Tunicate cytostatic factor TC14-3 induces a polycomb group gene and histone modification through Ca2+ binding and protein dimerization

Background As many invertebrate species have multipotent cells that undergo cell growth and differentiation during regeneration and budding, many unique and interesting homeostatic factors are expected to exist in those animals. However, our understanding of such factors and global mechanisms remains very poor. Single zooids of the tunicate, Polyandrocarpa misakiensis, can give off as many as 40 buds during the life span. Bud development proceeds by means of transdifferentiation of very limited number of cells and tissues. TC14-3 is one of several different but closely related polypeptides isolated from P. misakiensis. It acts as a cytostatic factor that regulates proliferation, adhesion, and differentiation of multipotent cells, although the molecular mechanism remains uncertain. The Polycomb group (PcG) genes are involved in epigenetic control of genomic activity in mammals. In invertebrates except Drosophila, PcG and histone methylation have not been studied so extensively, and genome-wide gene regulation is poorly understood. Results When Phe65 of TC14-3 was mutated to an acidic amino acid, the resultant mutant protein failed to dimerize. The replacement of Thr69 with Arg69 made dimers unstable. When Glu106 was changed to Gly106, the resultant mutant protein completely lost Ca2+ binding. All these mutant proteins lacked cytostatic activity, indicating the requirement of protein dimerization and calcium for the activity. Polyandrocarpa Eed, a component of PcG, is highly expressed during budding, like TC14-3. When wild-type and mutant TC14-3s were applied in vivo and in vitro to Polyandrocarpa cells, only wild-type TC14-3 could induce Eed without affecting histone methyltransferase gene expression. Eed-expressing cells underwent trimethylation of histone H3 lysine27. PmEed knockdown by RNA interference rescued cultured cells from the growth-inhibitory effects of TC14-3. Conclusion These results show that in P. misakiensis, the cytostatic activity of TC14-3 is mediated by PmEed and resultant histone modification, and that the gene expression requires both the protein dimerization and Ca2+-binding of TC14-3. This system consisting of a humoral factor, PcG, and histone methylation would contribute to the homeostatic regulation of cell growth and terminal differentiation of invertebrate multipotent cells.


Background
Cell and tissue homeostasis are among the most important features of living organisms. In vertebrates, various types of extracellular molecules act as cell growth regulators. For example, angiostatin and endostatin are potent inhibitors of endothelial cell proliferation and angiogenesis [1,2]. They contribute to our understanding of in vivo cell growth homeostasis and therapeutic control of tumor angiogenesis [3]. Among invertebrates, many species have multipotent cells that undergo cell growth and differentiation during regeneration and budding [4,5]. Therefore, many unique and interesting homeostatic factors are expected to exist in invertebrates. However, our understanding of such factors and global mechanisms remains very poor.
The Polycomb group (PcG) genes are involved in epigenetic control of genomic activity. PcGs in Drosophila were initially identified as homeotic gene repressors [11,12]. PcG proteins bind in vivo to many discrete sites on the chromosome [13]. In mammals, PcG homologs play a role in genome-wide gene silencing [14]. They are essential for cell fate maintenance in embryonic stem cells [15] and hematopoietic stem cells [16]. In keratinocytes, PcG proteins regulate cell growth, differentiation, and senescence [17]. Polycomb repressive complex 2 (PRC2), a biochemically discernible component of PcG, is involved in gene repression by histone modification [18]. PRC2 contains several core proteins: Histone H3 methyltransferase (Ezh2) catalyzes trimethylation of H3 at Lys27 (H3K27me3); Eed and Suz12 are Ezh2 activators [16]. We found recently that a Polyandrocarpa homolog of Eed (PmEed) was remarkably induced during budding, an expression pattern similar to that of the TC14s [6,10]. It seems, therefore, likely that PmEed is involved in the cytostatic activity of TC14-3.
In this study, we aimed to disclose why and how only TC14-3 exerts the unique cytostatic activity in P. misakiensis. First, we examined amino acid moieties responsible for the cell growth-inhibitory activity of TC14-3. Using chimeric and mutant proteins, we demonstrate that protein dimerization and Ca 2+ binding motifs are essential for the cytostatic activity of TC14-3. Second, downstream genes of TC14-3 were looked for, using wild-type and mutant proteins. We present evidence that PmEed is up-regulated in vivo and in vitro by wildtype TC14-3. In relation to Eed induction, we show immunocytochemically histone H3 trimethylation in Polyandrocarpa cell nuclei. Using RNA interference (RNAi), rescue experiments were done to demonstrate that PmEed mediates the cell growth-inhibitory activity of TC14-3. Taken together, budding tunicates provide us with a unique and interesting system in which a coelomic polypeptide can induce a PcG gene and epigenetic histone modification.

Results
Survey of functional domains for cytostatic activity of TC14-3 Figure 1F shows the alignment of TC14-1, TC14-2, and TC14-3 sequences. All 3 proteins are composed of 145 amino acids, of which 20 N-terminal amino acids are signal peptides. The remaining 125 amino acids constitute the mature protein. The CRD of TC14s consists of 2 α helices, 5 β strands, and 4 loops ( Figure 1F) [19].
TC14-3 T69R exhibited no cytostatic activity on cultured tunicate cells ( Figure 3B,D). TC14-2 R69T , on the other hand, acquired the cytostatic activity to some extent ( Figure 3B,D). As a reference, the amino acid at position 70 was exchanged between TC14-2 and TC14-3. The cytostatic activity of the mutant proteins was unaffected (Table 1).
These results indicate that the amino acid at position 69 can modulate multiple characteristics of TC14s, such as electrophoretic mobility, stability of protein dimers, and cytostatic activity.
Both TC14-3 K113S and TC14-3 N114E retained their growth-inhibitory activities on cultured cells ( Figure  3C). On the other hand, the inhibitory activity was greatly diminished in the double mutant protein TC14-3 K113S.N114E ( Figure 3C,D). Mutations at C-terminal positions 136, 144, and 145 did not have any apparent influence on cell growth (Table 1).
In intact animals, PmEed was expressed abundantly from bud stages to juvenile zooid stages [see Additional file 1A, B, C, D], but diminished conspicuously at adult zooid stages except the gonad [see Additional file 1A, E, F] (More detailed results will be published elsewhere). In this study, adult zooids were cut into 3 pieces to facilitate TC14-3 infiltration, and treated with TC14-3 proteins for 2 days. Zooids of P. misakiensis possess a high potential for regeneration [20]. As expected, control zooid pieces treated with PBS could survive during the course of study. They did not exhibit any apparent signals for PmEed in most tissues and organs except the gonad ( Figure 7A-C), similar to intact adult zooids, indicating that the surgery by itself did not affect PmEed expression. In contrast to the control, zooid pieces that had been treated with wild-type TC14-3 ubiquitously expressed PmEed ( Figure 7D, G), the expression pattern similar to buds. The strongest signal was detected in coelomic cells in the hemocoel ( Figure 7F, H, I). The atrial, gastric, and perivisceral epithelia also expressed PmEed ( Figure 7F, I). The epidermis showed moderate expression of PmEed, but muscle cells did not ( Figure  7E). Results of RT-PCR showed that only wild-type TC14-3 could induce in vivo PmEed ( Figure 6C). By semi-quantitative PCR, the PmEed products became visible at the 25 th cycle ( Figure 6D), and increased exponentially thereafter ( Figure 6D, E). In the control, on the other  hand, PmEed products became first visible at the 27 th cycle ( Figure 6D), and increased parallel to the experiment ( Figure 6E). The result indicated that the amount of PmEed transcripts in wild-type TC14-3-treated animals was approximately 2-4-fold that of the control.

TC14-3 also induces mitochondrial respiratory gene
Our recent study showed that in P. misakiensis, PmEed and mitochondrial respiratory genes were both inactivated during zooidal senescence and reactivated remarkably during budding (Kawamura et al., submitted). We examined, therefore, whether wild-type TC14-3 could induce not only PmEed but also cytochrome c oxidase 1 (PmCOX1) in aged zooids. Results of in situ hybridization showed that in the control, signals were hardly detectable in the body wall, pharynx, and visceral organs ( Figure 8A-C). In contrast, when TC14-3 was applied to zooids, a portion of epithelial cells and coelomic cells in the pharynx expressed PmCOX1 strongly ( Figure 8D, E). The endostyle, digestive tract, and surrounding coelomic cells did not emit signals ( Figure 8F). The increasing curves of PCR products indicated that TC14-3-treated samples had larger amount of PmCOX1 transcripts than untreated controls, although the difference was not so high ( Figure 8G).

Trimethylation of histone H3 by TC14-3
Anti-H3K27me3 antibody stained the in vivo nuclei of epithelial cells and coelomic cells in buds ( Figure 9A, B). Nuclei of epidermal cells stained weakly ( Figure 9A), whereas those of the atrial epithelium, multipotent epithelial cells in P. misakiensis, stained heavily ( Figure  9A, C, D). In the hemocoel, many coelomic cells emitted strong signals ( Figure 9A, C), but differentiated cells such as morula cells did not have apparent signals in the nucleus ( Figure 9C black arrowheads, 9D white arrowheads). Cultured cells untreated with TC14-3 were not stained with anti-H3K27me3 antibody ( Figure 9E). Cells treated with mutant protein (TC14-3 E106G ) were stained weakly ( Figure 9F), whereas wild-type TC14-3-treated cells were stained heavily with the antibody ( Figure 9G). Western blotting of in vitro cultured cells showed that anti-histone H3 antibody stained a single band of approximately 17 kDa ( Figure 9H, lane 1). Anti-H3K27me3 antibody, on the other hand, did not stain any bands when cells were not treated or treated with TC14-3 E106G ( Figure 9H, lanes 2, 3), but stained a single band of 17 kDa when cultured cells were treated with wild-type TC14-3 ( Figure 9H, lane 4). We could not find in vivo differences in histone trimethylation between TC14-3-treated and untreated samples (not shown).
The gene expression of PmEzh2, a Polyandrocarpa homolog of Histone H3K27 methyltransferase, was examined. Adult zooid fragments treated with wild-type TC14-3 showed the same strength of signals as those of untreated zooids [see Additional file 2 lanes 1, 2). Cultured cells in the growth medium without TC14-3 showed a weak signal of PmEzh2 PCR products at 30 th cycle [see Additional file 2 lane 3]. When cells were treated in vitro with wild-type or mutant TC14-3s, the signals were approximately the same as those of the control [see Additional file 2 lanes 4-7). These results indicate that wild-type TC14-3 can induce H3K27me3 without affecting PmEzh2 gene expression.

Recovery from TC14-3-induced growth arrest by PmEed knockdown
We examined the effect of PmEed RNAi on cell growth arrest by wild-type TC14-3. Double-stranded RNA of PmEed (dsRNA PmEed ) was introduced into cultured cells by electroporation. In the positive control, blunt electroporation was performed in the absence of dsRNA PmEed , and the cells were allowed to grow for 3 days without TC14-3. Cells spread on the culture dish ( Figure 10A). In the negative control, cells were treated with TC14-3 after the blunt electroporation. Cells formed many aggregates ( Figure 10B). In dsRNA PmEed experiments, cells spread again in the presence of TC14-3 ( Figure  10C). The cell number was approximately twice as many as that of the negative control ( Figure 10D). The recovery value accounted for 65% compared to the positive control.
Discussion a2 helix and loop 3 are essential for the cytostatic activity of TC14-3 The results of the chimera experiments revealed that the amino acids at positions 61-145 in the C-terminal region of TC14-3 are responsible for cytostatic activity. The C-terminal region contains 1 α helix (α2), 4 β strands (β2-β5), and 4 loops (L1-L4) (see Figure 1F). In the α2 helix of TC14-1, hydrophobic amino acids (Ala 61 and Phe 65 ) play a key role in protein dimerization [19]. Our study, using site-directed mutagenesis and SDS-PAGE of recombinant proteins, confirmed that in TC14-3, Phe 65 of α2 helix is essential for protein dimerization and also critical for cytostatic activity.
TC14s are Ca 2+ -binding proteins [7]. The ligands for calcium are the side-chain oxygen atoms of Glu 106 (loop 3), Asn 109 (loop 4), Asp 127 (β4 strand), and Asp 128 (β4 strand), as well as the main-chain carbonyl oxygen of Asp 128 (see Figure 1F) [19]. In TC14-3, Glu 106 of loop 3 played a key role in Ca 2+ binding, and the loss of Ca 2+ binding was associated with the loss of cytostatic activity. Glu 106 and Asn 109 of TC14s correspond to Glu 185 and Asn 187 of mannose-binding protein A (MBP-A), respectively. In MBP-A, double mutations, Glu 185 Gln and Asn 187 Asp, alter the sugar substrate specificity from mannose to galactose [21]. In E-selectin, the sequence Trp-Ala-Pro-Gly-Glu-Pro (76-81) regulates carbohydrate-binding specificity [22]. If Ala at position 77 is replaced with Ser, the sugar specificity of the mutant E-selectin changes from sialic acid to mannose. An exactly identical sequence exists in loop 3 of TC14-3 (see Figure 1F, positions 102-107). The corresponding sequence of TC14-2 was Trp-Ser-Pro-Asp-Glu-Pro. Both TC14-3 A103S and TC14-3 G105D retained strong cytostatic activity (see Table 1). It is, therefore, unlikely that the loop 3 is responsible for the difference between TC14-2 and TC14-3, although the loop 3 is essential for determining biological and biochemical features of TC14s.
Angiostatin and endostatin are specific, potent inhibitors of endothelial proliferation and angiogenesis [1,2]. Endostatin is a 20-kDa C-terminal fragment of collagen XVIII. TC14-3 is similar to endostatin in several respects. The X-ray structure of murine endostatin is similar to that of C-type lectin [23]. It lacks a characteristic Ca 2+ -binding site, but instead binds zinc at the Nterminus. This metal binding enables the dimerization of human endostatin [24]. Similar to TC14-3, protein dimerization is essential for endostatin to carry out the antitumor activity [3].
The amino acids at position 69 of TC14-3 and TC14-2 are Thr and Arg, respectively. As Arg has a large side chain, it would interfere with the fitting and hydrophobic bonds at the α2 helix between juxtaposing proteins. As expected, TC14-3 T69R changed the electrophoretic mobility and the stability of protein dimers, and lost the cytostatic activity. In contrast, TC14-2 R69T could not form stable dimers comparable to that of wild-type TC14-3. This result suggests that additional as yet unidentified amino acids may contribute to the stability of protein dimers. However, it is undoubted that the amino acid at position 69 can modulate the biological and biochemical properties of TC14s.

Lys 113 and Asn 114 modulate Ca 2+ binding of TC14-3
The cytostatic activities of TC14-3 depend on calciumdependent galactose binding [6]. Therefore, we initially expected that the affinity of TC14-3 for calcium may be higher than that of TC14-2. However, contrary to our expectation, the Ca 2+ -binding affinity of TC14-3 was apparently lower than that of TC14-2.
Lys 113 and Asn 114 are specific for TC14-3. They are located at the boundary between loop 4 and the β3 strand. When both these amino acids were replaced with those of TC14-2, the resultant TC14-3 K113S.N114E exhibited an increase in Ca 2+ -binding affinity (> 0.6) and a decrease in cytostatic activity. As mentioned, TC14-3 N109G had low Ca 2+ -binding affinity (0.4), and exhibited reduced cytostatic activity. Taken together, TC14-3 appears to have the highest cytostatic activity when the binding ratio of protein to Ca 2+ is 1:0.5.

PmEed mediates cytostatic activity of TC14-3
In P. misakiensis, the atrial epithelium is a transdifferentiation-competent, multipotent tissue [5,25]. It undergoes the terminal differentiation into the pharynx, gut, and brain when growing buds enter the developmental stage [25]. TC14-3 is induced remarkably during budding, and it disappears from the morphogenesis domain where transdifferentiation takes place [6]. This disappearance of TC14-3 may be caused by retinoic acidinducible serine protease [26]. TC14-3 can block in vitro cell growth and differentiation in Polyandrocarpa cell lines that have been established from explants of the atrial epithelium [6,27]. Consequently, Matsumoto et al. [6] have argued that in P. misakiensis, TC14-3 serves as a negative regulator of terminal differentiation of multipotent cells.
In P. misakiensis, PmEed was developmentally regulated during budding cycle. The gene expression of PmEed was the highest at bud stages, gradually diminish during zooid growth, and was almost absent in somatic tissues and organs of adult zooids (Kawamura et al., submitted). This expression pattern was similar to that of TC14. In the present study, wild-type TC14-3 could induce PmEed in both cultured cells and adult zooid tissues, and interestingly, mutant proteins with abnormalities in protein dimerization or Ca 2+ binding failed to induce PmEed.
Semi-quantitative PCR analysis of zooid pieces revealed that in the presence of TC14-3, the amount of PmEed transcripts was 2-4-fold higher than that of the control. This value seemed smaller than that expected from the results of in situ hybridization. This may be due to strong signals from the gonads in the control as well as the experiment. In fact, many gonads are embedded in the ventral body wall (see Figure 1A), and they particularly expressed PmEed in adult tissues in a TC14-3-independent manner. Therefore, the net induction of PmEed may be much larger, if the background value in the gonad could be subtracted from the total signal.
In P. misakiensis, dsRNA PmEed rescued cultured cells from the growth-inhibitory effect of wild-type TC14-3. This result affords further evidence that PmEed is a downstream mediator of cytostatic TC14-3. In mammals, when Eed is deficient in ES cells, PcG target genes are de-repressed [14], leading to cell growth and differentiation. Therefore, PcG is thought to play roles in stem cell renewal and inhibition of cell differentiation in ES cells [15]. Our results are consistent with these findings and notion in mammals.

Other genes regulated by TC14-3
A previous study has shown that in P. misakiensis, TC14-3 up-regulates α-integrin gene expression [6]. In this study, wild-type TC14-3 suppressed the gene expression of both cyclin A and cyclin B. In Drosophila, PcG directly down-regulates cyclin A [28].
In P. misakiensis, mitochondrial respiratory complex genes are regulated in accordance with PmEed during budding life cycle (Kawamura et al., submitted). When wild-type TC14-3 was applied to zooid pieces of P. misakiensis, PmCOX1 gene was up-regulated. This gene regulation may also be related to PmEed. However, it should be noted that, unlike PmEed, the expression of PmCOX1 was not ubiquitous, but restricted around the pharynx. It is, therefore, possible that mitochondrial respiratory complex genes may be up-regulated via a route other than PmEed.

Epigenetic histone H3 trimethylation involved in cell growth and differentiation
Eed and Ezh2 are the components of PRC2 in PcG [18]. Eed acts as Ezh2 activator, and Ezh2 catalyzes H3K27me3 in the so-called histone tail [16]. Trimethylation of histone H3K27 recruits PRC1 to the chromatin. PRC1 possesses a discrete enzyme activity that modifies histone H2A, resulting in genome-wide, epigenetic gene repression [14]. Polyandrocarpa histone H3 showed 100% sequence similarity to mammalian histone H3.3 (not shown). Rabbit anti-histone H3K27me3 antibody indeed stained nuclei of the atrial epithelium and coelomic cells in intact buds of P. misakiensis. Our in vitro studies indicated that wild-type TC14-3 could induce H3K27me3 in Polyandrocarpa cultured cells. It is notable that TC14-3 up-regulated the PmEed gene expression, but not PmEzh2. Therefore, epigenetic trimethylation of histone H3K27 should be ascribable exclusively to enhanced PmEed gene expression.
In contrast with the atrial epithelium and coelomic cells, nuclei of epidermal cells and coelomic morula cells were stained very weakly with anti-H3K27me3 antibody. The epidermis is a specialized tissue to synthesize and secrete tunic components. Morula cells are differentiated cells engaged in self-defense mechanisms. In the light of multipotency of the atrial epithelium [5,25], it is probable that H3K27me3 is related to the block of terminal differentiation in budding tunicates. In ES cells, STAT3, Oct-3/4, and Sox2 induce Eed that influences H3K27me3 in the nucleus [29,30]. These transcription factors are essential for stem cell maintenance. Although the atrial epithelium in tunicates is quite different from ES cells in origin and developmental potential, the basic mechanism for keeping the multipotent cell state appears to be shared by tunicate cells and mammalian ES cells.
Trithorax group also modifies histone H3 by trimethylation of Lys4. However, the result of histone methylation is quite different from the case of PcG, making chromatin loose and activating differentiation genes [16]. In P. misakiensis, Lys4 trimethylation occurs in the process of transdifferentiation, which will be reported in the near future.

Conclusions
As mentioned, TC14-3 is similar to endostatin in several aspects, but there are, of course, important differences between them. Endostatin binds α5β1 integrin and Eselectin on the endothelium [31] and inhibits the activity of metalloproteinases [32]. TC14-3, on the other hand, exerts cell growth inhibition at least in part by inducing in vivo and in vitro PmEed. A major function of induced PmEed is to facilitate H3K27me3. This system of budding tunicates consisting of a humoral factor, PcG, and histone trimethylation can regulate cell growth and differentiation of multipotent cells. Consequently, the homeostatic maintenance of transdifferentiation-competent cells would support budding and regenerative activities in P. misakiensis. Further studies of how humoral growth inhibitors such as endostatin and TC14-3 work in dimerization-and cation-dependent manners will afford insight into therapeutic control of malignant and/ or multipotent cells and tissues.

Animals
Asexual individuals of P. misakiensis were reared in culture boxes placed in the Uranouchi Inlet near the Usa Marine Biological Institute, Kochi University.

Cell culture and bioassay
Polyandrocarpa cells were cultured as described previously [27]. Cells were harvested in cell dissociation medium (0.2% trypsin and 2 mM EDTA in DMEM). They were resuspended in the growth medium at a density of 1 × 10 5 cells/ml, and 100 μl of this solution was plated in each well of a 96-well multiplate. Recombinant TC14s were added to the cell suspension at a final concentration of 30 μg/ml. As a control, sterile PBS (10 μl) was added to each well. Cells were counted with a hemocytometer [6] or the 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT) method [33]. For in vivo bioassay, adult animals were cut transversely into 3 pieces and incubated for 2 days in sterile seawater in the presence or absence of 30 μg/ml of wild-type TC14-3. for 1 min, 55°C for 1 min, and 72°C for 4 min; and 1 cycle at 72°C for 4 min. PCR products were treated with T4 polymerase for 5 min to produce blunt ends. After the phosphorylation of the 5' end by polynucleotide kinase (Takara Bio Inc.), linear DNAs were made circular by DNA ligase (Takara Bio Inc.). Mutation was confirmed by DNA sequencing.

Chimeric TC14s
In both TC14-2 and TC14-3, a unique HindIII restriction site was created at amino acid positions 60-62 by site-directed mutagenesis [see Additional file 3]. After the digestion with restriction enzymes, 3' fragments of TC14-2 and TC14-3 were exchanged with each other, and were ligated to 5' fragments. The chimeric cDNAs were mutated again to restore the original KAI sequence [see Additional file 3].

DNA sequencing
For cycle sequencing, the Thermo Sequenase Dye Terminator cycle sequencing premix kit (Amersham Pharmacia Biotech., Piscataway, NJ, USA) was used. The products were analyzed using a DNA sequencer (373A; ABI, Foster City, CA, USA).

Electrophoresis
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed with the method of Laemmli [34]. Proteins were treated with SDS sample buffer with or without heat denaturation. After electrophoresis, the gels were stained with Coomassie Brilliant Blue G250.

Gel scanning
After acrylamide gel staining, the proteins were scanned with Kodak EDAS 290 (Eastman Kodak Ltd., Rochester, NY, USA). The staining intensity of each band was quantified using Image Analysis software (ver. 3.5) (Eastman Kodak Ltd.). PCR products were separated by agarose gel electrophoresis and stained with ethidium bromide. They were scanned and quantified using Ima-geJ free software developed by the National Institutes of Health.

Ca 2+ binding experiments
Protein-Ca 2+ binding was measured by the flow dialysis method, using 45 CaCl 2 (Amersham Pharmacia Biotech, CA, USA) in 0.1 M NaCl, 20 mM MOPS (pH 7.0) at 25°C. The protein concentration was adjusted to 25-75 μM. The loss of radioactive ligands during experiments and the nonspecific Ca 2+ binding to the apparatus were corrected. The resulting Ca 2+ binding data were analyzed by the Adair-Klots equation for a single binding site.

Semiquantitative PCR
Poly(A) + RNA was extracted and purified from cultured cells and adult zooids by the biotinyl magnet method, according to the manufacturer's protocol (Roche, Mannheim, Germany). Single-stranded DNA complementary to poly(A) + RNA was synthesized for 1 h at 42°C using StrataScript reverse transcriptase (Agilent Technologies, Santa Clara, CA, USA). The DNA pool was stored as templates for PCR. PCR was performed in 2 steps: 1 cycle for sense strand synthesis (30 s at 94°C, 2 min at 52°C, and 2 min at 72°C); 24-35 cycles of denaturation for 30 s at 94°C, annealing for 60 s at 52°C, and extension for 90 s at 72°C. As an internal standard, β-actin cDNA was amplified by PCR.

In situ hybridization
The protocol for in situ hybridization has been described previously [35]. In brief, specimens were fixed in 4% paraformaldehyde in PBS at 4°C for 10-16 h. The fixed specimens were rinsed in PBS containing 0.1% Tween 20 (PBST), digested with proteinase K, and postfixed in 4% paraformaldehyde and 1% glutaraldehyde in PBST. Specimens were hybridized with digoxigeninlabeled antisense RNA probe for 12-14 h at 58°C. After thorough washing, samples were incubated in blocking solution (1% skim milk in Tris-buffered salt solution containing 0.1% Tween 20) for 6 h in an ice bath, and then treated overnight on ice with anti-digoxigenin monoclonal antibody labeled with alkaline phosphatase (Roche, Mannheim, Germany). The samples were stained with the color development solution, dehydrated, and embedded in Technovit 8100 resin (Heraeus Kulzer, Wehrheim, Germany).

Double-stranded RNA
PmEed cDNA (approximately 1 kb) devoid of poly(A) tail was inserted into pGEM-T (Promega Co.). Using the T7 RNA polymerase transcription system, sense and antisense RNA strands were synthesized. Both RNA solutions were mixed and heat-denatured for 10 min at 95°C. Then, the temperature was gradually lowered to anneal the double-stranded RNA (dsRNA). Immediately before use, the dsRNA was dissolved in RNase-free seawater at the final concentration of 0.5 μg/ml.

Additional material
Additional file 3: Experimental procedure for chimeric protein production. Both cDNA of TC14-2 and TC14-3 were mutated at the position Ile 61 to make an unique site for Hind III (top). They were cut with Hind III to exchange the C-terminal fragments with each other (middle). After ligation, chimeric cDNAs were mutated again to change Phe61 to Ile61 before transferred to expression vectors (bottom).