Identification of Novel Proteases and Immunomodulators in the Secretions of Schistosome Cercariae That Facilitate Host Entry *S

Schistosomiasis, caused by parasitic helminths, remains a serious human disease in the tropics. Cercariae of Schistosoma mansoni infect their hosts by direct skin penetration, aided by secretions from acetabular and head glands. Both proteolytic and immunomodulatory properties have been ascribed to the released material, but to date only five isoforms of elastase and one putative anti-inflammatory protein (Sm16) have been cloned. We analyzed secretions from mechanically transformed cercariae by two-dimensional electrophoresis. An average gel image was created and compared with a separation of soluble larval extract, revealing a less complex spot pattern in the secretions with 60% of the spots matched to the larval extract. Subsequent tandem mass spectrometric analysis identified 48 spots from the released material, representing approximately 80% of its normalized volume. Twenty-nine of these are likely to originate in the vesicles, and 18 are likely to originate in the cytosol of the glands (the latter class being present due to holocrine secretion); one is unknown. The vesicular proteins were significantly more enriched than the cytosolic proteins in the released material when compared with the larval extract. A novel metalloproteinase (termed SmPepM8) was the second most abundant constituent after three isoforms of cercarial elastase. In addition, a dipeptidyl peptidase IV (SmDPP IV) was discovered but in much smaller quantity. A new serine protease inhibitor (SmSerp_c) was also prominent. Along with Sm16, four potential immunomodulators were identified, three with similarity to venom allergens (SmSCP_a, _b, and _c) and one with homology to the potassium channel blockers in scorpion venom (SmKK7). Interrogation of the expressed sequence tag database found transcripts encoding the majority of vesicular proteins present solely in the intramolluscan stages of the life cycle. Distinct patterns of radiolabel incorporation suggested three separate origins for the vesicular proteins. All the novel constituents merit investigation as vaccine candidates, and the potential immunomodulators merit investigation as therapeutic agents.

Schistosomiasis, caused by parasitic helminths, remains a serious human disease in the tropics. Cercariae of Schistosoma mansoni infect their hosts by direct skin penetration, aided by secretions from acetabular and head glands. Both proteolytic and immunomodulatory properties have been ascribed to the released material, but to date only five isoforms of elastase and one putative anti-inflammatory protein (Sm16) have been cloned. We analyzed secretions from mechanically transformed cercariae by two-dimensional electrophoresis. An average gel image was created and compared with a separation of soluble larval extract, revealing a less complex spot pattern in the secretions with 60% of the spots matched to the larval extract. Subsequent tandem mass spectrometric analysis identified 48 spots from the released material, representing approximately 80% of its normalized volume. Twenty-nine of these are likely to originate in the vesicles, and 18 are likely to originate in the cytosol of the glands (the latter class being present due to holocrine secretion); one is unknown. The vesicular proteins were significantly more enriched than the cytosolic proteins in the released material when compared with the larval extract. A novel metalloproteinase (termed SmPepM8) was the second most abundant constituent after three isoforms of cercarial elastase. In addition, a dipeptidyl peptidase IV (SmDPP IV) was discovered but in much smaller quantity. A new serine protease inhibitor (SmSerp_c) was also prominent. Along with Sm16, four potential immunomodulators were identified, three with similarity to venom allergens (SmSCP_a, _b, and _c) and one with homology to the potassium channel blockers in scorpion venom (SmKK7). Interrogation of the expressed sequence tag database found transcripts encoding the majority of vesicular proteins present solely in the intramolluscan stages of the life cycle. Distinct patterns of radiolabel incorporation suggested three separate origins for the vesicular proteins. All the novel constituents merit investigation as vaccine candidates, and the potential immu-nomodulators merit investigation as therapeutic agents.

Molecular & Cellular Proteomics 5:835-844, 2006.
Schistosomiasis is an important parasitic disease in many parts of the tropics, second only to malaria as a source of morbidity (1,2) and mortality (2). Schistosoma mansoni, one of the three principal species of blood flukes infecting humans, inhabits the blood vessels of the hepatic portal vasculature. Infection occurs when waterborne cercaria larvae penetrate the skin after emergence from a snail intermediate host. The larvae must then migrate via the vasculature to the portal system where the worms mate and the females deposit eggs in the blood vessels of the gut wall. The eggs pass to the gut lumen and continue the life cycle or travel downstream to lodge in the liver where they initiate the pathology that is the hallmark of the disease.
The newly emerged cercaria possesses pre-and postacetabular glands (3) and a head gland (4), which are implicated in the invasion process. Synthesis of gland contents occurs during cercarial development from a germball in the daughter sporocyst located in the snail's hepatopancreas, so that all materials are in place to effect rapid host entry upon their emergence from the snail. Contact with host skin stimulates secretion from the glands (5), the contents of which provide adhesion to (6), and proteolysis of (7), the outer layers of the skin. Simultaneously the cercarial tail detaches, initiating transformation to the schistosomulum larva, a process that involves shedding of the cercarial surface membranes and their replacement with preformed material from subtegumental cell bodies (8). Some functional properties of the gland secretions have been characterized, including proteolytic activity (e.g. Ref. 9) and immunomodulatory capacity (e.g. Ref. 10). Furthermore a small number of constituents have been cloned, namely several isoforms of a serine protease, termed cercarial elastase (SmCE) (e.g. Ref. 11), and Sm16, a putative anti-inflammatory protein (e.g. Ref. 12).
Cercarial proteins released into the skin should be accessible to the immune system and thus could serve as vaccine candidates. This is an attractive proposition because the skin is the parasite's first point of contact with the host. Proteomics, which permits characterization of complex mixtures of proteins, provides an ideal approach to characterize the larval secretions. It has been applied recently to S. mansoni cercariae (13), identifying the SmCE isoforms, proteins associated with calcium function, and cytosolic proteins such as heat shock protein chaperones. We report here our detailed analysis of cercarial secretions, using 2-DE, 1 MS/MS, and biosynthetic labeling, that significantly extends the findings of the recent study. The generation of a large expressed sequence tag (EST) database for S. mansoni, including 17,810 ESTs derived from germballs (14), together with the release of the draft genome sequence (available at www.schistodb.org) has assisted greatly in our identification of several novel molecules with putative proteolytic or immunomodulatory properties.

EXPERIMENTAL PROCEDURES
Processing of Cercariae and Their Secretions-A Puerto Rican isolate of S. mansoni was maintained by routine passage through albino Biomphalaria glabrata snails and MF1 outbred mice. Shedding of cercariae from snails was stimulated by exposure to bright light with approx. 0.5 million parasites produced at each collection. Mechanical transformation by vortexing (15) was used to stimulate the release of gland cell contents during a 3-h culture period in 30 ml of RPMI 1640 medium (Invitrogen) containing 300 units/ml penicillin and 300 g/ml streptomycin at 37°C in 5% CO 2 . The medium was removed and centrifuged at 130 ϫ g at 4°C for 8 min, and the supernatant was stored at Ϫ20°C with the addition of 20 l of 20ϫ general use protease inhibitor mixture (Sigma). This soluble preparation, termed the 0 -3-h released proteins (RPs), was filter-concentrated to ϳ1 ml using a 5-kDa-molecular mass cutoff Ultrafree-15 unit (Millipore, Watford, UK), and protein content was assayed using Coomassie Plus 200 reagent (Perbio, Tattenhall, Cheshire, UK). Cercarial proteins soluble in 40 mM Tris (termed SCAP) were also prepared (16).
Production and Analysis of 2-DE Gels-The 0 -3-h RPs were separated by 2-DE, which also allowed the relative abundance of components to be determined. Before isoelectric focusing, 250 g of 0 -3-h RPs was dialyzed against 10 mM Tris overnight at 4°C. Large (18 ϫ 18 ϫ 0.1-cm) 2-DE gels of 0 -3-h RPs and SCAP were produced and stained with SYPRO Ruby (Bio-Rad), and images were captured using a Molecular Imager FX (Bio-Rad) (16). To generate an "average" 2-DE image, three separate cercarial collections were used to produce three individual gels. Phoretix 2D Evolution software (Nonlinear Dynamics, Newcastle, UK) was used to analyze each gel with the specification that a spot must be matched between at least two of the three gels for inclusion in the average image. Background subtraction and spot volume normalization (each spot being expressed as a percentage of the total spot volume on that gel) were also performed. Manual editing removed artifacts (e.g. SYPRO Ruby speckles), deleted or split any poorly assigned spots, and improved the image warping and spot matching. A SYPRO Ruby-stained SCAP gel containing 200 g of protein was also subjected to software analysis and matched to the average 0 -3-h RP image to enable the -fold difference in normalized volume (NV) between matched spots to be calculated.
In-gel Digestion and Mass Spectrometric Analysis-Spots chosen for analysis were manually excised from BioSafe Coomassie-(Bio-Rad) or MS-compatible silver (17)-stained 2-DE gels containing 250 -500 g of 0 -3-h RPs. Following in-gel digestion with trypsin (0.2 g in 40 l of 20 mM ammonium bicarbonate at 37°C for 18 h), 2 ϫ 1 l of eluted peptides were sequentially spotted onto a MALDI target plate, air-dried, and followed by 0.6 l of matrix (a saturated solution of ␣-cyanohydroxycinnamic acid (Sigma) in 50% acetonitrile, 0.1% TFA diluted to half-strength with the same solvent). Where necessary remaining peptides were desalted and concentrated using a C 18 ZipTip (Millipore). A 4700 Proteomics Analyzer with TOF-TOF Optics (Applied Biosystems, Framingham, MA) calibrated using CalMix2 (Applied Biosystems) was used to obtain MS and MS/MS spectra in positive reflector mode. For each target spot, a mass spectrum was collected between m/z 700 and 4000; where present, two trypsin autolysis peaks were used as an additional, internal calibrant. The 15 most intense peaks from each spectra were automatically selected for MS/MS, ignoring masses in the "exclusion list" (e.g. common trypsin autolysis peaks and keratin contamination (18)). Peptide CID was performed at a collision energy of 1 keV with air as the collision gas.
Database Searching-Matches to the MS/MS data were retrieved using GPS Explorer software (Applied Biosystems) underpinned by Mascot version 1.9 (www.matrixscience.com). The National Center for Biotechnology Information (NCBI) non-redundant protein database was searched first, and then a compilation of clusters, termed "SchistoCDS," was searched. The latter comprises all publicly available S. mansoni ESTs (at NCBI dbEST and The Wellcome Trust Sanger Institute, ftp.sanger.ac.uk/pub/pathogens/Schistosoma/mansoni) and gene sequences predicted from the assembled schistosome genome using three different heuristics (Snap, PHAT, and Glim-merHMM at www.schistodb.org). Search parameters allowed for one missed tryptic cleavage site, the carbamidomethylation of cysteine, and the possible oxidation of methionine; precursor and product ion mass error tolerance was Ϯ0.1 Da. A protein was considered positively identified if the GPS total ion score confidence interval exceeded 99.9%. Identified proteins were classified as vesicular in origin when their full-length sequences possessed a signal peptide as determined by SignalP (19) or by inference from their nearest homolog as determined by BLAST. PSORT II (20) was used to determine whether proteins contained any transmembrane domains. Where possible, the pattern of life cycle stage expression for each identified protein was determined by reference to the ESTs in SchistoCDS. Where matches were made to the NCBI database, BLAST analysis was performed to identify the relevant EST cluster in SchistoCDS.
Extension of Partial Gene Sequences Using Genomic Data-Where possible, when MS data retrieved hits to partial EST sequences or incomplete gene predictions, the coding sequence of the gene was extended by manual inspection of the genome (www.schistodb.org). The genomic contig was retrieved and edited in Artemis (21). Briefly where multiple partial sequences were found, these were concatenated, ensuring that the intron splice donor and acceptor sequences surrounding the exons conformed to those expected in S. mansoni. The putative coding sequence was then searched against the NCBI non-redundant database using BLAST. Where gaps were indicated in the query, the sequence retrieved by BLAST was itself searched against the genomic assembly to pinpoint any missing exons, which were then added. This process was repeated until no more exons could be found without disrupting the correct intron splice donor and acceptor sites and the single open reading frame.
protein was assessed for the presence of "PL," the preferred residues N-terminal to the scissile bond (positions P1 and P2 counting outward (11)). The program Compute pI/Mw (available at www.expasy.ch) was used to predict the size and pI of theoretical polypeptide fragments resulting from cleavage at this point. To assess whether each potential cleavage site was likely to be accessible to SmCE-1a in the folded protein, the S. mansoni amino acid sequence was modeled onto the three-dimensional structure of the most similar full-length protein available using Cn3D 4.1 software (available from www.ncbi.nlm.nih. gov/Structure/CN3D/cn3d.shtml). Uncomplexed rabbit actin (Protein Data Bank code 1J6Z) was used for modeling S. mansoni actin 2 (UniProt accession number P53471), the mature form of Leishmania major leishmanolysin (Protein Data Bank code 1LML) was used for SmPepM8, and creatine kinase from Limulus polyphemus (Protein Data Bank code 1RL9) was used for ATP:guanidine kinase (ATP:GK) (UniProt accession number P16641).
Metabolic Labeling of 0 -3-h RPs-To identify proteins synthesized in the developing sporocyst, infected snails were exposed to radiolabeled amino acids (22) using Ͼ1000 Ci/mmol [ 35 S]Cys/Met (Pro-mix L TM , Amersham Biosciences) instead of 75 Se-methionine. Three, 5, 7, and 9 days after labeling, 0 -3-h RPs were collected as described above, and the processed sample was counted in a liquid scintillation counter (PerkinElmer Life Sciences) to determine cpm/g of protein.
The incorporation of radiolabel per parasite was assessed by counting 10 aliquots of 10 cercariae with background contributions from radioactivity in the water (e.g. from labeled snail mucous) subtracted.
Detection of Metabolically Labeled Proteins and Calculation of "Specific Activity"-Labeled 0 -3-h RPs were separated by 2-DE minigel; the preparation and method was as for the large gels except a 125-l final volume was applied to each 7-cm IPG-strip (Amersham Biosciences), and an approximate total of 14-kV-h focusing was achieved on an IPGphor (Amersham Biosciences). Second dimension separation used NuPage Novex 4 -12% bis-Tris ZOOM gels and MES running buffer (Invitrogen). Minigels were stained with SYPRO Ruby, and the images were captured as described above before drying onto cellulose film (Perbio Science UK Ltd., Northumberland, UK). Dried gels were exposed to Fuji storage phosphor screens (Raytek Scientific Ltd., Sheffield, UK) for 3 days before images were captured using a Molecular Imager FX. Minigel images were analyzed using the Evolution software with the volume of each spot expressed as a sum of the pixel intensities within the spot boundary after "mode of nonspot" background subtraction. This value was used to represent the "amount of protein" in a given spot. Storage phosphorimages were manipulated and quantified using QuantityOne software (Bio-Rad) to determine the "amount of radioactivity" associated with each spot expressed as the sum of pixel values in a designated area following "lowest on boundary" background subtraction. These values were used to calculate the specific activity (S.A.) of selected spots.

Proteins of a Vesicular Origin Are Highly Enriched in the 0 -3-h
RPs-Analysis of the three replicate 0 -3-h RP preparations produced an average gel image with a relatively simple pattern of 144 spots (Fig. 1). The 50 most abundant spots, based on their NV and accounting for 79% of the total protein on the gel, were selected for MS analysis (specimen spectrum, Supplemental Fig. 1). The 44 identities obtained represented 16 separate proteins, seven of which were present as two or more spots (Table I). SignalP classified seven proteins as vesicular in origin, but none contained transmembrane domains on the basis of PSORT II analysis, which also designated the remaining nine proteins as cytosolic in origin. The vesicular proteins accounted for 54% of the normalized spot volume (or 69% of the selected spot volume) with two proteases alone, SmCE and SmPepM8 (see below for details), comprising 34.4 and 12.8% NV, respectively. The remaining  proteins with vesicular locations are all potential immunomodulators. Three novel proteins, each containing a spermcoating protein (SCP) domain (pfam accession number PF00188) and with homology to venom allergens, were identified. These were designated SmSCP_a, _b, and _c and represented 1.91, 0.61, and 0.44% NV, respectively. A further novel protein (0.56% NV) had homology to potassium channel inhibitors found in scorpion venom (named SmKK7 after its nearest homolog in Mesobuthus martensii, BmKK7). The final vesicular protein is the well characterized, putative anti-inflammatory protein Sm16 (3.38% NV). Of the cytosolic proteins identified (25% NV in total), a novel serpin (named SmSerpin_c as two serpins are already sequenced from S. mansoni; 4.89% NV) and several cyclophilins were dominant, whereas the remainder (actin, 28-kDa GST (GST28), fatty acid-binding protein (FaBP), triose-phosphate isomerase (TPI), aldolase, and thioredoxin) are ubiquitous throughout the schistosome life cycle.
A separation of SCAP (of which 0 -3-h RPs should be a subset) revealed a much more complex pattern of 1314 spots (Supplemental Fig. 2). Software comparison of the two images showed 60% of the 0 -3-h RP spots matched to those in the SCAP (data not shown). Normalized volumes for matched spots were obtained from each gel, and the -fold difference in the 0 -3-h RPs, relative to the SCAP, was calculated (Table I). Due to the lower complexity of the 0 -3-h RPs, all but one of the normalized spot volumes (actin, spot 47) were higher in this preparation than in the SCAP, but the -fold increase of vesicular protein spots (mean Ϯ S.E., 11.95xϮ 2.93) was significantly greater than that of the cytosolic protein spots (mean Ϯ S.E., 4.18xϮ 1.21; two-sample t test with pooled variances, t ϭ 2.45, adjusted degrees of freedom ϭ 24, p ϳ 0.02). There was a small number of exceptions within the cytosolic group, the two SmSerpin_c spots (spots 6 and 16) and a cyclophilin_1 spot (spot 9) whose -fold increase approached that of the vesicular proteins. Of the unidentified proteins with unknown origin, two fell into the vesicular category and one fell into the cytosolic category on the basis of their -fold increase. Four spots, not among the top 50, were highly enriched in the 0 -3-h RP sample and were therefore subjected to MS analysis. Two proved to be of vesicular origin, one an apparent high molecular weight version of SmCE (spot 68) and the other (spot 74) homologous to dipeptidyl peptidase IV (see below for details). One cytosolic protein (spot 57, ATP:GK) was found at an unexpected gel position, and a hit (spot 97) was made to an S. mansoni EST with no homology.
When the expression profile for each protein was determined by reference to SchistoCDS, two clear patterns were evident ( Table I). Proteins of vesicular origin had a restricted expression profile (e.g. SmCE, Sm16, and SmSCP_a), whereas those of cytosolic origin were present in virtually all stages (e.g. cyclophilins, actin, aldolase, FaBP, ATP:GK, and TPI). The cytosolic protein thioredoxin showed a more restricted expression pattern, whereas SmKK7 and SmDPP IV were detected in only two life cycle stages. Expression profiling was not possible where identification was made only to gene predictions from the genome.
Further Characterization of Novel Proteins-MS data were searched against predictions of coding regions derived from the S. mansoni genomic assembly. Of the seven spots designated SmPepM8, six hit three separate Snap predictions, each with homology to proteins with Peptidase_M8 domains (PF01547). Manual inspection of the genomic data identified 16 putative exons ( Fig. 2A) with the full-length sequence containing an entire Peptidase_M8 domain and a perfect metalloprotease metal-binding site; the eight distinct peptide hits were distributed throughout the sequence (Fig. 2B). A BLAST search with the full-length sequence gave an expect score of 2.00EϪ74 to an Anopheles gambiae sequence (NCBI accession number XP_311944) that also contains the Peptidase_M8 domain and an expect score of 9.00EϪ24 to the Leishmania donovani leishmanolysin (gp63) surface metalloprotease (A45621). PSORT II failed to detect a transmembrane domain at the C terminus of the predicted schistosome molecule. The inclusion of an EST contig (Sm09573) in the extended SmPepM8 sequence enabled us to determine that transcripts were present only in the germball stage.
Two EST contigs were hit for spot 74, both having homology to dipeptidyl peptidase IV; thus the protein was termed SmDPP IV. Manual inspection of the genomic data showed that they are arranged in tandem with no missing exons. The exon pattern was largely replicated by data from three independent Snap predictions (Fig. 2C). A molecular mass of b The number of fragmentation spectra that matched each identity is designated by the peptide count. Where more than one sequence was hit and subsequently shown to be part of the same gene, the accession number and peptide coverage for all hits are shown. NS, the Mascot score for the peptides matching this sequence was not significant. approximately 90 kDa is predicted from the 2-DE gel in agreement with that of the nearest homolog in Drosophila melanogaster (AAF49712; BLAST expect score of 2.00EϪ63), but the inferred coding sequence predicts a protein of only 67 kDa (Fig. 2D), and its N terminus could not be obtained. Nevertheless two domains, dipeptidyl peptidase IV N-terminal region (PF00930) and Peptidase_S9 (PF00326) were evident in common with the Drosophila protein.
It is probable that the complete coding regions of SmSer-pin_c and SmKK7 were present in the EST contigs hit by MS data. This is borne out by the presence of the entire Serpin domain (PF00079) in Snap prediction 06756, whereas the theoretical pI (9.32) and molecular mass (44 kDa) of the coding region correspond to those observed on the 2-DE gel (Fig.  1). No specific domain is available to aid further characterization of SmKK7, but comparison with its nearest homologs shows that features of those molecules are conserved (Fig. 2E). It was not possible to perform the gene reconstruction procedure for the three SCP domain-containing proteins using the current, incomplete version of the S. mansoni genome assembly.
Three SmCE Isoforms Are Present in the 0 -3-h RPs-Fourteen gel spots in the 0 -3-h RPs were identified as SmCE , and Peptidase_S9 domain (dotted line below sequence). E, ClustalW alignment of SmKK7 with BmKK7 (nearest BLAST homolog) and BeKm-1 (nearest homolog for which a crystal structure exists). Symbols beneath the sequence indicate the following: "*," identity between all residues in alignment; ":," conserved substitutions according to amino acid properties; ".," semiconserved substitutions. A vertical line indicates a putative signal peptide cleavage site. Cysteine residues are shaded and numbered to indicate those which form disulfide bonds in the scorpion molecules. Boxes around residues show those amino acids thought to be most crucial to toxin-K ϩ channel interactions (36). and appeared at the expected molecular weight, although only five genes encoding schistosome cercarial SmCEs have been described to date (Supplemental Fig. 3A). To discover whether individual spots corresponded to separate isoenzymes, we determined the theoretical masses of diagnostic tryptic peptides for the five predicted amino acid sequences; between 18 and 41 peptides were novel for each. A peptide mass fingerprint produced for each of the 14 spots revealed the presence of between two and 10 masses ostensibly corresponding to selected diagnostic peptides from all five isoenzymes. However, fragmentation of these peptides and Mascot searches with the resulting data confirmed the presence of only SmCE-1a, -1b, and -2a (Supplemental Fig. 3B).

Fragments of Proteins in the 0 -3-h RPs May
Result from SmCE Action-Three 0 -3-h RP spots, identified as actin, SmPepM8, and ATP:GK, were present in gel positions that were not consistent with their predicted pI and below their expected molecular mass. However, because parasites were cultured in the absence of protease inhibitors (to preserve viability) the anomalous spots could result from proteolytic cleavage. We located PL residues, the cut site for SmCE, in the full-length sequences of all three proteins and calculated the theoretical masses and isoelectric points of the fragments that would result from cleavage. Digestion of actin at the single PL site would give fragments of molecular mass 29.5 kDa, pI 5.5 and molecular mass 12.5 kDa, pI 5.05 (Fig. 3), the first of which corresponds exactly with the observed position of spot 24 in Fig. 1. Furthermore, mapping of S. mansoni actin onto the crystal structure of uncomplexed rabbit actin (91% sequence identity) showed the cleavage site to be present in an exposed surface loop (Supplemental Fig. 4). The situation was less clear for SmPepM8 and ATP:GK where the lower sequence identity (26 and 45%, respectively) to proteins with determined structure meant that a surface location for appropriate PL cut sites was harder to ascertain.
Released Proteins Display Varying Patterns of Synthesis-The level of [ 35 S]Cys/Met incorporated into intact cercariae and 0 -3-h RPs was determined when they were collected 3, 5, 7, and 9 days after labeling (Fig. 4A). Cercariae exposed 3 days prior to shedding had the lowest incorporation with levels 4 -6 times higher at the three later times. However, incorporation of [ 35 S]Cys/Met into the 0 -3-h RPs was lowest at the day 9 collection with levels up to 4 times greater prior to this. A 25-g aliquot of 0 -3-h RPs from each time point was separated by a 2-DE minigel. The images of stained proteins were consistent between samples, whereas the phosphorimages of radioactivity exhibited marked differences in spot pattern (Fig. 4B). Three distinct patterns of biosynthesis were revealed by the determination of S.A. (Fig. 4C) with maxima before cercarial release from the snail at 3 days (SmPepM8 only), 5 days (SmSerpin_c, TPI, and SmSCP_b), and 7 days (Sm16, SmSCP_c, actin, enolase, and SmCE). No individual protein had its highest level of radioactivity incorporation at the 9-day sampling point. The magnitude of the maximum S.A. values varied between 13 (SmSerpin_c) and 65 (enolase). DISCUSSION We undertook an in-depth analysis of the proteins released during transformation of S. mansoni cercariae into schistosomula. We are confident that the 0 -3-h RP preparation, generated by mechanical induction of transformation, faithfully represents the gland contents rather than proteins leaking from dying and dead parasites. Viable larvae are produced, 95% of which develop to the lung stage in vitro and are capable of normal migration and maturation if surgically transplanted into mice (23). This contrasts with an alternative method involving exposure of cercariae to skin lipids in water that leads to very high mortality levels (up to 100% (24)) with an inevitable artifactual contribution of somatic proteins to the preparation. It may explain the marked differences in the findings reported here with those of the recent "shotgun" proteomics study by Knudsen et al. (13). In addition, we also searched the S. mansoni genome database, and many of our identifications are derived from that rather than EST databases. Significant enrichment for vesicular proteins was achieved in our 0 -3-h RPs, including the known SmCE isoforms and Sm16, compared with a soluble protein extract of cercariae.
The function of the 0 -3-h RPs is to facilitate host entry, and there are numerous accounts of protease activity in the cercarial secretions (11). The SmCE variants were the most abundant proteins with three of the five isoforms (SmCE-1a, -1b, and -2a) identified as in previous investigations (11,13). We did not find cysteine proteases, reported as present on the basis of immunocytochemical localization (25), but we added two further proteases, SmPepM8 and SmDPP IV. SmPepM8, a metalloprotease of the leishmanolysin family, was the second most abundant constituent, a surprising finding because proteases of this class have only been detected in older schistosomula (26). Membrane-bound leishmanolysin enhances passage of the protozoan parasite Leishmania across the skin extracellular matrix and is also capable of degrading basement membrane components (27). These structures are encountered by S. mansoni larvae traversing the skin, so a similar role for SmPepM8 seems likely. Although we found no C-terminal membrane anchor in its putative full-length sequence, secreted forms of homologous leishmanolysin play a role in host invasion by Leishmania (28). The second protease DPP IV is primarily a membrane-bound protease (29), but no transmembrane domain was found in the S. mansoni version. However, an anchoring domain could be missing from our putative sequence that lacks the N-terminal region. The functions of DPP IV in other species imply that it may be involved in skin invasion by S. mansoni. Thus, the mammalian version, located within "invadopodia," assists cell migration through connective tissue during wound repair (30) and cancer metastasis (31), whereas a secreted form of DPP IV in Aspergillus fumigatus has a potential role in lung invasion by this fungus (32). The identification of these two proteases suggests that elastases are not the sole agents used by cercariae to invade the skin.
Proteins with immunomodulatory potential were the second most common class identified. The 0 -3-h RPs have been demonstrated to drive Th2 responses via dendritic cells and the T-cells they polarize (33), but the molecules responsible are unknown. The three proteins containing SCP domains with their similarity to venom allergens (34) could be involved, but their function in any organism has yet to be elucidated. The crystal structure of one such protein from Necator americanus shows it bears similarities to chemokines, perhaps explaining its immunomodulatory potential (35). SmKK7, another protein identified, has homology to a scorpion toxin that binds to and blocks K ϩ channels; the activity of the scorpion toxin is dependent on the position of several key amino acids within a three-dimensional structure that is stabilized by three disulfide bridges (36). The level of conservation between these important residues in the schistosome and scorpion proteins suggests a similar function for the schistosome molecule.
Because T-cell activation is known to be regulated by K ϩ channels (37), it is plausible that the invading schistosome modulates the immune response of the host by this mechanism. It is notable that some SCP domain-containing proteins may also block voltage-gated K ϩ channels (38).
Eleven cytosolic proteins were identified in the 0 -3-h RPs, probably the result of holocrine secretion by the cercarial acetabular glands in which the entire cell contents are forced though the ducts by muscular contraction. Most of the gland cell is occupied by vesicles with cytoplasm confined to the periphery (6), explaining the low abundance of cytosolic proteins compared with those in the whole cercarial extract, with two exceptions. The first, SmSerpin_c serine protease inhib- FIG. 4. A, incorporation of [ 35 S]Cys/Met into intact parasites (cpm/ cercaria) and into 0 -3-h RPs (cpm/g of protein). B, two-dimensional separation of 0 -3-h RPs metabolically labeled "in-snail" 3-9 days previously visualized by SYPRO Ruby and phosphorimaging. Gel spots are designated as follows: 1, SmPep8; 2, SmSerpin_c; 3, TPI; 4, SmSCP_b; 5, Sm16; 6, SmSCP_c; 7, actin; 8, enolase; 9, SmCE. C, specific activity (amount of radioactivity per unit of protein in a spot) was calculated at the time points after labeling using data acquired from analysis of the images in A as amount of radioactivity/amount of protein (units are arbitrary for both terms).
itor, is unlikely to be needed for protection of the gland cells because its potential SmCE targets are expressed as inactive proenzymes, but it could limit damage to the host or parasite upon release. The relative abundance of the second protein, cyclophilin, may be explained by a high demand for its peptidyl-prolyl cis-trans isomerase activity during vesicle content synthesis. Although GST28 was not enriched in the gland cells compared with the whole cercarial extract, its proposed prostaglandin-D 2 synthase activity (39) provides a biochemical basis for the observed eicosanoid production during skin penetration (40).
Three broad patterns of 0 -3-h RP synthesis could be categorized after in-snail pulse labeling with [ 35 S]Cys/Met. The height of each specific activity peak is dependent on the rate of protein synthesis during the short labeling period with a single dominant peak indicating a narrow window. Vesicular proteins, packaged soon after translation, would not be expected to turn over and so should retain a high S.A., whereas this is less likely for the cytosolic proteins. Because proteins within a vesicle have been synthesized at the same time, the three patterns of incorporation could reflect distinct cellular compartments, perhaps corresponding to the pre-and postacetabular glands and the head gland. Alternatively they could represent three vesicle types within a given gland (6,41). Only immunolocalization of the various products we identified will answer this question.
We believe that our study highlighted the principle proteins used by the parasite to enter and establish itself in the host. Many of the vesicular proteins are glycosylated, and we will report on the composition of the glycan residues in a separate publication. 2 Our next major task is to clone and express the various proteins to determine their functional properties in vitro and in vivo. This information may provide both potential candidates for a S. mansoni vaccine and novel immunomodulatory compounds with therapeutic value.