Schistosome Invasion of Human Skin and Degradation of Dermal Elastin Are Mediated by a Single Serine Protease*

Aquatic larvae (cercariae) of the trematode parasite Schistosoma mansoni rapidly penetrate human skin by degrading host proteins including elastin. Two serine proteases, one chymotrypsin-like and the second tryp-sin-like, have been proposed to be involved. To evaluate the relative roles of these two proteases in larval invasion, both were purified, identified by sequence, and then biochemically characterized. The trypsin-like activity was resolved into two distinct serine proteases 76% similar in predicted amino acid sequence. Southern blot analysis, genomic polymerase chain reaction, and immunolocalization demonstrated that the trypsin-like proteases are in fact not from the schistosome, but are released with larvae from the snail host Biomphalaria glabrata . Invasion inhibition assays using selective inhibitors confirmed that the chymotrypsin-like protease is the enzyme involved in skin penetration. Its ability to degrade skin elastin was confirmed, and the three sites of cleavage within elastin help define a new family of elastases.

Aquatic larvae (cercariae) of the trematode parasite Schistosoma mansoni rapidly penetrate human skin by degrading host proteins including elastin. Two serine proteases, one chymotrypsin-like and the second trypsin-like, have been proposed to be involved. To evaluate the relative roles of these two proteases in larval invasion, both were purified, identified by sequence, and then biochemically characterized. The trypsin-like activity was resolved into two distinct serine proteases 76% similar in predicted amino acid sequence. Southern blot analysis, genomic polymerase chain reaction, and immunolocalization demonstrated that the trypsin-like proteases are in fact not from the schistosome, but are released with larvae from the snail host Biomphalaria glabrata. Invasion inhibition assays using selective inhibitors confirmed that the chymotrypsin-like protease is the enzyme involved in skin penetration. Its ability to degrade skin elastin was confirmed, and the three sites of cleavage within elastin help define a new family of elastases.
Infection of a human host by the trematode parasite Schistosoma mansoni begins with invasion of intact skin by an aquatic larva, the cercaria (1). Exiting the freshwater snail Biomphalaria glabrata, cercariae locate a human host by thermal (2) and chemical (3) signals and rapidly penetrate the skin, entering the vascular system in the dermis (4). S. mansoni cercariae are ϳ150 m long and 70 m wide and require lysis of skin tissues to migrate into blood vessels. Host macromolecules representing barriers to cercarial invasion are known to be cleaved by proteolytic activities present in cercarial secretions. These include elastin (5); chondromucoprotein (6); keratin (7); fibronectin, laminin, and collagens IV and VIII (8). Two distinct serine proteases have been reported in extracts of cercariae or in secretions from cercariae induced upon contact with skin lipids. One is a "chymotrypsin-like" protease with a preference for large hydrophobic side chains at P1 (9). The second is a "trypsin-like" protease with a preference for positively charged side chains at P1 (10). The class of proteases responsible for host protein degradation has been demon-strated by several independent studies to be serine proteases (6), but the relative contributions to invasion of the trypsin-like or chymotrypsin-like proteases are not known.
To analyze the relative contributions of each of these proteases to the degradation of host proteins, cercarial secretions were fractionated, and the two proteases purified. The trypsinlike activity, which had not been previously purified or sequenced, was purified, and a cDNA was cloned by reverse transcription-PCR 1 based on amino-terminal amino acid sequence. Specific inhibitors were identified to evaluate the role of each protease in skin invasion. Proteases were localized by immunohistochemistry, and specific sites of cleavage in elastin, the most protease-resistant target in host skin, were analyzed.

Biologic Materials
S. mansoni-Approximately 5 ϫ 10 5 cercariae (Puerto Rican strain) were collected in 750 ml of distilled water from 200 -300 infected B. glabrata snails using a light induction method previously reported (11). Cercarial secretions were collected using a modification of a previously reported technique (12). The cercariae were placed in Petri dishes coated with linoleic acid (to simulate skin contact) and floated in a 37°C water bath to produce a thermal gradient. After 2 h, the cercariae had released the majority of their gland contents, and the conditioned water was collected and filtered to remove cercarial bodies and debris. The secretion sample was then lyophilized and stored at ϩ4°C until further purification.
B. glabrata-Snails were maintained with a diet of organic lettuce and school chalk as a calcium supplement (13). All snails were housed in the absence of light to increase yields of cercariae during light induction.
Escherichia coli-BL21 cells (Novagen, Madison, WI) were electroporated with the appropriate plasmid construct and selected overnight using LB medium/ampicillin (50 g/ml) plates. A single colony was picked and grown overnight in 4 ml of LB medium with 100 g/ml ampicillin. 1 ml of overnight culture was added to 1000 ml of LB medium with 100 g/ml ampicillin and grown to A 600 ϭ 0.6. Isopropyl-␤-D-thiogalactopyranoside was then added to 100 M; and after 4 h of continued incubation, the cells were pelleted and frozen at Ϫ70°C.

Ion Exchange Chromatography
Fractions from the benzamidine affinity column were diluted 10-fold into running buffer (20 mM Tris, pH 8.0) and pumped through an HR 5/5 Mono-Q column at 1 ml/min. The column was then washed with an additional 10 ml of running buffer. Elution was performed over 30 1-ml fractions using running buffer and a linear salt gradient from 0 to 1.0 M NaCl.

Nickel(II)-Nitrilotriacetic Acid Affinity Chromatography
E. coli pellets from 330 ml of inducted culture were thawed and resuspended in 5 ml of binding buffer (8 M urea and 50 mM Tris, pH 8.0). Cell disruption was completed with six rounds of 20-s sonications on ice. Histidine-tagged proteins (QIAGEN Inc., Valencia, CA) were then purified according to the manufacturer's instructions using 1-ml spin columns.

Enzyme and Inhibitor Assays
All assays were performed with native enzymes in buffer with 100 mM glycine, pH 9.0 (the pH optimum with small synthetic substrates), and 100 M pNA substrate at room temperature unless otherwise noted. Inhibitors were tested in the same buffer system under the same conditions. A 15-min preincubation of inhibitor with enzyme was carried out before substrate was added. A standard assay consisted of 10 l of sample and 100 l of assay buffer in 96-well nylon plates (Falcon polyvinyl chloride plates, Becton Dickinson Labware, Franklin Lakes, NJ). Absorbance was monitored using a UV-Max spectrophotometer and SoftMax Version 2.02 software (Molecular Devices, Sunnyvale, CA).

Skin Invasion Assays
Skin invasion assays were performed as described previously (11). Briefly, human skin samples were fixed on plastic wells over warm medium (RPMI 1640) while 200 l of inhibitor solution (2 mM in Me 2 SO) or controls, including Me 2 SO, were applied to the skin surface and allowed to permeate and dry for 30 min. Approximately 3000 cercariae in 3 ml of water were then applied using 15-mm plastic cylinders on the skin. After 120 min, the skin was fixed in 10% phosphate-buffered Formalin, embedded in paraffin, sectioned at 7 m, and stained with hematoxylin and eosin. Cercariae penetrating the skin were counted as described previously (11). Multiple sections of three separate skin samples were counted for each inhibitor and control.

Protein Blotting and N-terminal Peptide Sequencing
NuPage-polyacrylamide gradient gels (Invitrogen, Carlsbad, CA) were used according to the manufacture's instructions. Proteins were blotted onto polyvinylidene difluoride membranes using the Novex transfer system. Peptide sequencing was carried out using an ABI Procise 491 Protein Sequencer (Applied Biosystems, Inc., Foster City, CA) at the Biomolecular Resource Center of the University of California (San Diego).

Nucleic Acid Purification and cDNA Synthesis
Poly(A) ϩ mRNA was isolated by poly(T) affinity chromatography (Amersham Pharmacia Biotech) from the hapatopancreases of five infected snails. RNA was converted to cDNA using avian myeloblastosis virus reverse transcriptase (Life Technologies, Inc.) as described by the manufacturer. DNA isolation from snails required CsCl purification because of the high proportion of glycoproteins. DNA was isolated from S. mansoni by standard detergent lysis and phenol/chloroform extraction.

Cloning and Plasmid Construction
A nested PCR strategy was used to clone both BgSPs. Amplification conditions for the first PCR were two cycles of 94, 40, and 72°C each for 1 min and then 35 cycles of 94, 50, and 72°C each for 1 min. The second PCR conditions were 25 cycles of 94, 50, and 72°C each for 1 min. The  primers used for these reactions were as follows: BgSP-␣1 forward, 5Ј-ATCGTCGGNGGNAARGARTCNATGC-3Ј; BgSP-␤1 forward, 5Ј-AT-GGTCGGWGGWCARGARGCNGTNC-3Ј; BgSP-␣2 forward, 5Ј-CC-NAAYAAYCAYAWNTGYGG-3Ј; and BgSP-␤2 forward, 5Ј-GCGCCN-ACNCAYCAYTTYTGYGG-3Ј. The reverse primer for both the first and second reactions was oligo(dT) 15 . The E. coli expression construct pET21a-BgSP-␤ was assembled by inserting the active portion of BgSP-␤ into the restriction sites NdeI and XhoI of the vector pET21a (Invitrogen). The forward primer 5Ј-CGC-CATATGGTCGGWGGWCARGARGCNGTNC-3Ј created an NdeI site (underlined) and changed the first amino acid to a start codon (Ile 3 Met). This substitution creates an inactive protease. The reverse primer 5Ј-GCGCTCGAGTCTGTTGATGACGGTGTTA-3Ј added an XhoI site (underlined) and deleted the stop codon, creating an open reading frame in the pET21a vector that added a 6-histidine tag to the C terminus of the protein.

Southern Blotting
Genomic DNA (10 g) was digested overnight with the individual restriction enzymes EcoRI, PstI, XbaI, and XhoI (New England Biolabs Inc., Beverly, MA). The DNA was transferred to a Hybond-N ϩ charged nylon membrane (Amersham Pharmacia Biotech) using alkaline capillary transfer. The DNA was fixed to the membrane by UV cross-linking and prehybridized for 45 min in Rapid-Hyb solution (Amersham Pharmacia Biotech). 40 ng of probe (active full-length PCR product) were labeled with 32 P (PerkinElmer Life Sciences) using a High Prime nick translation kit (Roche Molecular Biochemicals) according to the manufacturer's instructions. Hybridization was carried out at 65°C for 1 h. 2ϫ SSC stringency washing was sufficient for the removal of nonspecific probe. The membrane was exposed for 3 days at Ϫ70°C to Hyperfilm (Amersham Pharmacia Biotech) using intensifying screens. The membrane was then stripped for 30 min in 0.4 M NaOH at 45°C, washed twice for 10 min at room temperature in strip solution (200 mM Tris-HCl, pH 7.0, 0.1ϫ SSC, and 0.1% (w/v) SDS), and reused.

Antibody Production, Immunohistochemistry, and Fluorescent Microscopy
Recombinant His-tagged protein from E. coli was used to produce rabbit antiserum using standard commercial procedures (Corvance, Richmond, CA). Materials prepared for immunohistochemistry were incubated in fixative (2% paraformaldehyde, 1% glutaraldehyde, and 0.1 M phosphate buffer, pH 7.4) for Ͼ2 h and then embedded in JB4 plastic; 7-m sections were cut; and antibody localization was visualized using an ABC kit (Vector Labs, Inc., Burlingame, CA).

Identification of Trypsin-like Activity within Cercarial
Secretions-To identify and isolate a previously reported trypsinlike (10) activity from cercariae, gel filtration was chosen as the first chromatography step to fractionate the complex protein mixture of cercarial secretions (Fig. 1). The chromogenic substrate LGR-pNA was used to detect the trypsin activity, and AAPF-pNA was used to detect the chymotrypsin-like activity (SmCE). The trypsin activity was identified and resolved from the SmCE activity, and although it was 30-fold more active than SmCE against its small tripeptide substrate, it had no significant activity against macromolecular elastin.
Using molecular mass standards as a reference, the trypsinlike activity had an apparent molecular mass of 25 kDa, whereas the SmCE activity peak eluted at an apparent molecular mass of 15 kDa. SmCE has a pI Ͼ9.0 and may interact with the column resin, retarding its migration relative to other proteins. At higher salt concentrations, the peak of SmCE was shifted to the left (data not shown).
Purification of Native Trypsin Activity-To further characterize this native trypsin activity, benzamidine conjugated to Sepharose was employed as a first step affinity column, yielding a 1000-fold purification (Table I). Ion exchange chromatography was employed as the second and final step. The trypsin activity was resolved into two distinct proteases. Sufficient material was generated to obtain unambiguous N-terminal sequences (Fig. 2) for both peaks of activity. These two activities were designated BgSP-␣ and BgSP-␤, corresponding to the order in which they eluted from the column.
Biochemical Characterization of BgSP-␣ and BgSP-␤-As shown in Table II, both native BgSP-␣ and native BgSP-␤ have a very similar pH optimum and calcium optimum. BgSP-␤ was significantly more effective at degrading Azocoll (denatured collagen) than BgSP-␣. BgSP-␤ cleaved several different peptide substrates that have a charged amino acid in the P1 position, whereas BgSP-␣ preferred the substrate LGR-pNA.
Cloning of BgSP-␣ and BgSP-␤ cDNAs by Reverse Transcription-PCR-Based on N-terminal sequence data, a set of nested degenerate PCR primers was designed for each protease. Reverse transcription-PCR was performed on mRNA isolated from the hepatopancreases of infected snails. This is the location where the parasite replicates and develops. A SmCE cDNA had previously been cloned from this source (14). Once the cercaria leaves the host snail, there is relatively little transcription or translation until after it enters the human host.
The nested primers in conjunction with a poly(A) reverse primer yielded two bands of the predicted size. Cloning and sequencing of these two bands revealed two serine proteases of the chymotrypsin-fold family that are 76% similar (Fig. 3). The amino termini matched those of the purified proteases, and the S1 subsites were consistent with trypsin-like serine proteases. Both S1 pockets contained an aspartic acid at the position corresponding to trypsin 195d. A BLAST (15) search confirmed that both sequences are unique, but similar to other serine proteases with a chymotrypsin tertiary structure. BgSP-␣ has a putative glycosylation site at Asn 51 . The utilization of this glycosylation site is consistent with the observed molecular mass determined by SDS-polyacrylamide gel electrophoresis. BgSP-␣ has an apparent molecular mass 2-3 kDa larger than that of BgSP-␤, yet their calculated molecular masses are  4. Southern blot of S. mansoni, B. glabrata, and B.  BgSP-␣ and BgSP-␤ Are Produced by the Host Snail B. glabrata-Southern blot analysis showed that both protease sequences are present only in the genome of the intermediate host snail (Fig. 4). Additionally, genomic PCR using B. glabrata DNA produced a fragment that contained both the protease sequence and the sequence of an intron with the consensus sequence for an acceptor site at the predicted junction (data not shown).
Localization of BgSP-␤ and SmCE by Microscopy-Polyclonal rabbit antiserum was generated against recombinant BgSP-␤ produced in E. coli and purified using a 6-histidine tag. Specific localization of the antiserum was visualized in granules produced by the epithelial cells in the hepatopancreas of the snail. No staining was seen at any stage of cercarial development or in or on free swimming cercariae. In contrast, polyclonal antiserum reactive against SmCE, generated with recombinant SmCE produced in E. coli, localized only within the secretory glands of the cercaria and did not stain the adjoining snail tissue (Fig. 5).
Similarity of BgSPs to Other Serine Proteases of Mollusca-A search of the gene data base revealed only five proteases that have been identified within the phylum Mollusca (NCBI accession number txid6447). Three of these proteases belong to be furin/Kex family of serine proteases (Lymnaea stagnalis (Gen-Bank TM /EBI accession numbers AF140362 and AF140361) and Helix aspersa (AF107213)), a structurally unrelated serine protease (i.e. a protease with a different protein fold). The two remaining proteases belong to the chymotrypsin family (B. glabrata (AA547777) and Haliotis rufescens (X71438)). The amino acid similarities between BgSP-␣, BgSP-␤, and these two family members range from 35 to 43%.
BgSP Production under Various Conditions-Two conditions were tested to determine what factors may influence the production of BgSPs within the snail. The total activity of BgSPs within the hepatopancreas of the snail was not affected by feeding. No significant differences were seen over an 8-h time course of feeding following 24 h of starvation. Total activity within the hepatopancreas was affected by schistosome infection. The displacement and diminution of the snail hepatopancreas by the dividing parasite caused a "crowding out" effect and reduced the total activity of the BgSPs on average 10-fold (data not shown).
Identification of Inhibitors for SmCE and BgSPs for Use in "Chemical Knockout"-The tripeptide inhibitor FPR-CMK was highly effective against the pooled native BgSPs (IC 50 ϭ 0.9 nM) and also had some activity against native SmCE (IC 50 ϭ 1700 nM). In contrast, AAPF-CMK was effective against SmCE (IC 50 ϭ 1200 nM), but had no measurable activity against the BgSPs (IC 50 Ͼ 10,000 nM) (Table III). To evaluate the role of these proteases in the invasion process, an in vitro human skin invasion assay was used (9,11). AAPF-CMK was effective in reducing the number of cercariae penetrating by Ͼ80% (Fig. 6), whereas FPR-CMK produced a 50% reduction in cercarial penetration.
Relative Rates of Elastin Degradation by BgSP, SmCE, and Chymotrypsin-Elastin is the skin protein most resistant to proteolysis. It is a major component of the dermal barrier to cercarial invasion (16,17). Native BgSP, native SmCE, and sequence-grade bovine chymotrypsin were compared on a molar basis against insoluble native bovine elastin. Only SmCE demonstrated significant elastase activity (Fig. 7).

DISCUSSION
S. mansoni cercariae directly penetrate human skin within 5-10 min after contact (19). During this process, 10 glands that compose ϳ30% of the volume of a cercaria release their contents. Previous studies have shown that these glands contain and release proteolytic activity (12,14) and that cercarial invasion is a lytic event, not solely a mechanical event (20).
Two proteases were proposed to be involved in skin invasion. The first, a chymotrypsin-like serine protease also known as SmCE, cleaves a variety of human skin macromolecules and has been studied in detail (9,14,21,22). The second, a trypsinlike serine protease, has only recently been identified through the use of small synthetic substrates (10). Secretions from cercariae were collected and assayed for trypsin-like activity using the substrate LGR-pNA. A purification strategy was developed, and the activity was found to result from two similar serine proteases designated BgSP-␣ and BgSP-␤. 2 R. P. Mecham, personal communication.

FIG. 5. Immunolocalization of proteases within the infected host B. glabrata.
A shows anti-cercarial elastase serum (identified by the orange-brown color of the peroxidase reaction) localizing within the pre-acetabular glands (PG) of a cercaria (SC with bar) developing within a daughter sporocyst that is within the host snail. B shows anti-B. glabrata serine protease serum reacting within vesicles of the host snail secretory epithelium (HE), but not within a parasitic cercaria (SC with bar). Solid bars ϭ 50 m. Southern blot analysis and genomic PCR both demonstrated that the trypsin-like proteases are of snail host origin. The predominance of single bands and their similar arrangements between the two blots suggest that both genes are present in single copies and may be contiguous. Immunohistochemistry localized the protease to secretory vesicles within the epithelium of the snail hepatopancreas. The function of the trypsins in the snail is likely digestion. The trypsin-like activity is released by the snail during the release of cercariae and would therefore contaminate preparations of cercariae or their secretions. Expression of protease activity in the snail was "constitutive" without induction or enhanced release during feeding. However, less activity was found in infected snails due to replacement of the snail tissue by the developing parasites.
Elastin is one of the most difficult host macromolecules to degrade, and few proteases are "true" elastases capable of degrading native insoluble elastin (23). To gain insight into the mechanism of elastin degradation by parasite larvae, the site of cleavage by SmCE was determined. Solubilized fragments from native bovine elastin degraded by SmCE were purified and sequenced. Three cleavage sites localizing to two exons (exons 12 and 13, amino acids 202-239) within bovine elastin were discovered.
To confirm the role of SmCE in the invasion process and to rule out any possible contribution of BgSP (e.g. being passively carried to skin by cercariae), a human skin invasion assay was used in combination with selective protease inhibitors. One inhibitor (AAPF-CMK) inhibited SmCE alone, whereas the other (LGR-CMK) inhibited both the proteases, but was 1900fold more effective against BgSP. If BgSP contributed to invasion, FPR-CMK should have the predominant effect. If SmCE alone were responsible, both inhibitors should be effective, but AAPF-CMK would be more active. AAPF-CMK inhibited 80% of cercariae invasion, whereas LGR-CMK inhibited 50%. This is the expected result if SmCE was acting alone to facilitate cercarial invasion. SmCE is a biologically potent histolytic protease with activity against many of the macromolecular barriers of skin (5)(6)(7)(8). It is one of the few true elastases capable of degrading insoluble elastin itself. Its cleavage pattern on elastin suggests that exons 12 and 13 (amino acids 202-239) are solvent-accessible domains. Exon 12 is also targeted by macrophage elastase, and the observations reported here support the conclusion of Mecham et al. (23) that "elastases" are a more diverse family of proteases than originally thought. Insoluble elastin can be degraded by proteases acting outside the alanine-rich regions targeted by mammalian pancreatic elastases.  (24). Lowercase letters represent the sequenced portions of the peptide fragments generated (described under "Experimental Procedures").