avian malaria parasite Plasmodium gallinaceum and cloned the gene, PgCHT1, encoding it. PgCHT1 encodes catalytic and substrate-binding sites characteristic of family 18 glycohydrolases. Expressed in Escherichia coli strain AD494 (DE3), recombinant PgCHT1 was found to hydrolyze polymeric chitin, native chitin oligosaccharides, and 4-methylumbelliferone derivatives of chitin oligosaccharides. Allosamidin inhibited recombinant PgCHT1 with an IC(50) of 7 microM and differentially inhibited two chromatographically separable P. gallinaceum ookinete-produced chitinase activities with IC(50) values of 7 and 12 microM, respectively. These two chitinase activities also had different pH activity profiles. These data suggest that the P. gallinaceum ookinete uses products of more than one chitinase gene to initiate mosquito midgut invasion.

karyotes and eukaryotes (1); their biologic roles include cell wall modification (e.g. fungi (2), Entamoebae (3) and filaria parasites (4), carbon source degradation (e.g. Streptomyces spp. (5,6), Serratia marcescens (7), and Vibrio spp. (8)) and plant and fungal host defense against chitin-containing pathogens (1). One other protozoan pathogen of man, Leishmania donovani, the agent of human visceral leishmaniasis, is known to use a chitinase in its life cycle (9,10). The Leishmania chitinase is thought to disrupt the sand fly cardiac valve, allowing amastigotes to be regurgitated from the midgut into the skin of the vertebrate host. The Leishmania chitinase is not thought to function in invasion of the arthropod vector per se (11). In contrast, Plasmodium chitinase is thought to be required for the parasite to invade the mosquito midgut after being taken up in a blood meal (12). Because of its critical biological function in the life cycle of the malaria parasite, the Plasmodium chitinase is a potential target for blocking transmission from the vertebrate host to the mosquito vector (12).
How the Plasmodium ookinete penetrates the chitin-containing peritrophic matrix to begin its invasion of the mosquito has been outlined. By using transmission electron microscopy, Sieber et al. (13) demonstrated that the motile ookinete, fully mature 20 -25 h after zygote formation, actively penetrates the peritrophic matrix (PM) 1 in the mosquito midgut en route to invading the midgut epithelium. Huber et al. (14) showed that the Plasmodium gallinaceum ookinete secretes chitinolytic activity. Native activity gel electrophoresis of ookinete cell extracts and culture supernatants showed two predominant bands of chitinase activity, but the precise molecular masses of the two forms could not be determined because active enzyme could not be obtained after SDS-PAGE. Shahabuddin et al. (12) showed that addition of an oligosaccharide chitinase antagonist, allosamidin, to a blood meal prevents oocyst formation in the mosquito, in both the Aedes aegypti-P. gallinaceum and the Anopheles freeborni Plasmodium falciparum model systems. This effect could be completely reversed by enzymatic degradation of the PM in vivo, by adding exogenous chitinase to the blood meal. These observations demonstrated that a chitinase is necessary for malaria parasites to invade the mosquito and initiate sporogonic development.
Because of intrinsic biologic interest and the potential for Plasmodium chitinases to be targets of interfering with malaria transmission (for a review of malaria transmission-blocking vaccines and the potential of Plasmodium chitinases as targets, see Refs. 15 and 16), we sought to characterize the malaria parasite chitinase genes. We began our studies with the chitinase of the avian malaria parasite, P. gallinaceum; this parasite is thought to be closely related to the lethal human malaria parasite, P. falciparum (17,18). Conveniently, large numbers of P. gallinaceum chitinase-producing ookinetes can be obtained in vitro, allowing for the direct study of the protein. In addition, P. gallinaceum serves as a useful model system for studying malaria parasite transmission (19,20).

EXPERIMENTAL PROCEDURES
Preparation of P. gallinaceum Ookinete Chitinase-The 8A strain of P. gallinaceum was used to infect 4 -6-week-old White Leghorn chickens. A gametocyte-producing line was maintained by subpassage in chickens and periodic passage through mosquitoes. Ookinetes were cultured from purified zygotes in serum-free and protease-free M199 culture medium as described previously (21). Preparations routinely yielded 5-15 ϫ 10 7 ookinetes per 5 chickens, with transformation efficiencies of 50 -90%.
Twenty four to 30 h cultures of ookinetes were centrifuged, and the pellet and supernatants were pooled separately and frozen at Ϫ20°C. Extracts of ookinetes were prepared by addition of 20 mM sodium phosphate, pH 6.8, to the ookinete cell pellet, usually without protease inhibitors, followed by vigorous vortexing, three cycles of freeze-thawing (dry ice to room temperature), and sonication (6 cycles for 20 s on ice). For the experiment in which the time course of chitinase expression was determined with Western immunoblotting, a mixture of protease inhibitors was added directly to the fresh cell pellet in 20 mM sodium phosphate, pH 6.8 (2 mM AEBSF, 5 mM EDTA, 200 M N-tosyl-L-phenylalanine chloromethyl ketone, 100 M tosyl-L-lysine chloromethyl ketone, 1 mg/ml pepstatin, 2 mg/ml leupeptin, 2 mg/ml aprotinin, 1% Triton X-100). Pooled supernatants were 200-fold concentrated by centrifugal ultrafiltration (Centriprep 10, Amicon, Beverly, MA), dialyzed against 20 mM sodium phosphate, pH 6.8, and frozen at Ϫ20°C until further use.
Protein Purification and Chitinase Detection-The high pressure liquid chromatography (HPLC) system consisted of a Thermoseparation Constametric Pump and a Spectromonitor 4100 UV detector (dual wavelength at 220 and 280 nm). The initial HPLC step consisted of 5.0 ml of pooled, dialyzed supernatants combined with soluble ookinete extracts injected into a quaternary ammonium anion exchange column (Q column) (Vydac 300VHP575. 0.75 ϫ 5 cm, Hesperia, CA). Buffer A was 20 mM Tris, pH 8.0. Buffer B was 1 M NaCl in 20 mM Tris, pH 8.0. The gradient was developed over 30 min, from 100% Buffer A to 50% Buffer B at a flow rate of 1.0 ml/min. Fractions were collected at 1-min intervals. To individual wells of a 96-well black microfluorimetry plate (Microfluor B, Catalog #011-010-7205, Dynatek. Chantilly, VA), 10 l of each fraction was added to 160 l of 20 mM Tris, pH 8.0, to which 30 l of 4-methylumbelliferyl-N,NЈ,NЉ-␤-D-triacetylchitotrioside (4-MU Glc-NAc 3 ) (Calbiochem, 125 M solution in water) was added. Enzyme reactions were incubated at room temperature. A Dynatek Fluorolite 1000 (filters, excitation 365 nm and emission 450 nm) was used for kinetic fluorescence detection for 60 min.
Chitinase-containing fractions (those which produced linearly increasing fluorescence over the course of the kinetic assay) were identified, pooled, diluted 10-fold with saturated ammonium sulfate, pH 8.0 (to a total volume 5 ml), and injected into a Bio-Gel TSK-phenyl 5PW hydrophobic interaction column (Bio-Rad). Buffer A was 2 M ammonium sulfate with 20 mM Tris, pH 8.0. Buffer B was 20 mM Tris, pH 8.0. A gradient from 100% Buffer A to 100% Buffer B was developed over 30 min at a flow rate of 0.5 ml/min. For the anion exchange and hydrophobic interaction steps, chitinase activity was assessed without addition of any other buffers to the tested fractions and without changing the pH of the final solution before detection. Pilot experiments showed that chitinase activity was readily detectable in buffers with 10 mM to 1.5 M NaCl, and at pH 5.0 to pH 8.5 in 20 mM Tris (data not shown).
Chitinase-containing fractions from the hydrophobic interaction HPLC were pooled, mixed with an equal volume of 50% acetonitrile, 0.1% trifluoroacetic acid, injected into a reverse-phase C-18 PRP Infinity column (Hamilton, Reno, NV), and eluted with a 10%-70% gradient of acetonitrile, 0.1% trifluoroacetic acid at a flow rate of 0.5 ml/min.
Amino-terminal Peptide Sequence Analysis-For purified, native, P. gallinaceum ookinete-produced chitinase, 10% of the volume of the acetonitrile fractions from the reverse-phase step was used to obtain amino-terminal sequence. Recovery of the purified protein from the sample tube was maximized by adding neat trifluoroacetic acid to a final concentration of 10%. One hundred percent of the recovered protein was applied to the biphasic column of a Hewlett-Packard G1005A (Palo Alto, CA), followed by a 1-ml wash with the manufacturer's sample loading solution. The purified protein was then subjected to automated Edman degradation using the manufacturer's recommended protocols and Chemistry Routine 3.5. For amino-terminal sequencing of endoproteinase Lys-C-treated rPgCHT1-NT1, the cleaved recombinant protein was run on SDS-PAGE, electroblotted to PVDF, stained with 0.05% Coomassie Blue in 40% methanol, 1% acetic acid, and destained with 50% methanol. The stained band was excised and submitted to automated Edman degradation on an Applied Biosystems 494/HT Procise Sequencing System in the University of Texas Medical Branch Protein Chemistry Core Facility.
Tryptic Digestion, HPLC Separation, and Microsequencing-To obtain amino acid sequence of internal tryptic peptide fragments, the remaining 90% of the acetonitrile fractions from the reverse-phase step was electrophoretically purified by SDS-PAGE. 50 l of 2ϫ SDS-PAGE sample buffer (Novex, San Diego, CA) was added to each 0.5-ml fraction from the reverse-phase step and concentrated by vacuum centrifugation (Hetovac, Heto Labs, Denmark). These fractions were size-fractionated by SDS-PAGE on a 4 -20% gel (Novex), which was stained with Coomassie Blue R-250 (Bio-Rad) and destained for 4 h with Gel-Stain Destain Solution (Novex) with three changes of destaining solution. Stained bands were excised from the gel, rinsed twice with 50% acetonitrile in HPLC-grade water, and frozen on dry ice. Subsequent protein sequencing steps were performed at the Harvard University Microchemistry Laboratory (William Lane, Director, Cambridge, MA). The band was subjected to in-gel reduction, S-carboxy-amidomethylation, and tryptic digestion (Promega) (22), and a 10% aliquot of the resultant mixture was analyzed. Sequence information was determined by capillary (180 m ϫ 15 cm column, LC Packings, Amsterdam) reverse-phase chromatography coupled to the electrospray ionization source of a quadrupole ion trap mass spectrometer (Finnigan LCQ, San Jose, CA). The instrument was programmed to acquire successive sets of three scan modes consisting of full scan MS over the m/z 395-1200 atomic mass unit, followed by two data-dependent scans on the most abundant ion in that full scan. These data-dependent scans allowed the automatic acquisition of a high resolution (zoom) scan to determine charge state and exact mass and MS/MS spectra for peptide sequence information. The remainder (90%) of the peptide mixture was separated by microbore high performance liquid chromatography using a Zorbax C18 1.0 ϫ 150 mm reverse-phase column on a Hewlett-Packard 1090 HPLC/1040 diode array detector. Optimum fractions were chosen based on differential UV absorbance at 205, 277, and 292 nm, peak symmetry, and resolution and then further screened for length and homogeneity by matrix-assisted laser desorption time-of-flight mass spectrometry on a Thermo BioAnalysis Lasermat 2000 (Hemel, UK). Strategies for peak selection, reverse-phase separation, and Edman microsequencing have been previously described (22). Tryptic peptides were submitted to automated Edman degradation of a Perkin-Elmer/Applied Biosystems 477A or Procise 494-HT protein sequencer (Foster City, CA). When possible, complementary Edman degradation data and MS/MS spectra were used to improve the final sequence interpretation.
Determination of DNA Sequence of Chitinase Gene and Non-translated Flanking Sequence-Degenerate oligodeoxynucleotides were designed based on the amino acid sequences of the following peptides ( ). First-strand cDNA synthesis, using 1 g of total RNA, was prepared using the Capfinder system (CLONTECH, Palo Alto, CA). By using a Perkin-Elmer 9600 thermal cycler and Klen-Taq DNA polymerase (CLONTECH), the following polymerase chain reaction (PCR) cycling protocol was used: 94°C for 3 min for 1 cycle; 94°C for 30 s, 47°C for 30 s, 68°C for 3 min for 35 cycles; 4°C on hold.
The full-length transcript was PCR-amplified using the ookinete first-strand cDNA prepared as described above as template (Fig. 2). To amplify the 5Ј end of the transcript, an antisense nondegenerate genespecific oligonucleotide primer 2501 was synthesized (GGG TTT TCA GTT ATA GTA AGG TC) based on the internal sequence of the PCR product generated by degenerate oligonucleotides derived from the amino acid sequences of GT84 and GT29; the 5Ј PCR primer from the Capfinder kit (AAG CAG TGG TAA CAA CGC AGA GT) was used as the sense primer. Similarly, a sense nondegenerate gene-specific primer 2503 (GAA AAA ATA TGC GAT GGG AAA GCA) was paired with the antisense cDNA synthesis primer (T 30 A/C/G A/C/G/T). The cycling protocol was as follows: 95°C for 3 min for 1 cycle; then 95°C for 30 s, 54°C for 30 s, 68°C for 3 min for 30 cycles; then 4°C. PCR products were ethanol-precipitated and resuspended in water. The DNA was phosphorylated and the ends made blunt (10 l of 10ϫ React 1 buffer (Life Technologies, Inc.), 10 l of 10 mM ATP, 2 l of 10 mM dNTP (dATP, dCTP, dGTP, dTTP), 10 units of T4 DNA polymerase, 10 units of T4 kinase, water to 100 l, incubated at 37°C for 60 min). The products of this reaction were electrophoresed and then purified from a Tris acetate/EDTA agarose gel using the Geneclean II kit (Bio 101, San Diego, CA). The DNA was ligated into pUC18 (SmaI-digested, bacterial alkaline phosphatase-treated, Amersham Pharmacia Biotech), electroporated (1.8 kV, 25 microfarads capacitance, 200 ohms resistance) (Gene Pulser, Bio-Rad) into electrocompetent DH10B E. coli (Life Technologies, Inc.), incubated in SOC (Life Technologies, Inc.) at 37°C in a shaking incubator for 1 h, and plated on LB/ampicillin (100 g/ml) plates. Plasmids from transformants were isolated by alkaline lysis (Wizard MiniPrep, Promega, Madison, WI). Clones containing the appropriately sized insert as determined by restriction analysis were sequenced using dye terminator reactions according to the manufacturer's instructions (DNA Sequencing Kit, Part Number 402079, Perkin-Elmer Applied Biosystems, Foster City, CA) and analyzed by an automated ABI sequencer (ABI Prism, 377 DNA Sequencer, Perkin-Elmer). Independent, cloned full-length genomic DNA and cDNA PCR products were sequenced in their entirety to verify the final sequence.
Expression and Preparation of Recombinant PgCHT1-The NT1 form of rPgCHT1 PCR was amplified from a synthetic DNA template constructed in Escherichia coli-preferred codons (ECPC) (Operon, Alameda, CA). NT1 refers to the amino-terminal sequence of the larger of the two proteins in the purified 60-kDa chitinase chitinase doublet ( Fig.  1e) that was determined by Edman degradation analysis of the purified native protein. The sense primer used to make the rPgCHT1-NT1 construct was GCG CCA TGG GTT ACG GTA GCT ATT GTG GCG; the antisense primer was GCG CTC GAG TTG CAG CGG CAG GTC CAC (the Nco and XhoI sites, respectively, are underlined). PCR cycling conditions were 94°C ϫ 6 min and then 18 cycles of 94°C ϫ 30 s, 50°C ϫ 30 s and then 68°C for 1.5 min. The PCR product was digested and ligated into the NcoI and XhoI restriction sites of the expression vector pET32b. This vector expresses genes of interest as fusion proteins with a 105-amino acid thioredoxin leader sequence, amino-and carboxyl-terminal His 6 tags to facilitate purification by nickel-chelating chromatography, and an enterokinase cleavage site for removal of the amino-terminal fusion partner from the expressed protein. Transformants of the appropriate construct, verified by restriction enzyme analysis and automated sequencing of the construct in DH10B E. coli cells (Life Technologies, Inc.), were then transfected into E. coli strain AD494 (DE3) (24) (Novagen, Madison, WI). The expression of rPgCHT1-NT1 in AD494 (DE3) was further characterized by analyzing protein expression by Coomassie Blue-stained SDS-PAGE gels, Western immunoblotting, nickel-chelating chromatography, and assays of chitinase activity to demonstrate that protein of the appropriate size and activity was expressed.
Conditions for expressing protein for rPgCHT1-NT1 were as follows: 1) growth to A 600 ϭ 0.600 at 37°C in a shaking incubator at 300 rpm; 2) addition of isopropyl-1-thio-␤-D-galactopyranoside to 0.1 mM; 3) growth in a shaking incubator at 300 rpm at 18°C for 16 h.
For enzymatic analysis of recombinant rPgCHT1-NT1, recombinant protein was partially purified in a single nickel-chelating chromatography step. Pellets of 2 liters of induced E. coli cells were pooled and resuspended in a lysis buffer of 20 mM Tris, 10 mM imidazole, 300 mM NaCl at 4°C, without the addition of protease inhibitors. The cell suspension was sonicated, centrifuged at 10,000 ϫ g, and then the supernatant was filtered through a 0.22-m filter and batch-adsorbed to 1 ml of nickel-nitrilotriacetic acid (Ni-NTA)-agarose resin (Qiagen, Chatsworth, CA) at 4°C. Protein was eluted from the Ni-NTA beads with 250 mM imidazole in a buffer of 20 mM Tris, pH 8.0, 300 mM NaCl, after first washing with 10 column volumes of 10 mM and then 20 mM imidazole in the same buffer.
Protease Treatment of Recombinant NT1-PgCHT1-In triplicate, aliquots of rPgCHT1 were incubated with sequence grade endoproteinase Lys-C for 1 h at 23°C (Roche Molecular Biochemicals) or recombinant enterokinase for 4 h at 23°C. Additional samples were heat-inactivated (10 min, 90°C) or left untreated as controls. Amino-terminal protein sequencing was done in the University of Texas Medical Branch Protein Chemistry Core Facility using an Applied Biosystems 494/HT Procise Sequencing System.
Antibody Preparation and Immunoblotting-Two synthetic peptides were designed based on amino acid sequences found in the predicted open reading frame of the P. gallinaceum chitinase (numbered amino acids refer to Fig. 2): 219-DLDGVDIDWEPHGK-232 (active site) and 506-CDGKAAHYYNTDYKE-520 (carboxyl terminus). The peptides were chosen based on predicted antigenicity from a Kyle-Doolittle hydrophilicity plot. Peptide synthesis, carrier coupling, and structural analysis of the conjugates were performed by Mark Garfield and Jan Luszko, NIAID (Twinbrook Facility, Rockville, MD). Peptide purity was verified by reverse-phase HPLC and mass spectroscopy. The peptides were coupled to keyhole limpet hemocyanin via the terminal cysteine (carboxyl-terminal peptide) or lysine (active site peptide). Molar coupling ratios of peptide:carrier, as determined by amino acid analysis, were 39 for the carboxyl-terminal peptide and 413 for the active site peptide.
Animal-use protocols for obtaining polyclonal mouse antisera were approved by the Animal Care and Use Committee, NIAID, National Institutes of Health. Mice were immunized according to the following schedule: each mouse received an intraperitoneal injection of 100 g of conjugate emulsified in 100 l of complete Freund's adjuvant (Sigma) for primary immunization, followed by 50 g of conjugate emulsified in 100 l of incomplete Freund's adjuvant three times at 3-week intervals.
For immunoblotting, proteins were separated on 4 -20% SDS-PAGE gels (Novex) and electroblotted to nitrocellulose using the Novex Xcell Blot II module. After blocking with 5% dried skim milk in PBS, 0.05% Tween 20 (PBS-T), blots were incubated in primary polyclonal antisera at 1:1000 dilution in PBS, 0.05% Tween 20 for 1 h at room temperature. After three washes over 30 min with PBS-T, blots were incubated in secondary antibody (goat anti-mouse IgG heavy and light chain, alkaline phosphatase-conjugated, Kirkegaard & Perry Laboratories, Gaithersburg, MD) at 1:5000 dilution in PBS, 0.05% Tween 20, for 1 h. After three washes over 30 min with PBS-T, the blots were developed in an alkaline phosphatase substrate (Western Blue, Promega, Madison, WI).
Analysis of Chitinase Activity-Chitinase activity of both native and recombinant chitinases was assessed in three ways. First, enzyme preparations were analyzed for their ability to degrade polymeric chitin, as described previously (14). Second, microfluorimetry (HTS7000, Perkin-Elmer, excitation 360 nm and emission 465 nm) was used to measure the hydrolysis of 4-MU GlcNAc, 4-MU GlcNAc 2 , 4-MU GlcNAc 3 , and 4-MU GlcNAc 4 (Sigma) as described previously. Initial enzyme reaction rates were measured. Enzymatic activity is reported as relative fluorescence units or fold change. Third, TLC was used to analyze the products of recombinant or ookinete-produced P. gallinaceum chitinase using native chitin oligosaccharides and synthetic 4-MU derivatives of chitin oligosaccharides as substrates. With native chitin oligosaccharides (GlcNAc 1-6 , Calbiochem), 6 l of 5 mM substrate was mixed with 4 l of 5ϫ citrate/phosphate, pH 3-7 (McIlvaine buffer), to which was added 10 l of enzyme. The reaction mixtures were incubated at 37°C overnight and then analyzed by TLC. 3 l of the reaction mixture were applied to Silica Gel-60 TLC plates, 20 ϫ 20 cm (EM Science, Gibbstown, NJ) and chromatographed in isopropyl alcohol: ethanol:water (5:2:1). The plates were developed by spraying the plates with 10% sulfuric acid in ethanol followed by heating at 120°C for 10 -20 min to detect dark spots. The chromatograms were scanned on a flat-bed scanner and images processed using Adobe Photoshop 4.0 (Adobe Systems, Inc., San Jose, CA). Samples of enzyme reacted with synthetic 4-MU substrates (5 l, containing 0.5 nmol of substrate/ product) were applied to 10-cm Silica Gel-60 TLC plates following overnight incubation at 37°C. Products were separated as above. These chromatograms were visualized with a FLUOR-S imager using 366 nm excitation/456 nm emission filters (Bio-Rad).

Determination of pH Activity Profiles and Allosamidin Inhibition
Curves-By using a citrate/sodium phosphate buffer with pH ranging from 3.0 to 7.0 in 0.5 pH unit increments, aliquots of enzyme were incubated with 4-MU GlcNAc 3 (100 M). Chitinase reaction rates were analyzed by microfluorimetry. Allosamidin (courtesy of Dr. Shosei Sakuda, University of Tokyo, Japan, and Dr. Jon Mynderse, Eli Lilly and Co., Indianapolis) was made as a 2 mg ml Ϫ1 stock in water and diluted.
Identification of a Fragment of a P. falciparum Chitinase-824 bp of preliminary sequence data of the P. falciparum chitinase, found by random shotgun sequencing in the chromosome 14 sequencing project, was obtained from The Institute for Genomic Research website. The plasmid clone, PNADI77, was obtained from The Institute for Genomic Research and the entire insert sequenced. The Institute for Genomic Research sequence, plus the additional sequence that we determined, has been submitted to GenBank TM with accession number AF072442. The sequencing of chromosome 14 was part of the International Malaria Genome Sequencing Project.
Computer Analysis-Analysis of DNA sequences was performed using the Lasergene set of programs (DNASTAR, Madison, WI). Homology searches were performed with gapped BLAST, with further profile analysis performed with PSI BLAST. The multiple alignments were generated using the GIBBS sampling procedure. The signal sequence was predicted using the algorithm of von Heijne (25).

RESULTS
Purification and Microsequencing of the P. gallinaceum 60-kDa Chitinase Protein-A mixture of supernatants and soluble extracts of 2 ϫ 10 9 P. gallinaceum ookinetes was sequentially subjected to anion exchange, hydrophobic interaction, and reverse-phase HPLC (Fig. 1). 4-MU GlcNAc 3 was used as the substrate for following endochitinase activity through the anion exchange and hydrophobic interaction steps; acetonitrile/ trifluoroacetic acid in the reverse-phase step irreversibly disrupted the chitinase activity. Coomassie Blue staining of the major peak (fractions 62-64) in the final purification step showed an apparently pure doublet with apparent molecular mass of 60 kDa (Fig. 1e). Fractions 62-64 contained a 60-kDa protein doublet; fractions 65-66 contained a single 60-kDa protein and a 210-kDa protein. The 60-kDa doublet was suspected to be a chitinase for two reasons as follows: most reported family 18 chitinases are between 35-and 80-kDa molecular mass; and polyclonal antisera raised to a synthetic peptide derived from the active site of the Entamoeba histolytica chitinase recognize an ookinete stage-specific 60-kDa doublet (26). When an aliquot of pooled fractions 62-66 was subjected to immunoblot analysis, antisera to the E. histolytica chitinase active site recognized the 60-kDa doublet in fractions 62-64 and the 60 and 210-kDa proteins in fractions 65-66 (data not shown).
Ten percent of the pooled reverse-phase fractions 62-64 was analyzed by direct amino-terminal sequencing. The remaining 90% was lyophilized in the presence of SDS, subjected to SDS-PAGE, and stained with Coomassie Blue. The purified doublet was excised from the gel and subjected to microsequencing.
Direct Edman degradation of an aliquot of pooled fractions 62-64 ( Fig. 1e; 5 pmol of protein as determined by amino acid analysis) showed two distinguishable amino-terminal sequences (amino-terminal 1 and amino-terminal 2, Fig. 3). Three peptides, GT29, GT33, and GT84 (Fig. 3), were produced by in situ trypsin digestion of the purified protein doublet and isolated by reverse-phase HPLC; these protein fragments were further analyzed by mass spectroscopy and Edman degradation.
Determination of the DNA Sequence of the P. gallinaceum Chitinase Gene, PgCHT1, That Encodes the 60-kDa Chitinase-Degenerate oligodeoxynucleotide primers were synthe- sized based on the tryptic peptide sequences GT33, GT29, and GT84. PCR, using first-strand cDNA of mature P. gallinaceum ookinetes as template and the pairing of degenerate oligodeoxynucleotide primers GT33 with GT29 and GT29 with GT84, respectively, generated single products that were cloned and sequenced (data not shown). These PCR fragments encoded amino acid sequences without recognizable homology on BLAST search. Nondegenerate gene-specific oligodeoxynucleotides, synthesized based on the internal sequences of these PCR products, were used as primers in two separate PCR reactions; these reactions used first-strand cDNA of mature P. gallinaceum ookinetes, prepared with the CLONTECH Capfinder system as template (Fig. 2). The first PCR reaction used the CLONTECH 5Ј Capfinder PCR primer and a gene-specific antisense 3Ј primer; similarly, the second PCR used the oligo(dT) primer as the 3Ј antisense primer and a gene-specific 5Ј sense primer. When amplifying the 5Ј end of the cDNA template, a single band was obtained, cloned, and sequenced; when amplifying the 3Ј end, several bands were obtained, and only the highest molecular weight band was cloned and sequenced (data not shown). These two sequences (using PCR primers to cross the overlap between the two gene-specific primers) were combined to produce a 2508-bp cDNA corresponding to the full-length mRNA transcript from the initiation of transcription to the poly-adenylation tail (GenBank TM accession number AF064079). Fig. 3 depicts the full-length translated open reading frame. The two amino-terminal sequences of the purified 60-kDa doublet determined by direct Edman degradation and the amino acid sequences of tryptic digest fragments GT29, GT33, and GT84 of the 60-kDa doublet were all found in the amino acid sequence encoded by PgCHT1.
PCR products derived from either gDNA or cDNA as template had the same size and sequence (data not shown), demonstrating that the open reading frame was a single exon. A Southern blot of P. gallinaceum gDNA probed with a 42-mer oligodeoxynucleotide probe directed against the active site was consistent with a single or low copy number copy gene (data not shown).
Analysis of the Primary Structure of the Encoded P. gallinaceum Chitinase-The cloned cDNA PCR product representing a full-length chitinase cDNA contains 275 bp of untranslated sequence at the 5Ј end, a 1761-bp opening reading frame encoding 587 amino acids with a predicted molecular mass of 67,927 Da (Fig. 3), and 432 bp of untranslated sequence at the 3Ј end. Although a Kozak consensus sequence (27) is not present at the presumptive site of the initiation of transcription, a consensus sequence AAA(A/A)TG at the predicted transcrip-tional start site typical of Plasmodium spp. is present (28). A secretory signal sequence of 22 amino acids is predicted, followed by a 42-amino acid lysine/asparagine-rich precursor region not found in the enzymatically active, purified 60-kDa chitinase doublet. Amino-terminal sequences 1 and 2 are colinear near the amino terminus. The two forms of the purified chitinase are predicted to be of molecular mass 60,595 and 59,117 Da, respectively, similar to the apparent molecular masses observed for the protein doublet on SDS-PAGE (Fig. 1e).
BLAST search with the full-length P. gallinaceum amino acid sequence revealed homologies to numerous other chitinases; similarities were limited to the substrate binding and catalytic active site motifs (Fig. 4). On searching the P. falciparum chromosome 14 genome data base (29), a partial sequence of a P. falciparum gene encoding a family 18 glycohydrolase catalytic domain was found. This P. falciparum fragment was approximately 1000 bp (Fig. 4) and encoded motifs characteristic of the substrate-binding and catalytic active sites of family 18 glycohydrolases (Fig. 4) (30). This P. falciparum chitinase has been further characterized (31). The substrate-binding site (amino acids 187-194) contains the consensus sequence XXXSXGG, where X represents hydrophobic amino acids, except at the amino-terminal end where there is an unusual non-conservative isoleucine substitution for a lysine. Both Plasmodium chitinases are typical of family 18 glycohydrolases, with one exception: uniquely the Plasmodium chitinases lack a highly conserved tyrosine (or a conservative change such as phenylalanine) at position 229. A putative chitin-binding domain is present that, although not sharing significant amino acid sequence similarity with chitin-binding domains of any eukaryotic chitinases, has significant secondary structural homology to bacterial chitin-binding domains (data not shown).
Analysis of Different Forms of P. gallinaceum Chitinases and Time Course of Expression-To confirm further that the P. gallinaceum gene sequence determined above encoded the purified 60-kDa doublet that we identified as a chitinase, antisera for use in Western immunoblots were prepared to synthetic FIG. 2. Cloning strategy of PgCHT1 cDNA. Reverse transcriptase PCR was performed using as template first strand cDNA (prepared with the Capfinder system; see "Experimental Procedures") synthesized from P. gallinaceum ookinete total RNA and degenerate oligonucleotide primers based on the amino acid sequences of GT29 and GT84, as indicated by arrowheads within rectangles. The resulting PCR product was cloned and sequenced, from which the non-degenerate PCR primers 2503 and 2501 were designed. Primer 2501 paired with the Capfinder 5Ј oligonucleotide primer and Primer 2503 paired with the oligo(dT) primer (see "Experimental Procedures" for primer sequences) were used for two separate PCR reactions using first strand ookinete cDNA as template. A single product was obtained with the primer pair 2501 ϫ 5Ј PCR primer; this PCR product was cloned and sequenced. Several discrete bands were obtained with the primer pair 2503 ϫ oligo(dT); only the largest of these bands was cloned and sequenced. The sequences of these two reverse transcriptase-PCR products overlapped and gave rise to the full-length 2508-bp cDNA of PgCHT1 (GenBank TM accession number AF064079). peptides consisting of the catalytic active site and a peptide near the carboxyl terminus (Fig. 3). Antibody specificity of antisera raised against the synthetic peptides derived from the P. gallinaceum 60-kDa chitinase was confirmed by Western immunoblotting P. gallinaceum zygotes and ookinetes with immune and preimmune sera against both synthetic peptides; immune sera recognize proteins in ookinete but not zygote extracts (see below), whereas preimmune sera do not detect any proteins in the same preparations (data not shown). Western immunoblot of purified chitinase (fractions 65 and 66 from Fig. 1e) with antisera to the carboxyl terminus detects only the 60-kDa doublet (Fig. 5a, right panel). Unexpectedly, antisera to the catalytic active site recognize, in addition to the 60-kDa doublet, a 210-kDa protein (Fig. 5a, left panel). This latter result suggests that the 210-kDa protein may lack the carboxyl-terminal epitope present in the putative chitin-binding domain of the 60-kDa doublet. The 60-kDa doublet and the 210-kDa protein co-eluted through a multiple step purification protocol (Fig. 1e), suggesting the possibility that P. gallinaceum ookinetes secrete at least two structurally related forms of chitinase, a 60-kDa doublet and a 210-kDa protein.
P. gallinaceum ookinetes produce a number of proteins (210-, 160-, 66-, 60-kDa doublet, 35-kDa) that are stage-specifically expressed in mature ookinetes and are not present in zygotes, as determined by analysis with Western immunoblot; all of these proteins react with antisera raised to a synthetic peptide derived from the active site of PgCHT1 (Fig. 5b, left panel). This antiserum lacked reactivity to P. gallinaceum zygote extracts (Fig. 5b, left panel) and non-recombinant E. coli extracts (data not shown). Only a subset of these bands (160-, 66-, and 60-kDa doublet) react with antisera directed against the carboxyl-terminal region (Fig. 5b, right panel); the 66-kDa band (thin arrow) is the precursor to the 60-kDa doublet as shown by Western immunoblot of ookinete extracts with antisera raised to a synthetic peptide from the region between the signal peptide and the NT1 cleavage site (data not shown). The appearance of these bands correlates with the appearance of chitinase activity during the development of zygotes to ookinetes (Fig.  5c), consistent with the hypothesis that these bands are those of chitinases.
P. gallinaceum Ookinetes Produce at Least Two Chromatographically and Immunologically Distinct Chitinases-Ini-  tially, only one broad peak of endochitinase activity was isolated by anion exchange chromatography of P. gallinaceum ookinete extracts. When a slower flow rate and more gradual salt gradient were used to chromatograph ookinete extracts by strong anion exchange HPLC, two peaks of endochitinase activity (hereafter referred to as peak 1 and peak 2), as assessed by hydrolysis of 4-MU GlcNAc 3 were reproducibly resolved (Fig. 6a).
The separate peaks of chitinase activity were concentrated, subjected to SDS-PAGE under both reducing and non-reducing conditions, and Western immunoblotted using antisera that recognize the catalytic active site and carboxyl terminus (Fig.  6, b-e). Several proteins reacted with antibody to the active site of PgCHT1. Western immunoblot of peak 1 (fractions [35][36][37] with active site antisera demonstrated the 60-kDa chitinase doublet both under non-reducing (Fig. 6b) and reducing (Fig.  6d) conditions. In addition, a 35-kDa protein in peak 1, of unknown function, reacted with the active site antisera. This 35-kDa protein is stage-specifically expressed in ookinetes, is not found in zygotes (Fig. 5b, left panel), and could be either a degradation product of another chitinase protein, the product of another chitinase gene, or simply an unrelated protein.
Western immunoblot of peak 2 (fractions 42-44) with active site antisera demonstrated a predominant band of 210 kDa under non-reducing conditions (Fig. 6b) that, under reducing conditions (Fig. 6d), was not detected and was replaced by a 35-kDa protein. In addition, in fraction 42, a band of 160 kDa reacted with antiserum to the active site but not the carboxyl terminus (Fig. 6, b-e). This 160-kDa protein is stage-specifically expressed in ookinetes and is not detectable in zygotes (Fig. 6b, both panels). Additional resolution of this second peak of chitinase activity with additional anion exchange chromatography and enzymatic analysis suggests that both the 210and the 160-kDa proteins are chitinases (data not shown). The active site antisera recognized both the 210-and 35-kDa bands, but the carboxyl-terminal antisera recognized neither (Fig. 6, b  and c). This finding suggests that the carboxyl-terminal epitope is absent in the 210-and 35-kDa bands in the second peak of chitinase activity, as well as the 35-kDa protein in the first peak of chitinase activity.
Sites of Action of Endoproteinase Lys-C on the 60-kDa P. gallinaceum Chitinase-P. gallinaceum chitinase has been reported to be secreted as an inactive zymogen that mosquito midgut proteases activate to a fully active enzyme (12); this finding has become established in the literature (12,(32)(33)(34). The serine protease endoproteinase Lys-C (Endo Lys-C) was reported to increase the P. gallinaceum chitinase activity in culture supernatants up to 13-fold (12). To characterize further this phenomenon, we used Western immunoblotting of native, ookinete-secreted chitinase (Fig. 7) and amino-terminal sequencing of Endo Lys-C-cleaved rPgCHT1 to delineate the sites where Endo Lys-C cleaves the 60-kDa P. gallinaceum chitinase.
Concentrated ookinete supernatants were treated with Endo Lys-C and the reactions terminated with the serine protease inhibitor, AEBSF. We found that chitinase activity, as assessed by 4-MU GlcNAc 3 hydrolysis, was unaffected by treatment with Endo Lys-C, in contrast to the results of others (12) (data not shown). Immunoblots were performed on the reaction mixtures, under non-reducing and reducing conditions, with polyclonal antisera to the active site-and carboxyl terminus-derived peptides (Fig. 7). Endo Lys-C had two effects on the 60-kDa doublet as follows: 1) both bands of the doublet were cleaved, resulting in a single band running at a slightly smaller molecular mass than the bottom band of the doublet; and 2) a time-dependent disappearance of the carboxyl-terminal epitope occurred. Antisera to the active site demonstrated that a single full-length chitinase remained after Endo Lys-C treatment under non-reducing conditions. Under reducing conditions a small fragment was removed, resulting in a slightly faster migration of the processed chitinase (Fig. 7, right top  arrowhead). The precise epitope recognized by the carboxylterminal antisera, which demarcates the putative chitin-binding domain, contains a predicted Endo Lys-C site at Lys 509 - ). An equal number of parasite equivalents were loaded into each lane or assayed for chitinase activity. Single arrowheads indicate proteins associated with peak 1 of chitinase activity (see Fig. 9); double arrowheads indicate proteins associated with peak 2 of chitinase activity. The thin arrow indicates the precursor of the protein doublet of 68-kDa indicated by the upper single arrowhead, as demonstrated by a Western immunoblot using antisera directed against the pro-enzyme domain (data not shown). Protein bands that do not increase in intensity during the course of ookinete development, and thus presumably are not chitinases, are not indicated with arrowheads. c, chitinase activity, as detected with 4-MU GlcNAc 3 , as a function of time after zygote formation.
Ala 510 , consistent with the experimental findings. Cys 506 , at the amino terminus of the epitope, may form a disulfide bridge with the only cysteine downstream from it, Cys 557 (Fig. 3).
Since Endo Lys-C converts the 60-kDa chitinase doublet to a slightly smaller single band, we determined the amino-terminal site of action of this protease. After treatment of rPgCHT1-NT1 with Endo Lys-C, the cleaved product was separated by SDS-PAGE, transferred to a PVDF membrane, and analyzed by Edman degradation. This analysis showed that Endo Lys-C cleaves rPgCHT1 on the carboxyl-terminal side of Lys 86 , just downstream from the NT2 cleavage site (Fig. 3). Immunoblot of Endo Lys-C-treated rPgCHT1 with the anti-carboxyl-terminal antiserum showed loss of the epitope (data not shown), similar to that found when native ookinete-produced 60-kDa chitinase was treated with Endo Lys-C (Fig. 7).
Expression of Recombinant PgCHT1-Numerous constructs of native codon-based PgCHT1, using different vectors expressed under a variety of temperature conditions and isopropyl-1-thio-␤-D-galactopyranoside concentrations in E. coli host cells, did not produce more than ϳ5 g of recombinant protein/ liter-induced E. coli cells. Undetectable quantities of recombinant protein were obtained when attempting to express similar constructs in a well characterized Saccharomyces cerevisiae expression system (35). Because the A ϩ T codon bias of this gene (70.6%) was suspected to be the primary barrier to producing recombinant protein, a synthetic PgCHT1 gene was constructed using E. coli-preferred codons and used as a PCR template for making the rPgCHT1-NT1construct (see "Experimental Procedures"). Recombinant PgCHT1-NT1 (depicted schematically in Fig. 8a), expressed with a construct synthesized in E. coli-preferred codons and expressed in E. coli AD494 (DE3) cells, produced ϳ5-10 mg of recombinant protein/liter of induced cells, of which ϳ1-3% was soluble and active (data not shown). Chitinase activity was readily detectable in crude soluble extracts of the cells (data not shown). When the same constructs were expressed in E. coli BL21 (DE3) cells, no chitinase activity was detected, despite a comparable total quantity of protein produced (data not shown). Western immunoblot analysis of protein obtained by a Ni-NTA purification step demonstrated that the eluted, soluble proteins were Ͼ90% PgCHT1 (data not shown). Approximately 20% of the soluble protein in eluted fractions was rPgCHT1-NT1 of the predicted length; the rest were aggregated and truncated forms of rPgCHT1 (data not shown).

P. gallinaceum Chitinases Degrade Polymeric Chitinase and Have Identical Substrate Preferences and Reaction Product
Profiles-A previous report demonstrated that several bands of chitinolytic activity were present in crude extracts and culture supernatants of P. gallinaceum ookinetes, as determined in a glycol chitin activity gel (14). rPgCHT1-NT1, whether or not treated with proteases, also degraded polymeric chitin in a glycol chitin activity gel (Fig. 8b); quantitation of chitin degradation was not possible from this experiment. As negative controls, Endo Lys-C alone and heat-inactivated rPgCHT1NT1 (data not shown) had no detectable chitinase activity in the activity gel. The ability of rPgCHT-NT1 to cleave 4-MU Glc-NAc 3 was then assessed. When rPgCHT1-NT1 was treated with enterokinase to remove the thioredoxin fusion partner, there was no change in enzymatic rate of 4-MU GlcNAc 3 (Fig.  8c). However, to exploit the finding that Endo Lys-C cleaves PgCHT1 just downstream of the NT2 cleavage site, rPgCHT1 was treated with Endo Lys-C. This protease treatment increased chitinase activity 73-fold, using the 4-MU GlcNAc 3 substrate (Fig. 8c). As an additional control to show that the FIG. 6. Two chromatographically separable chitinase activities produced by P. gallinaceum ookinetes. a, quaternary ammonium anion exchange HPLC of ookinete extracts using a shallower gradient than that used in Fig. 1a was able to two separate peaks of chitinase activity. Proteins from each peak of chitinase activity were subjected to SDS-PAGE with a 4 -20% polyacrylamide gel under both non-reducing and reducing conditions and immunoblotted (b-e). b and d, non-reduced and reduced, respectively, probed with active site antiserum; c and e, non-reduced and reduced, respectively, probed with carboxyl-terminal antiserum.

FIG. 7. Effect of Endo Lys-C on P.
gallinaceum ookinete-produced chitinase. Concentrated ookinete supernatants (from ϳ2 ϫ 10 7 parasites per lane) were either treated with Endo Lys-C or buffer alone. Aliquots were taken for immunoblotting at 5, 30, and 120 min. An equal volume of 2ϫ sample buffer with or without 10% ␤-mercaptoethanol was added to an aliquot of each fraction, and the samples were subjected to SDS-PAGE with a 16% polyacrylamide gel. After electroblotting, the blots were probed with active site or carboxyl terminus antiserum.
Endo Lys-C and enterokinase did in fact cleave the recombinant protein, Western immunoblot analysis performed before and after protease treatment demonstrated the appropriately sized cleavage products (data not shown).
To characterize the substrate specificity and reaction product profiles of P. gallinaceum chitinases, the activities of ookinete-produced chitinase and rPgCHT1-NT1 were assessed with native chitin oligomers (Fig. 9) and 4-MU GlcNAc substrates (Fig. 10). P. gallinaceum ookinete-produced chitinase (peak 1, peak 2, and unfractionated extracts) and rPgCHT1-NT1 had identical substrate preferences and reaction product profiles, with the exception that the crude extract had an N-acetylglucosaminidase activity not found in peak 1 or peak 2 of anionexchanged chromatography fractionated ookinete extracts (Figs. 9 and 10). Both parasite-produced chitinase and rPgCHT1-NT1 hydrolyzed 4-MU GlcNAc 3 and 4-MU GlcNAc 4 (Fig. 10). Treatment of rPgCHT1-NT1 with Endo Lys-C had no effect on the pattern of substrates preferred by the enzyme nor on the reaction product profile (data not shown). Regardless of whether rPgCHT1-NT1 was treated with Endo Lys-C, it did not (amino acids (aa) Tyr 65 to Gln 587 ) was amplified from a synthetic PgCHT1 gene constructed in E. coli preferred codons with NcoI and XhoI restriction sites included in the 5Ј and 3Ј ends of the PCR primers, respectively. The NcoI and XhoI restriction enzyme-digested PCR product was cloned into the NcoI and XhoI restriction sites of the bacterial expression vector, pET32b. This vector expresses proteins fused to a 105-amino acid thioredoxin (trx) leader sequence and to hexahistidine tags (His 6 ) at both the amino and carboxyl termini. An enterokinase cleavage site allows for removal of the amino-terminal fusion protein, leaving the correct NT1 amino terminus of rPgCHT1-NT1. Endo Lys-C was experimentally determined (see Fig. 7 and text) to cleave rPgCHT1-NT1 as shown schematically. The PgCHT1-NT1 construct used for enzymological analysis corresponds to one of the forms of the PgCHT1 gene product secreted by ookinetes, as determined by direct aminoterminal sequencing of the purified 60-kDa chitinase ( Fig. 1e and Fig.  3). b, rPgCHT1-NT1 degrades polymeric chitin in a glycol chitin activity gel. Equal amounts of rPgCHT1-NT1 (prepared and treated as described under "Experimental Procedures") were electrophoresed in a native, non-denaturing 8% polyacrylamide gel into which 0.02% glycol chitin had been incorporated. After the gel was run and was incubated in 0.1 M sodium phosphate, pH 6.8, the gel was counter-stained with Calcofluor White and visualized by transillumination with UV light. c, rPgCHT1-NT1 cleaves 4-MU GlcNAc 3 and is enhanced by Endo Lys-C but not enterokinase proteolytic treatment. Chitinase activity is represented as fold change of relative fluorescence units. P. gallinaceum Chitinases Have Different pH Activity Profiles and Susceptibility to the Inhibitor Allosamidin-pH activity profiles were determined by microfluorimetry for peaks 1 and 2 of chitinase activity and for Endo Lys-C-treated rPgCHT1 (Fig. 11a) (peak 1 contains PgCHT1), using 4-MU GlcNAc 3 as substrate. Peak 2 chitinase activity had a broad pH optimum of pH 4.0 -5.0. Peak 1 chitinase activity was optimal at pH 5.0. Similar to peak 1, rPgCHT1 also had a pH optimum of 5.0. This reproducible difference in pH activity profiles suggests that the two peaks of chitinase activity are comprised of chitinases with different amino acids present in the catalytic sites of the enzymes and thus are products of different genes.
To confirm and extend the suggestion that P. gallinaceum ookinetes may secrete at least two chitinases derived from different genes, the sensitivity of each of the two peaks of chitinase activity and rPgCHT1 to the chitinase inhibitor allosamidin was determined (Fig. 11b). The IC 50 value estimated for peak 1 and rPgCHT1 were similar (7 and 12 M, respectively). In contrast, the IC 50 value for peak 2 was 0.3 M, about 30-fold less than that found for peak 1 or rPgCHT1. The allosamidin inhibition data provides additional evidence for a second P. gallinaceum chitinase gene and that peak 1 is the product of the PgCHT1 gene. The 1 and 0.1 mM concentrations of allosamidin used in previous studies to block oocyst development (12) far exceed the IC 50 values for both peaks of chitinase activity and would completely inhibit chitinase activity in ookinete extracts. DISCUSSION We report here the purification of a 60-kDa P. gallinaceum ookinete-secreted chitinase and characterize the gene, PgCHT1, encoding it. The experiments presented here identify at least two developmentally regulated chitinases expressed by P. gallinaceum ookinetes. Both are inhibited by allosamidin at concentrations far less than those used in in vivo studies for blocking ookinete penetration of the mosquito peritrophic membrane (12). Both are secreted and act as endochitinases, a property that would be expected of enzymes that allows the ookinete to penetrate and traverse the PM in the mosquito midgut.
At least two chitinase activities are separable by HPLC. The first, encoded by the gene, PgCHT1, was identified from peak 1 as a 60-kDa doublet (Fig. 1), which is composed of two forms of the protein, NT1 and NT2, which differ in size by 14 amino acids (Fig. 3). The NT1 form of PgCHT1, expressed as a recombinant protein in E. coli, has chitinase activity that is increased 73-fold by treatment with Endo Lys-C. The Endo Lys-C cleavage site is between eight and nine amino acid residues down-stream from the amino terminus of the NT2 form, strongly suggesting that the NT2 form is an active chitinase. It is not clear if the NT1 form is fully active or needs to be converted further by parasite-produced proteases to the NT2 form for full activity. Whereas activity of the recombinant fusion protein rPgCHT1-NT1 is enhanced by Endo Lys-C, we found that Endo Lys-C has no effect on the activity of crude native ookineteproduced chitinase. One explanation of this discrepancy is that the 105-amino acid fusion partner of the recombinant fusion protein could be acting as the pro-enzyme domain does in the native produced gene product to inhibit enzyme activity and that the four amino acids remaining amino-terminal to the NT1 site remaining after enterokinase treatment prevent full activity of the enzyme. It is also possible that Endo Lys-C could inactivate the activity of some chitinases in crude ookinete culture supernatants while increasing the activity of other chitinases, leading to the finding of no overall change of chitinase activity of ookinete culture supernatants when treated by Endo Lys-C. The essential point is that we have demonstrated that the ookinete itself is capable of processing a pro-chitinase to a fully active form, and mosquito midgut proteases are not required for fully active ookinete-produced chitinase activity. This interpretation stands in contrast to previous models that hypothesize that the late trypsins expressed by the mosquito midgut more than 10 h after blood meal ingestion activate ookinete-produced chitinase (to increase the activity as much as 13-fold) for the purpose of optimizing the timing of mosquito midgut invasion (12,16,32). We have recently demonstrated that the P. falciparum chitinase homolog, PfCHT1, has no proenzyme domain and is a chitinase that does not require proteolytic activation for its enzymatic activity (31). The comparative activities of the NT1 and NT2 forms of PgCHT1 will be the focus of further study.
Peak 2 chitinase hydrolyzes 4-MU derivatives in a pattern similar to that produced by peak 1 and rPgCHT1 (data not shown). However, peak 2 chitinase has a distinct pH activity profile and is about 30-fold more sensitive to allosamidin (Fig.  11b) than rPgCHT1 and native peak 1 chitinase. Under reducing and denaturing conditions, a number of proteins, including a 35-kDa protein in peak 2, were identified by Western immunoblotting that increase in expression in parallel with an increased chitinase activity (Figs. 5 and 6). The 35-kDa protein reacts with antisera prepared from a peptide from the catalytic domain of PgCHT1, but not with antisera from a carboxylterminal peptide of PgCHT1. Collectively, the different enzymatic properties and immunological reactivity of the peak 2 chitinase support the hypothesis that this chitinase is the product of a different gene, PgCHT2. Under non-reducing conditions, peak 2 chitinase migrates as a ϳ210-kDa protein. It is FIG. 11. Analysis of pH activity profile and allosamidin sensitivity of the two peaks of chitinase activity chromatographically separated from P. gallinaceum ookinete extracts as in Fig. 7 and Endo Lys-C-activated rPgCHT1-NT1. Chitinase activity is expressed as percent activity of enzyme in the absence of inhibitor (as detected with the 4-MU GlcNAc 3 substrate). O, peak 1 of chitinase activity; q, rPgCHT1; छ, peak 2 of chitinase activity. These experiments were repeated two times each with two different preparations of enzyme. not clear whether the complex would be composed solely of PgCHT2 protein subunits.
Other proteins (ϳ30-kDa in peak 1 and ϳ160-kDa in peak 2) that cross-react with the active site antisera have not yet been further characterized, although these proteins also are stagespecifically expressed by ookinetes (Fig. 5). Additional chromatography of peak 2 suggests that the ϳ160-kDa protein in fact does have chitinase activity, as determined with the 4-MU GlcNAc 3 substrate (data not shown). This latter observation is consistent with the possibility that these unidentified crossreactive proteins are also chitinases. Experiments to identify and characterize the other chitinase proteins in peaks 1 and 2 at the molecular level are underway.
In the work presented here, we first identified a fragment of a putative P. falciparum chitinase. This fragment contains a catalytic domain but not a recognizable chitin binding domain; a stop codon appears at the end of the predicted catalytic domain. Work is in progress to characterize further the role of the P. falciparum chitinase gene, PfCHT1, in parasite invasion of the Anopheles midgut (31).
An important goal of research on the malaria parasite chitinases is to develop novel ways to interrupt malaria transmission. With the cloning of a full-length P. gallinaceum chitinase gene, recombinant chitinase can be synthesized and tested as a transmission-blocking vaccine in an in vivo avian malaria transmission model. The Plasmodium chitinases potentially share B cell epitopes with similar proteins in humans, particularly in the catalytic active site (Fig. 4 (38, 39)). Detailed structural analysis may provide insight into regions of the Plasmodium chitinases that could be safe and effective transmission-blocking vaccine candidates, avoiding potential autoimmune reactions. Determining three-dimensional structures of the Plasmodium chitinase domains may be possible and would provide the basis for rational design of a transmissionblocking drug. It may be possible to develop peptide chitinase inhibitors, the genes of which may be used for transforming mosquitoes (40,41) to make them refractory to malaria transmission. Finally, we have initiated experiments to knock out the P. falciparum chitinase gene to determine whether its gene product is necessary for mosquito invasion or whether, perhaps, P. falciparum too has additional chitinase genes.