We present evidence supporting novel collaborations between the serine protease inhibitor (serpin) and the trefoil factor during the budding stage of the tunicate Polyandrocarpa misakiensis. Using a maltose-binding protein/P-serpin fusion protein, two polypeptides of 40 kDa and 45 kDa were pulled down from Polyandrocarpa homogenates. Based on their partial amino acid sequence data, a single cDNA (928 bp) was cloned. It encodes a polypeptide that has five tandem repeats of a trefoil consensus motif. Thus, we termed the cDNA P-trefoil. Both P-trefoil and P-serpin were expressed exclusively by coelomic cells during budding. P-Trefoil was expressed mainly by coelomic cells throughout the asexual life cycle of Polyandrocarpa, while P-Serpin was localized particularly in coelomic cells and in the extracellular matrix in developing buds. The native P-Trefoil protein showed aminopeptidase activity. It induced cell growth in cultured Polyandrocarpa cells at a concentration of 8 µg/mL. P-Serpin reinforced this activity of P-Trefoil. Further, a mixture of P-Trefoil and P-Serpin exhibited the in vitro induction of a gut-specific alkaline phosphatase. These results show for the first time that a serpin can interact with a trefoil factor to play a role in the cellular growth and differentiation of the gastric epithelium.
Budding, which is a mode of asexual reproduction, affords evidence that a limited number of somatic cells are still capable of forming new individuals. It is adopted mainly by aquatic organisms in the animal kingdom including those in the phyla Porifera, Cnidaria, Platyhelminthes, Kamptozoa, Annelida, Bryozoa, Echinodermata, Hemichordata (Pterobranchia) and Chordata (Urochordata) (Brien 1968). In these animals, particular types of somatic cells have considerable developmental plasticity that enables cell growth, cell differentiation and morphogenesis. In some cases, these are undifferentiated stem cells defined by several criteria such as self-renewal, asymmetric division and irreversible differentiation (Lajtha 1979; Potten & Morris 1988; Hall & Watt 1989). In other cases, differentiated cells maintain multipotency. In the budding tunicate Polyandrocarpa misakiensis, endodermal differentiated cells (atrial epithelium) enter the cell division cycle during the early phase of budding (Kawamura & Nakauchi 1986). These cells then transdifferentiate into many organ placodes, for example, those of the brain, the pharynx and the gut (Kawamura & Fujiwara 1994). Our goal is to reveal the mechanisms by which multipotent somatic cells give rise to organs.
Previously, we constructed a catalog of genes that are expressed during budding in P. misakiensis (Kawamura et al. 1998). The expression of the serine protease inhibitor (serpin) in animals in the budding stages was fivefold that observed in the prebudding animals. However, there is no information on the type of cells that express Polyandrocarpa serpin (P-serpin) and the role that P-Serpin plays during budding. In mammals, serpins are responsible for regulating a variety of proteolytic processes. They bind to serine proteases to induce cell proliferation (Miyata et al. 2002) and to regulate apoptosis (Fell et al. 2002). In some cases, serpins function as tumor suppressor proteins (Blacque & Worrall 2002), and in other cases they facilitate the migration and invasion of tumor cells (Arroyo De Prada et al. 2002). Although it is known that P-Serpin inhibits the enzymatic activities of trypsin and elastase in P. misakiensis, little is known about the molecular partner of P-Serpin (Kawamura et al. 1998).
Trefoil peptides are small and stable molecules that have highly conserved cysteine-rich domains consisting of 45 amino acids. They are assumed to form a three-lobed leaf by the formation of six half-cystines (Chinery & Coffey 1996). Spasmolytic polypeptide (SP) inhibits gastrointestinal motility and gastric acid secretion. In mice, pS2 and intestinal trefoil factor (ITF) serve as negative or positive regulators of cell growth and are involved in gastrointestinal maintenance and repair (Lefebvre et al. 1996; Mashimo et al. 1996; Playford et al. 1996). In such cases, the trefoil factor interacts cooperatively with a mucin glycoprotein (Kindon et al. 1995).
We happened to observe a molecular interaction between P-Serpin and P-Trefoil in P. misakiensis. In this paper, we first present biochemical evidence for the binding of P-Serpin with P-Trefoil. Next, we describe the unique structure of P-Trefoil that has five tandem repeats of the trefoil motif and a single thiol protease consensus-like sequence. In fact, we present evidence supporting the aminopeptidase activity of P-Trefoil. We then reveal the spatiotemporal expression of the P-serpin and P-trefoil genes and their corresponding proteins. Finally, we provide in vitro evidence demonstrating that the P-Serpin/P-Trefoil complex serves as a cell growth-promoting factor and a cell differentiation-inducing factor in P. misakiensis. The results are discussed in the context of the instructive signal for the differentiation of multipotent endodermal cells into the gastric epithelium in budding tunicates.
Materials and methods
Asexual colonies (W strain) of P. misakiensis (Watanabe & Tokioka 1972) were reared in culture boxes settled in the Uranouchi Inlet near the Usa Marine Research Center, Kochi University.
A cDNA for P-serpin was ligated in flame to maltose-binding protein (MBP) expression vector (pMAL-C2X, BioLab) at sites of SalI and HindIII. Bacteria were induced to express MBP-serpin fusion protein by 0.1 mm isopropylthio-β-D-galactopyranoside for 18 h at 27°C. They were collected and sonicated in phosphate-buffered saline (PBS). After centrifugation, the supernatant (about 20 mL) was mixed with amylose resin (300 µL) for more than 6 h at 4°C. After washing twice with PBS, the recombinant protein was eluted with 0.1% maltose.
Colonies (12 g) of P. misakiensis were powdered in liquid nitrogen and extracted with 20 mL of 10 mm phosphate buffer (pH 7.0). After centrifugation at 12 000 g for 15 min, extracts were equilibrated with 0.1 m ammonium acetate and applied to a gel filtration column (3 cm × 100 cm; Ultrogel; Sepracor, Villeneuve, France), as described elsewhere (Kawamura et al. 1991). For anion exchange chromatography, samples were equilibrated with 20 mm Tris-HCl (pH 8.0), applied to a column of 16 mm × 40 mm (DEAE Toyopearl-650M; Tosoh, Tokyo, Japan) and eluted under a linear gradient of 0–0.5 m NaCl, as described elsewhere (Matsumoto et al. 2001).
For MBP-P-serpin affinity chromatography, samples were equilibrated with PBS, and then mixed overnight with MBP-serpin at 4°C. They were applied to a amylase resin column, washed thoroughly with PBS, and eluted with 0.2% maltose. For immunoaffinity chromatography, a HiTrap protein G column (0.7 cm × 2.5 cm; Amersham, Little Chalfont, Buckinghamshire, UK) was equilibrated with 20 mm phosphate buffer (pH 7.0). The void volume (1 mL) after gel filtration chromatography was equilibrated with the same buffer and mixed with anti-P-Serpin antiserum (0.1 mL) before applied to the column. After thorough washing, the column was eluted with the elution buffer (0.1 m glycine-HCl, pH 2.7).
Measurement of protease activity
After chromatography, aliquots (10 µL) of each fraction were diluted with 240 µL of 50 mm phosphate buffer (pH 7.0). A synthetic substrate for aminopeptidase, leu-4-methylcoumaryl-7-amide (MCA; Peptide Institute, Osaka, Japan) was dissolved in dimethylsulfoxide as 10 mm stock solution. This was added to the solution mentioned above in a final concentration of 0.1 mm. Proteolytic digestion and subsequent measurement of the activity were performed, as described elsewhere (Kawamura et al. 1998; Ohashi et al. 1999). One unit is the proteolytic activity to cleave 10−7M substrate for 1 h. The amount of protein was determined by the method of Lowry et al. (1951).
A portion of purified proteins was cleaved at the C-terminus of internal methionine residues by cyanogen bromide in 70% formic acid at 37°C for 2 h. Intact or fragmented polypeptides pretreated with 4-vinylpyridine were electrophoresed and blotted onto polyvinylidene fluoride membrane (Millipore, Billerica, MA, USA). The membrane was washed thoroughly with 50% methanol and applied to a 476A ABI peptide sequencer (Applied Biosystems, Foster, CA, USA).
For cycle sequencing, Thermo Sequenase Dye Terminator Cycle Sequencing Premix Kit (Amersham Pharmacia Biotech, Piscataway, NJ, USA) was used. The products were analyzed by a 373A, ABI DNA sequencer (Applied Biosystems).
Digoxigenin (DIG)-labeled DNA probes were prepared, as described in the instruction manual (Dig DNA labeling kit, Boehringer Mannheim, Indianapolis, IN, USA). A λgt11 cDNA library was prepared from Polyandrocarpa colonies (Kawamura et al. 1998). A total of 100 000 plaques were hybridized with DIG-labeled probes and visualized by anti-DIG antibody system.
Northern blot hybridization
cDNA were ligated to pBluescript II SK+ (Stratagene, La Jolla, CA, USA). Transcription of RNA probes was done for 2 h at 37°C, using T3 or T7 RNA polymerase in the presence of DIG-RNA labeling mixture (Roche, Basel, Switzerland). Total RNA extracted from Polyandrocarpa colonies was separated on agarose-formaldehyde gels and blotted on Hybond-N+ membrane (Amersham). Hybridization was performed at 50°C in the buffer consisting of 5× standard saline citrate (SSC), 0.1% sodium lauroylsarcosinate, 0.02% sodium dodecylsulfate (SDS), 1% blocking reagent (Boehringer Mannheim) and 50% formamide. The membrane was washed twice in 2 × SSC containing 0.1% SDS at room temperature for 5 min, then twice in 0.1 × SSC containing 0.1% SDS at 65°C for 15 min. Hybridization signals were detected using the anti-DIG antibody conjugated with alkaline phosphatase and the chemiluminescent substrate, Lumigen PPD (Boehringer Mannheim).
In situ hybridization
RNA probes were synthesized as described above. Specimens were fixed in 4% paraformaldehyde in PBS for 12 h at 4°C. They were washed three times with chilled PBS containing 0.1% Tween-20 (PBST), dehydrated with serial methanol and reserved in 100% methanol at −20°C. They were incubated in cold xylene for 15 min, hydrated in serial methanol and finally washed three times with PBST. After proteinase K (10 µg/mL PBST) treatment for 20 min, specimens were washed with glycine (2 mg/mL PBST) and then PBST, and postfixed in the mixture of 4% paraformaldehyde and 0.2% glutalaldehyde. Prehybridization and hybridization were made for 1 h and 12 h at 65°C, respectively.
The blocking was done overnight at 4°C in 2% skim milk. Specimens were reacted with anti-DIG antibody containing 2% skim milk in PBST for 4 h at 4°C. They were colored and cleared, as described elsewhere (Ohashi et al. 1999). Some specimens were dehydrated and embedded in plastic resin. They were cut serially with glass knives and mounted on glass slips for microscopy (Matsumoto et al. 2001).
Recombinant P-Serpin 1 protein was prepared, as described elsewhere (Kawamura et al. 1998). Native P-Trefoil was prepared from frozen colonies. BALB/c 3T3 mice were immunized five times with these antigens conjugated with Freund's adjuvant. Spleen cells were fused with NS1/AG4 myeloma cells and screened by HAT medium. Positive hybridomas were cloned, and the culture media were stored in a refrigerator in the presence of 0.05% NaN3. Monoclonal antibodies (APE, APG) that recognize specifically the endodermal (atrial) epithelium and the gastric epithelium have been described elsewhere (Kawamura & Fujiwara 1994).
Sodium dodecyl sulfate–polyacrylamide gel electrophoresis and western blots
Electrophoresis was done on 15% polyacrylamide gel containing 0.1% SDS in 0.375 m Tris-HCl (pH 8.8) (Laemmli 1970). Proteins were blotted electrically onto nitrocellulose (BioRad, Hercules, CA, USA) at 300 mA for 1.5 h in Tris-glycine buffer (25 mm Tris, 195 mm glycine, 20% methanol).
Immunostaining was carried out as described elsewhere (Matsumoto et al. 2001). In brief, specimens were fixed in Zamboni's fixative at 4°C for 60 min, followed by acetone at −20°C for 10 min. They were blocked by 2% skim milk for 60 min. They were then incubated in the primary antibody for 4 h and the secondary antibody labeled with alkaline phosphatase or rhodamine for 1 h, respectively.
The cell line established from the multipotent epithelium of P. misakiensis was cultured in growth medium containing 3% fetal bovine serum in the basal medium that consists of Millipore-filtered seawater and Dulbecco's modified Eagle's medium (5:1; Kawamura & Fujiwara 1995). The bioassay was done as described elsewhere (Ohashi et al. 1999). In brief, cells were suspended in the serum-free basal medium at the density of 1 × 105 cells/mL. Aliquots of native P-serpin dissolved in PBS were added to the cell suspension. Each 100 µL was plated on 96-well multiplates. In controls, PBS was added to cell suspensions. Cells were counted everyday using a hemocytometer.
Identification of the molecular partner of P-Serpin
In order to identify the molecule(s) with which P-Serpin interacts, we prepared a fusion protein consisting of a maltose-binding protein (MBP) and mature P-Serpin. The 55 kDa fusion protein was mixed with crude extracts from Polyandrocarpa colonies (Fig. 1A lane 1). This mixture was applied to an amylose resin column, washed with PBS (Fig. 1A lane 2) and eluted with maltose (Fig. 1A lane 3). The eluate contained the fusion protein and two novel polypeptides of 40 kDa and 45 kDa. As a negative control, a MBP-phospholipase A2 fusion protein (Arai et al. 2004) was conjugated with the crude Polyandrocarpa extracts (Fig. 1B lane 1). Neither the 40 kDa polypeptide nor the 45 kDa polypeptide was pulled down together with the fusion protein (Fig. 1B lanes 2,3).
We attempted to isolate the two polypeptides on a large scale. After gel filtration chromatography of the crude Polyandrocarpa extracts, the 40 and 45 kDa polypeptides were eluted mainly in the void peak (Fig. 1C lane 1, arrows, and Fig. 2A). Weak bands were observed in the third peak (Fig. 1C lane 6 and Fig. 2A). The peak fractions were collected and applied to an anion-exchange column. Both the 40 and 45 kDa polypeptides were eluted with 0.3 m NaCl as a single peak (Fig. 1D lane 1 and Fig. 2B). The two bands were separated even under non-reducing conditions by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE; not shown). Finally, 3.5 mg protein was isolated from 12 g of the colonies (Table 1). Using aliquots of the 40/45 kDa antigens, a few monoclonal antibodies that recognized only the 45 kDa polypeptide (Fig. 1D lane 2) or both polypeptides (Fig. 1D lane 3) were prepared. These antibodies recognized the antigens in the eluate obtained after the MBP-P-Serpin affinity chromatography (Fig. 1A lane 4), thereby indicating that the 40 and 45 kDa polypeptides that were purified by anion-exchange chromatography were identical to the pull-down products.
Table 1. A summary of the purification of P-Trefoil from Polyandrocarpa colonies (12 g) with reference to the effect of P-Serpin on aminopeptidase activity of P-Trefoil
The P-Serpin was added to the P-Trefoil with a molar ratio of 4:1.
Crude extract (12 g)
Gel filtration chromatography
Anion exchange chromatography
102 (15 200)
Finally, we confirmed the molecular coupling of P-Serpin and the 40/45 kDa polypeptides by immunoprecipitation. Anti-P-Serpin monoclonal antibody was prepared against recombinant P-Serpin. It recognized the 16 kDa native P-Serpin (Fig. 1E lane 1). After gel filtration chromatography of the crude extracts, the void- and third-peak fractions were mixed with the anti-P-Serpin antibody and applied to a protein G column. Besides the immunoglobulin heavy and light chains, the eluate contained several other bands (Fig. 1E lane 2). These were the 40 and 45 kDa polypeptides and the 16 kDa native P-Serpin (Fig. 1E lane 3).
Molecular characterization of P-Trefoil
The N-terminal amino acid sequence of the 40 kDa polypeptide was DQSAECSTAAENRVDCGYVGI (Fig. 2C). Following cyanogen bromide cleavage, the 40 kDa polypeptide split into 22 and 18 kDa polypeptides (Fig. 2D). The former had the same N-terminal sequence as that of the native (40 kDa) polypeptide. The N-terminal sequence of the latter was GITEVQCAEK (Fig. 2D), indicating the internal amino acid sequence. Surprisingly, the 45 kDa polypeptide exhibited the same amino acid sequence as that of the 40 kDa polypeptide (not shown).
Degenerate primers were designed based on the amino acid sequences. A DNA fragment of approximately 300 bp was amplified by polymerase chain reaction (PCR). Using this as a DIG-labeled probe, three positive clones were isolated from a Polyandrocarpa cDNA library. The longest cDNA was 928 bp long. Northern blot hybridization also yielded a single band of approximately 1 kb (Fig. 2E). The deduced open reading frame (ORF) consisted of 266 amino acids (Fig. 3A). Residues 17–37 (DQSAECSTAAENRVDCGYVGI) were identical to the N-terminal amino acid sequence of the 40 kDa polypeptide, indicating that the 16 amino acids at the N-terminus of the ORF comprised the signal sequence (Fig. 3A). The internal amino acids (GITEVQCAEK) of the 40 kDa polypeptide were found between residues 130 and 139 of the ORF (Fig. 3A). Residue 129 was methionine (Met), and this is consistent with the notion that cyanogen bromide cleaves peptide bonds at the carboxyl side of Met.
The ORF had five similar repeats consisting of 21 amino acids, of which three repeats completely satisfied the trefoil consensus pattern (RxxCx[FYPST]xxx[ST] xxxCxxxxCC[FYWH]) (Fig. 3A,B). This main motif and surrounding amino acids included six conserved cysteine residues. (Fig. 3B, asterisks). In vertebrates, they are known as the trefoil domain that is characteristic of the trefoil growth factor family. Therefore, we termed the 928 bp cDNA as ‘Polyandrocarpa trefoil factor’ (P-trefoil) and the 40/45 kDa polypeptides as ‘P-Trefoil’ (GenBank accession number AB234894).
The ORF also had a domain similar to the thiol (cysteine) protease active site (Fig. 3B, box), which contains the consensus pattern Qxxx[GE]xC[YW] xx[STAGC][STAGCV]. In fact, native P-Trefoil showed the enzymatic activity of an aminopeptidase (Fig. 2B). On anion-exchange chromatography, its specific activity was found to be 14 900 U/h·per mg (Table 1). When P-Serpin was added to P-Trefoil at a molar ratio of 4:1, the aminopeptidase activity was not substantially affected (Table 1).
Spatiotemporal expression of the mRNA and protein of P-serpin
In P. misakiensis, budding occurs by evagination of the parent body wall (Fig. 4A). Whole-mount in situ hybridization demonstrated that the intense P-serpin mRNA signal first appeared in the bud primordium, and it was maintained throughout the period of bud growth (Fig. 4A,B). No signals could be detected in the sense control (Fig. 4C). In sections, the coelomic cells in growing buds were found to express the signal (Fig. 4E). They were mainly granular leukocytes characterized by the presence of large granules in the cytoplasm (Fig. 4E, inset; cf. Sugino et al. 1993). Interestingly, the signal could scarcely be seen in the parent body, although the coelomic space was connected to that of the bud primordium (Fig. 4A,B).
In the developing buds, the granular leukocytes continued to express P-serpin mRNA at least 2 days after they were extirpated from the parent (Fig. 4F). By this stage of development, the multipotent endodermal cells (from the atrial epithelium) began to proliferate and underwent morphogenesis (cf. Kawamura & Fujiwara 1994). They had weak but evident P-serpin signals (Fig. 4F). Within approximately 1 week, the buds developed into young functional animals. They did not show any staining (Fig. 4D).
The spatiotemporal expression of the proteins was examined by immunohistochemistry and western blotting using an anti-P-Serpin monoclonal antibody (cf. Fig. 1E lane 1). Whole-mount immunohistochemistry showed that the signal was visible in growing buds, while it could not be detected in the parent bearing the buds (Fig. 5A). The buds entered the morphogenesis stage after isolation from the parent. In 1 day developing buds, the signal was localized in the morphogenetic region (Fig. 5B). In 2 day developing buds, the extracellular matrix (ECM) in the coelomic space became prominent (Fig. 5C). In sections, the coelomic cells, the multipotent endodermal cells and the ECM were stained (Fig. 5D). Thereafter, the P-Serpin signal gradually weakened, and in approximately 1 week, it was almost lost from the young functional animals (Fig. 5E,F).
Homogenates for western blotting were prepared from young functional animals, fully grown animals and growing buds as well as from 1, 2 and 3 day developing buds. The same amount of the total protein from each homogenate was applied to the gel. After blotting, the nitrocellulose membrane was stained with the anti-P-Serpin monoclonal antibody. The young functional animals did not show any signals (Fig. 5G lane 1). The adult animals, from whom the growing buds were extirpated in advance, showed only a trace of the signal with a relative electrophoretic mobility of 16 kDa (Fig. 5G lane 2). On the other hand, the growing buds and the developing buds showed strong signals (Fig. 5G lanes 3,4,5). The strongest signal lasted for 2 days. From day 3 of bud development, the signal weakened considerably (Fig. 5G lane 6).
Expression pattern of the mRNA and protein of P-trefoil
The coelomic cells and the multipotent endodermal cells demonstrated weak P-trefoil mRNA signals during bud formation (Fig. 6A). In 1 day developing buds, the coelomic cells showed evident signals (Fig. 6B). The larger cells showed stronger signals. They were identified as granular leukocytes. In 2 day developing buds, the signal weakened to some extent and spread from the coelomic cells to the epidermal cells (Fig. 6C). Both young and adult animals did not show any detectable signals (not shown). In the negative controls, no tissues showed any staining on treatment with the sense probe (Fig. 6D).
The expression and distribution of the P-Trefoil protein was examined by immunohistochemistry and western blotting using the anti-P-Trefoil antibody. In the growing buds, a few coelomic cells were stained (Fig. 6E). The signals were widely spread in the coelomic cells of 1 day buds (Fig. 6F). In 2 and 3 day buds, the epidermal cells and the coelomic cells facing the epidermis emitted the strongest signal (Fig. 6G,H). No difference was observed in the staining pattern between the anti-40 kDa/45 kDa antibody and the anti-45 kDa antibody (not shown).
Homogenates were prepared from adult animals and buds in the same manner as those prepared for the detection of P-Serpin. The signals of the 40 kDa polypeptide were relatively strong throughout the life span except in adult animals (Fig. 6I lane 2). On the other hand, those of the 45 kDa polypeptide changed evidently. While the signal was the strongest in the 2 day buds (Fig. 6I lane 5), it became weak in the young functional animals (Fig. 6I lane 1) and was almost lost in the adult animals (Fig. 6I lane 2).
Effects of the P-Trefoil/P-Serpin complex on cell growth and differentiation
The biological activities of P-Trefoil and/or P-Serpin were examined using Polyandrocarpa cell lines (Kawamura & Fujiwara 1995). In serum-free basal medium, the cells grew very slowly, while addition of the P-Trefoil isolated by anion-exchange chromatography (cf. Table 1) promoted cell growth and cell spreading (Fig. 7A,B,E). The minimum amount of protein required for this was 8.8 µg/mL (130 U/mL). Since the samples isolated by gel filtration chromatography showed higher activity (Fig. 7A), the purified P-Trefoil was mixed with P-Serpin before addition to the cell suspension. The cell growth-promoting activity approximated that of the gel filtration chromatography sample (Fig. 7A). P-Serpin by itself did not show any apparent activity (Fig. 7A).
The formation of the gastrointestinal epithelium is one of the major pathways involving the differentiation of the multipotent endodermal epithelium. Both epithelia express alkaline phosphatases, but are serologically distinguishable (Fig. 7C, inset, 7D, left inset) (Kawamura & Fujiwara 1994). The monoclonal antibody APE is specific to the endodermal epithelium, and APG is specific to the gastric epithelium. We cultured Polyandrocarpa cell lines for 3 or 7 days in the presence or absence of P-Trefoil/P-Serpin. In the controls, the cells were not stained with APE or APG (Fig. 7C,D). The cells that were cultured for 3 days in the presence of the P-Trefoil/P-Serpin complex showed staining with APE. (Fig. 7F). The cells that were cultured with P-Trefoil/P-Serpin for 1 week became larger and showed staining with APG (Fig. 7G).
The 40/45 kDa polypeptides have trefoil domains and proteolytic activity
Serpins are relatively small molecules that inhibit the enzymatic activities of serine proteases. They can form complexes with thrombin (Cunningham & Gurwitz 1989), plasminogen activator (Martin et al. 1993), kallikrein (Chen et al. 2000) and trypsin (Ye et al. 2001). In addition to the association with proteases, antithrombin III is known to bind heparin in porcine aqueous humor (Rao et al. 2000). Caspin is a collagen-associated serpin produced by murine colon adenocarcinoma cells (Kozaki et al. 1998). We have shown here for the first time that a serpin can interact with a trefoil factor in tunicates. By affinity chromatographies using MBP-P-Serpin and anti-P-Serpin antiserum, 40/45 kDa polypeptides were identified. Surprisingly, they appeared to be two products from a single mRNA due to the following reasons: (i) The 40 kDa polypeptide had the same amino acid sequences as the 45 kDa polypeptide at the N-terminus and in the internal position; (ii) the anti-P-Trefoil monoclonal antibody recognized both the 40 and 45 kDa polypeptides; (iii) a single cDNA fragment was amplified by PCR using degenerate primers designed from the amino acid sequence data; (iv) northern blot hybridization yielded a single band of approximately 1 kb, which provides evidence against the alternative splicing of immature mRNA; and (v) the amino acid sequence (266 residues) deduced from a full-length cDNA (928 bp long) did not correspond with the relative electrophoretic mobility of the 40/45 kDa polypeptides. In fact, the molecular mass of the recombinant P-Trefoil protein synthesized by a bacterial expression system was found to be approximately 30 kDa by SDS–PAGE; this is evidently smaller than the native polypeptides (our unpubl. data, 1998). Thus, several lines of evidence have indicated that both the 40 and 45 kDa polypeptides are encoded by a single gene and that they may differ in size from each other due to a post-translational modification such as glycosylation.
The amino acid sequence deduced from the cDNA of the 40/45 kDa polypeptides had five tandem repeats (21 residues) of which the third, fourth and fifth satisfied the trefoil consensus motif. Each trefoil domain (45 residues) included these 21 residues and contained six conserved cysteine residues except the fifth trefoil domain (cf. Fig. 3B). In vertebrates, trefoil peptides have three disulfide bonds between the first and the fifth, the second and the fourth as well as the third and the sixth half-cystines, forming a three-lobed leaf (Thim 1989). Generally, they are small protease-resistant glycoproteins (Poulsom & Wright 1993). In mammals, they are principally secreted by the gastrointestinal tract (Kindon et al. 1995). In P. misakiensis, unlike in mammals, the trefoil signals were very weak in the stomach and the intestine (Y. Ono, unpubl. data, 1998). Instead, they were found mainly in the coelomic cells.
Interestingly, P-Trefoil showed strong aminopeptidase enzymatic activity. Aminopeptidase belongs to a thiol protease family wherein the active site is cysteine (Stennicke & Salvesen 1999). In fact, the fourth trefoil domain of P-Trefoil contained a unique sequence very similar to the thiol protease consensus pattern (cf. Fig. 3B), which may explain why P-Trefoil shows the proteolytic activity. P-Serpin inhibits the enzymatic activities of trypsin and elastase (Kawamura et al. 1998), but it did not have a dramatic effect on the aminopeptidase activity of P-Trefoil (cf. Table 1). Therefore, it is unlikely that P-Serpin binds and masks the active cysteine of thiol proteases. The mechanism of the molecular interaction between P-Serpin and P-Trefoil is currently unknown.
Overlapping expression of P-Serpin and P-Trefoil during budding
The second finding of this paper is that in P. misakiensis, both P-Serpin and P-Trefoil are expressed simultaneously during budding. Previous expressed sequence tag (EST) analysis showed that the amount of P-serpin in budding animals was approximately fivefold that in prebudding animals (Kawamura et al. 1998), but little precise information is available about the spatiotemporal expression of P-serpin. As described in this paper, P-serpin gene expression and protein accumulation begins with budding. Granular leukocytes, which are a type of free coelomic cells, were responsible for this extraordinarily high expression of P-serpin in the growing and developing buds. Fujimoto and Watanabe (1976) have reported that in a related species, Polyzoa vesiculiphora, the granular leukocytes increase in number during budding. They accumulate oil droplets and other substances in their vacuoles, and these serve as nutrient reservoirs during budding. This paper shows that even in P. misakiensis, granular leukocytes are involved in protein synthesis and accumulation during budding.
With regard to P-trefoil, the previous EST analysis did not hit any trefoil factors (Kawamura et al. 1998). However, it turned out later that P-trefoil has been described as integumentary mucin A or C. The amount of the 40 kDa polypeptide was relatively constant throughout the life span of the animals; this may be related to the fact that trefoil factors are generally stable protease-resistant glycoproteins (Poulsom & Wright 1993). In contrast, the signals of both the P-trefoil mRNA and the 45 kDa polypeptide were maximum in the early phase of bud development. The strongest signals were emitted by the large coelomic cells that were identified as the granular leukocytes. Therefore, the synthesis and secretion of P-Trefoil overlapped spatiotemporally with those of P-Serpin. Together with the biochemical data obtained using several types of chromatography, these findings have provided evidence supporting the in vivo and in vitro molecular coupling of P-Serpin and P-Trefoil.
We speculated whether P-Serpin binds to serine protease(s) since P-Serpin can interact with trypsin and elastase and inhibit their enzymatic activities (Kawamura et al. 1998). For example, tunicate retinoic acid-inducible modular protease (Tramp) is expressed specifically in the coelomic space during budding (Ohashi et al. 1999). However, within the scope of our study, there is no evidence indicating the association of P-Serpin with Tramp (K. Ohta, unpubl. data, 2002).
Collaborative effect of P-Trefoil and P-Serpin on multipotent endodermal cells
The third finding of this paper is that the P-Trefoil/P-Serpin complex promotes in vitro cell growth, cell motility and cell differentiation. In P. misakiensis, budding involves the dedifferentiation and cell division of the multipotent endodermal epithelium (Fig. 8; Kawamura & Nakauchi 1986; Kawamura & Fujiwara 1994). Tramp and phospholipids have thus far been reported as the endogenous dedifferentiation factors (Ohashi et al. 1999; Arai et al. 2004). The calcium-dependent galactose-binding C-type lectin TC14-3 antagonizes these dedifferentiation factors (Matsumoto et al. 2001). The present study has shown that in P. misakiensis, the P-Trefoil/P-Serpin complex is a novel member that is involved in the dedifferentiation and redifferentiation of multipotent cells during budding.
In vertebrates, serpins act as mitogens (Hochstrasser et al. 1989) and motogens (Marchbank et al. 1996). In P. misakiensis, however, P-Serpin did not have any such effects on the cultured cells. It is evident that P-Trefoil plays a key role in cell growth and differentiation, and its molecular collaboration with P-Serpin enhances these activities.
In vertebrates, trefoil factors act on the digestive tract, but different species of trefoil peptides appear to have different activities. The ability of intestinal trefoil factor (ITF)-deficient mice to withstand injury to the intestine was lower (Mashimo et al. 1996). On the other hand, pS2-null mice did not exhibit this deficiency in injury repair, but demonstrated multifocal carcinoma (Lefebvre et al. 1996). In P. misakiensis, epidermal injury induces P-trefoil gene expression (our unpubl. data, 2000). The P-Trefoil/P-Serpin complex caused the in vitro induction of the endodermal cell-specific alkaline phosphatase (AP-E) that is recognized by the monoclonal antibody APE. This antigen is usually absent in confluent cultures (Kawamura & Fujiwara 1995). Long-term (1 week) exposure to the P-Trefoil/P-Serpin complex resulted in the expression of another differentiation antigen specific to the gastric epithelium (AP-G). This change in the antigen expressed in the cultured cells is consistent with the in vivo differentiation pathway (Kawamura & Fujiwara 1994). The P-Trefoil/P-Serpin complex was ineffective in inducing a stomach-specific digestive enzyme, trypsin (K. Ohta, unpubl. data, 2002). This may suggest that it is insufficient for the terminal differentiation of the gastric epithelium.
In conclusion, we assume that the P-Trefoil/ P-Serpin complex serves as one of the instructive signals to induce the differentiation of the gastric epithelium.
We thank Dr Tom Suzuki, Laboratory of Biochemistry, Kochi University for sequencing the P-Trefoil protein. We also thank Dr Toshi Yubisui for valuable advice and encouragement throughout the course of the study. We are also grateful to the staff of the USA-Marine Biological Laboratory for culturing the animals.