The budding ascidian Polyandrocarpa misakiensis proliferates asexually by budding. The atrial epithelium is a multipotent but differentiated tissue, which transdifferentiates into various tissues and organs after the bud separates from the parental body. We isolated cDNA clones homologous to the myc proto-oncogene from P. misakiensis. The cDNA, named Pm-myc, encoded a polypeptide of 639 amino acid residues, containing Myc-specific functional motifs, Myc box I and Myc box II, and the basic helix-loop-helix domain. Expression of Pm-myc was observed in the atrial epithelium in the organ-forming region of the developing bud, where the epithelial cells dedifferentiate and re-enter the cell cycle. The expression was also observed in fibroblast-like cells, which are known to participate in the organogenesis together with the epithelial cells. Unexpectedly, the atrial epithelium expressed Pm-myc more than one day before the dedifferentiation. The organogenesis was disturbed by Pm-myc-specific double-stranded RNA. In situ hybridization revealed that Pm-myc-positive fibroblast-like cells disappeared around the organ primordium of the dsRNA-treated bud. The results suggest that the mesenchymal-epithelial transition of fibroblast-like cells is important for the organogenesis in this budding ascidian species.
The Myc protein is linked to the dedifferentiation and proliferation of animal cells (e.g. Freytag 1988; Géraudie et al. 1990; Iovanna et al. 1992; Agata et al. 1993). Since deregulated Myc stimulates terminally differentiated cells to re-enter the cell cycle, functions of Myc have been extensively studied in the context of tumor formation (Nesbit et al. 1999). Recently, Myc became spotlighted because pluripotent stem cells were artificially induced by forced expression of a few transcription factors including Myc (Takahashi & Yamanaka 2006). Myc’s physiological functions are necessary in highly proliferative embryonic cells and adult tissue stem cells. Targeted disruption of the c-myc gene in mice resulted in embryonic lethality (Davis et al. 1993). Conditional knockout experiments revealed that c-myc regulated the balance between self-renewal and differentiation of tissue stem cells (Waikel et al. 2001; Wilson et al. 2004). Myc also seems involved in the dedifferentiation during regeneration in vertebrates. During the transdifferentiation of retinal pigment epithelial cells to lens cells, expression of c-myc is activated in dedifferentiated proliferative cells (Agata et al. 1993). Similarly, c-myc expression is upregulated prior to the transdifferentiation of hepatic stellate cells to myofibroblast-like cells (Potter et al. 1999).
The protochordate ascidian Polyandrocarpa misakiensis proliferates asexually by budding (Kawamura & Watanabe 1982). During bud development in this species, the entire digestive tract is formed mainly by transdifferentiation of the atrial epithelium (cf. Fig. 2a–c; Fujiwara & Kawamura 1992; Kawamura & Fujiwara 1994). The epidermis in the future posterior region of the bud produces retinoic acid (Kawamura et al. 1993). Retinoic acid induces mesenchyme cells to secrete transdifferentiation factor(s) that, in turn, directly induce dedifferentiation of the atrial epithelium (Hara et al. 1992). Our previous studies identified a number of secreted factors that regulate the differentiated state and mitotic activity of the atrial epithelium (reviewed by Kawamura et al. 2008a). However, information is not sufficient regarding the expression and function of genes in the atrial epithelium during dedifferentiation and proliferation prior to redifferentiation into the gut. In addition to the transdifferentiation of the atrial epithelium, mesenchyme cells take part in the formation of the gut (reviewed by Kawamura et al. 2008a). Fibroblast-like mesenchyme cells form aggregates near the gut primordium, and are integrated into the atrial epithelium (Kawamura et al. 1991).
In the present study, we isolated cDNA clones homologous to myc from P. misakiensis. The cDNA, named Pm-myc, is expressed during budding. The expression was detected in the atrial epithelium more than 1 day before dedifferentiation. The fibroblast-like mesenchyme cells also expressed Pm-myc. Treatment of the developing bud with a Pm-myc-specific double-stranded RNA (dsRNA) resulted in impaired formation of the gut primordium. In the dsRNA-treated buds, the Pm-myc-positive mesenchyme cells were not observed, suggesting that the hypoplasia of the gut results from a failure of mesenchyme cells to integrate into the dedifferentiated atrial epithelium.
Materials and methods
Polyandrocarpa misakiensis was attached on glass slides and cultivated in acrylic cages, settled in Tosa Bay close to Usa Marine Biological Institute of Kochi University.
Isolation of cDNA clones
A cDNA fragment corresponding to the N-terminal domain was amplified by polymerase chain reaction (PCR) using degenerate primers MbI-DEG-F and MbII-DEG-R (Table S1). The 5′ region of the cDNA was isolated by the “rapid amplification of cDNA ends” method (Frohman et al. 1988), with modifications described by Barns (1994) and Cheng et al. (1994). Myc-R4 was used as a specific primer for PCR (Table S1). A 506 bp cDNA fragment was obtained. This fragment was used as a template for synthesizing dsRNA as described below. For isolation of the 3′ region of the cDNA, PCR was carried out using a DNA sample purified from the λgt11 cDNA library constructed from a P. misakiensis colony (Shimada et al. 1995). First, nested PCR was carried out, using specific primers (Myc-F5 and Myc-F6 in this order) and another degenerate primer, HLH-DEG-R (Table S1). An approximately 1 kb cDNA fragment was obtained. This fragment was used for synthesizing digoxigenin-labeled probes and dsRNA described below. Then the 3′ end of the cDNA was obtained by another series of nested PCRs using Pm-myc-specific primers (Myc-F7, Myc-F8 and Myc-F9 in this order) and λgt11-specific primers (gt11-R1–R3 in this order) (Table S1). Finally, the entire translated region of the cDNA was amplified from the library, using primers Myc-5′-F and Myc-3′-R that corresponded to the 5′ and 3′ untranslated regions, respectively (Table S1). In every step, the product of the PCR was purified from an agarose gel, and inserted into pGEM-T (Promega).
Extraction of RNA was carried out as described by Fujiwara et al. (1993). For semi-quantitative reverse transcription (RT)–PCR, RNA was purified by using an RNeasy kit (Qiagen). Reverse transcription was carried out as described by Kaneko et al. (2010), except that 20 μL of the reaction solution contained 5 μg of total RNA as a template. The cDNA was diluted with 80 μL of TE (10 mM Tris-HCl, 1 mM ethylenediaminetetraacetic acid [EDTA], pH 7.5). PCR was carried out in 12.5 μL of reaction solution containing 1.25 μL of the cDNA pool. The primers MycRT-F1 and MycRT-R1 were used for amplification of the Pm-myc cDNA fragment from normally developing buds (Table S1). MycRT-F2 and MycRT-R2 were used for amplification from dsRNA-treated buds (Table S1). As a reference, a fragment of the cytoskeletal actin-encoding cDNA was amplified, using Act6RT-F1 and Act6RT-R1 (Table S1). The products of the PCR were separated in an agarose gel, stained with ethidium bromide, and quantified by ImageJ software (http://rsbweb.nih.gov/ij/).
In situ hybridization
The 1 kb fragment of the cDNA described above was inserted into pBluescript II SK+ (Stratagene). The cDNA was then amplified by PCR using M13-forward and M13-reverse primers. Using the product of the PCR as a template, digoxigenin-labeled RNA probe was prepared, according to the protocol supplied by the manufacturer (Roche). T3 RNA polymerase (Promega) was used to produce an antisense probe. Buds on glass slides were fixed on ice with 4% paraformaldehyde in phosphate-buffered saline for 12 h. In situ hybridization was performed as described by Sunanaga et al. (2007). After the immunological detection of hybridization signals, the specimens were embedded in JB4 plastic resin (Polyscience), and sectioned at a thickness of 2.0–2.5 μm.
The 506 bp and 1 kb cDNA fragments described above were used as templates for production of dsRNA. Each strand of the cDNA was separately transcribed using T3 RNA polymerase or T7 RNA polymerase (Takara). After treatment with DNaseI, the RNA samples were purified by phenol/chloroform extraction and ethanol precipitation. The RNA samples were mixed in a solution containing 750 mmol/L NaCl and 75 mmol/L sodium citrate. The mixture was denatured at 65°C for 10 min, and slowly cooled to form dsRNA. After phenol/chloroform extraction and ethanol precipitation, dsRNA was dissolved in sterile water.
Growing buds were separated from the parental body with a razor blade. The buds were treated with the dsRNA solution at room temperature for 1 h. The buds were cultured for a week, and fixed with 5% formalin for 30 min. The staining for the β-galactosidase (β-gal) activity was carried out as described by Kaneko et al. (2010).
Characterization of the P. misakiensis myc cDNA clones
Using PCR-based techniques, we obtained nine clones of cDNA homologous to the proto-oncogene myc, using a P. misakiensis library as a template. The clones contained the entire translated region encoding a polypeptide of 639 amino acids (Fig. S1). They were divided into two groups; each consisted of four and five clones (Figs S1, S2). Base substitutions within each group were <0.4% (data not shown). In contrast, 42 of 1959 nucleotides (2.1%) and 12 of 639 amino acid residues (1.9%) were different between the groups (Figs S1, S2). Although we infer that they are allelic variants (as discussed below), we tentatively named them Pm-myc1 and Pm-myc2. Their sequences have been submitted to the DNA Data Bank of Japan/European Molecular Biology Laboratory (DDBJ/EMBL)/GenBank databases under accession numbers AB669184 and AB669185. Analyses of expression and function described below were not designed to distinguish Pm-myc1 and Pm-myc2 because of high sequence similarity. In such cases, we collectively call them Pm-myc. The difference was concentrated in the C-terminal half of the coding region (Fig. S2). Amino acid sequences of the Myc box I (MbI), Myc box II (MbII) and basic helix-loop-helix (bHLH) motifs were completely identical between Pm-myc1 and Pm-myc2 (Figs 1, S2).
We kept four distinct strains of P. misakiensis, which were allowed to reproduce only asexually. It was possible that Pm-myc1 and Pm-myc2 derived from two different strains. RT–PCR was carried out using the Pm-myc1-specific or Pm-myc2-specific primer set and total RNA extracted from a single colony as a template. Individuals comprising a single colony were asexually reproduced clones. The two primer sets generated the product almost at the same level, indicating that both Pm-myc1 and Pm-myc2 derive from the genome of a single strain (Fig. S3).
MbI and MbII are Myc-specific motifs, involved in Myc’s biological activities, such as transcriptional regulation and cell cycle progression (e.g. Oster et al. 2003). The amino acid sequence of MbI in Pm-Myc was about 79% (11/14) identical to that in human Myc proteins (Fig. 1a). Phosphorylation at the threonine and serine residues within MbI is functionally important (Lutterbach & Hann 1994). These residues were conserved in Pm-Myc (Fig. 1a). The amino acid sequence of MbII in Pm-Myc was 57% (4/7) identical to that in human Myc proteins (Fig. 1b). The tryptophan residue in MbII, important for Myc’s functions (Brough et al. 1995), was also conserved in Pm-Myc (Fig. 1b). The bHLH domain is responsible for binding DNA and dimerization with the Max protein (e.g. Blackwood et al. 1991; Kerkhoff & Bister 1991; Amati et al. 1992). The bHLH sequence in Pm-Myc was about 50% identical to that in other Myc proteins examined (Fig. 1c).
Figure 1d shows a molecular phylogenetic tree, constructed with the maximum likelihood method (Felsenstein 1981). Pm-Myc1 and Pm-Myc2 clustered with the Myc protein of another ascidian species, Ciona intestinalis (Fig. 1d). The C. intestinalis genome contained a single myc homologue (Satou et al. 2003). Vertebrates have three paralogous genes, called c-myc, N-myc, and L-myc (Nesbit et al. 1999). Proteins encoded by these genes formed a cluster (Fig. 1d). The result suggested that these three genes have arisen by duplication during the evolution of vertebrates.
Expression of Pm-myc during bud development
The bud of P. misakiensis arises as a protrusion of the parental body wall (Fig. 2a). The body wall consists of the epidermis and atrial epithelium (Fig. 2a). Between these epithelia, there are several types of mesenchyme cells. Organogenesis in the bud starts after separation from the parent (Fig. 2a–c). Transdifferentiation of the atrial epithelium occurs to form the gut primordium at the proximal end of the bud (Fig. 2b; Fujiwara & Kawamura 1992; Kawamura & Fujiwara 1994). The branchial primordium also derives from the atrial epithelium (Fig. 2c). The Pm-myc mRNA was detected strongly in the proximal half of the developing bud (Fig. 2d). The atrial epithelium was one of the main tissues that expressed Pm-myc (Fig. 2d). Pm-myc-expressing epithelial cells looked cuboidal and had a relatively large nucleus, indicating that they were dedifferentiated (Fig. 2g,h). The branchial and gut primordia strongly expressed Pm-myc (Fig. 2g,h). A type of mesenchyme cell near the organ primordia also expressed Pm-myc (Fig. 2g,h). Fibroblast-like cells near the organ primordia are integrated into the epithelium and contribute to organogenesis (Kawamura et al. 1991, 2008a). The morphology and localization of the Pm-myc-positive cells were similar to those of the fibroblast-like cells (Fig. 2g,h). The epidermis did not express Pm-myc anywhere in the bud (Fig. 2d,e). The mRNA was not observed in the atrial epithelium of the distal region (Fig. 2d,f). RT–PCR analysis supported that the level of expression of Pm-myc was higher in the posterior half of the developing bud (Fig. 2i,j).
Since the stronger expression in the posterior region was evident as early as 24 h after separation from the parent (Fig. 2j), we examined the expression of Pm-myc before separation from the parent. We observed the expression of Pm-myc in the atrial epithelium of the growing buds (Fig. 3). The level of expression was higher in the proximal region (Fig. 3). Mesenchyme cells did not express Pm-myc before the bud separated from the parent (Fig. 3). No signal was observed in any tissue or organ in the parental body (Fig. 3e).
Disruption of the function of Pm-myc by RNAi
We attempted RNA interference (RNAi) to assess the requirement of Pm-myc function in the process of bud development. We used the 506 bp double-stranded RNA (dsRNA) corresponding to the 5′ untranslated region (5′ UTR) and a part of the translated region encompassing the MbI and MbII motifs (see Figs S1, S2). The nucleotide sequence of this region is identical between Pm-myc1 and Pm-myc2, except for a single nucleotide substitution (Fig. S1). To obtain staged developing buds, we cut the proximal part of growing buds with a razor blade. Just after the amputation, the buds were soaked in a solution containing 0.2 μg/μL of dsRNA for 1 h. The buds were then allowed to develop for 7 days. We examined the activity of β-gal in the digestive tract to evaluate the effect of the dsRNA treatment (Fig. 4a–c). As a control, we used a lacZ-specific dsRNA. Although lacZ encodes β-gal in Escherichia coli, the β-gal activity in the P. misakiensis digestive tract cannot be suppressed by the lacZ-specific dsRNA because their sequences are unrelated (K. Kawamura, unpubl. data, 2003). In a control experiment, eight of nine buds treated with lacZ-specific dsRNA developed normally, and showed strong β-gal activity in the stomach and intestine (grade 3: Fig. 4c,d). The other bud did not express the β-gal activity (grade 1; Fig. 4d). In contrast, seven of 10 buds treated with Pm-myc-specific dsRNA exhibited faint β-gal signals in the relatively undeveloped intestine (grade 2; Fig. 4b,d). The other two buds did not show β-gal activity (grade 1; Fig. 4a,d). Only one individual expressed strong β-gal activity (grade 3; Fig. 4d). Similar results were obtained when dsRNA was prepared from a 1 kb cDNA fragment corresponding to a part of the translated region (data not shown).
Since the developmental defects were not severe, we examined the formation of the gut primordium at an earlier stage (Fig. 5). In this experiment, buds were treated with 0.5 μg/μL dsRNA before amputation. Two days after separation from the parent, the buds were fixed and sectioned for histological analysis. The experimental buds showed various phenotypes, categorized into five grades (Fig. 5a–e). The grade 1 buds did not show any indication of organogenesis. The atrial epithelium stayed squamous (Fig. 5a). In the grade 2 buds, the atrial epithelium more or less invaginated into the mesenchymal space (Fig. 5b). However, the epithelial cells were all squamous, and their nucleus was not swollen (Fig. 5b). This morphology suggests that the epithelial cells were not dedifferentiated. In the grade 3 buds, the atrial epithelium became cuboidal, and large nuclei were obvious (Fig. 5c). However, the epithelium was not invaginated (Fig. 5c). The grade 4 buds developed a gut primordium (Fig. 5d). Morphological changes in the atrial epithelium were indistinguishable from those observed in normally developing buds (Fig. 5d). However, the size of the gut primordium was relatively small (Fig. 5d). The grade 5 buds developed a large gut primordium of normal morphology (Fig. 5e). The results of this experiment are summarized in Figure 5f. Seven of 10 control buds, treated with lacZ-specific dsRNA, developed a gut primordium (grade 4 or 5). The other three buds were grade 3. In contrast, 10 of 12 experimental buds did not develop a gut primordium (grade 1, 2 or 3). The other two buds formed a gut primordium (grade 4 or 5).
As well as morphology, Figure 5 shows in situ staining of the Pm-myc mRNA in dsRNA-treated buds. Although panels (a–c) are examples of the buds treated with Pm-myc-specific dsRNA, the atrial epithelium exhibits strong signals. However, there was no Pm-myc-positive fibroblast-like cell in the mesenchymal space (Fig. 5a–c). In contrast, we observed many Pm-myc-positive mesenchyme cells in control buds, treated with lacZ-specific dsRNA (Fig. 5d,e). The amount of the Pm-myc mRNA was examined by a semi-quantitative RT–PCR in the dsRNA-treated buds (Fig. S4). As a reference, the amount of the cytoskeletal actin Pm-Act6 was monitored (Fig. S4; Fujiwara et al. 1997). Pm-Act6 was exponentially amplified from the 18th to 23rd cycles (Fig. S4a,c). Pm-myc was exponentially amplified from the 22nd to 26th cycles (Fig. S4b,d). We therefore chose the signals obtained at the 22nd and 23rd cycles for quantification. We calculated the relative amount of the PCR product corresponding to Pm-myc, with reference to that of Pm-Act6. Then, the relative amounts were compared between the buds treated with Pm-myc-specific dsRNA and lacZ-specific dsRNA. The average ratio of the relative amounts was 1.4. The data support the results of in situ hybridization. The gross amount of the Pm-myc mRNA did not decrease on treatment with the Pm-myc-specific dsRNA, although the data may not tell about the change in the mesenchymal expression.
Premature expression of Pm-myc prior to dedifferentiation of the atrial epithelium
We identified two different but very similar cDNA sequences, Pm-myc1 and Pm-myc2, from P. misakiensis. Their predicted amino acid sequences were 98% identical, while their functional domains (MbI, MbII and bHLH) were 100% identical. In the genome of C. intestinalis, the rate of allelic polymorphisms within a single individual is 1.2% (Dehal et al. 2002). The genome of the amphioxus Branchiostoma floridae contains 3.7% single nucleotide polymorphisms (Putnam et al. 2008). Given the high rate of allelic polymorphisms observed in non-vertebrate chordates, it is not strange that the P. misakiensis genome has a polymorphism rate of 2%. Since our probes for in situ hybridization and dsRNA for RNAi should not distinguish between Pm-myc1 and Pm-myc2, the two sequences are collectively referred to as “Pm-myc” in the following discussion.
The finding that Pm-myc was expressed in the atrial epithelium of the growing bud still attached to the parental body was unexpected. It is evident that the atrial epithelium, before separation from the parent, consists of quiescent cells, staying in the G0 phase of the cell cycle (Kawamura & Nakauchi 1991). Dedifferentiation and proliferation of the atrial epithelium start at least one day after separation from the parent (Kawamura & Nakauchi 1986; Fujiwara & Kawamura 1992; Kawamura & Fujiwara 1994). In vertebrates, expression of myc in quiescent cells immediately triggers re-entry into the cell cycle (e.g. Eilers et al. 1991). The half-life of the human c-Myc protein is about 25 min (Hann & Eisenman 1984). In contrast, Myc’s heterodimerization partner, Max, is extremely stable and ubiquitously expressed (Blackwood et al. 1992). Therefore, only a transient upregulation of Myc expression is sufficient to restart the cell cycle. During the transdifferentiation of retinal pigment epithelial cells to lens cells, expression of c-myc occurs concurrently with dedifferentiation (Agata et al. 1993). In another budding ascidian species, Botryllus primigenus, a homologue of myc is expressed in undifferentiated proliferative cells (Kawamura et al. 2008b). It is probable that in P. misakiensis, expression of myc has started at the beginning of bud formation. The myc-expressing atrial epithelium would come from a differentiation-competent, unlike the myc-negative parental atrial epithelium.
Given that Pm-myc is expressed more than 1 day before the onset of mitosis, it is possible that the translation or function of the Pm-Myc protein is repressed in the growing bud. A calcium-dependent galactose-binding lectin, TC14-3, was expressed in the atrial epithelium and mesenchyme cells in the growing bud (Matsumoto et al. 2001). The TC14-3 mRNA disappeared from the region where the atrial epithelium dedifferentiates (Matsumoto et al. 2001). TC14-3 inhibited the growth of a cell line derived from the atrial epithelium (Matsumoto et al. 2001). From these results, TC14-3 is thought to stabilize the differentiated state of the atrial epithelium. Therefore, TC14-3 is a candidate suppressor of Pm-Myc. More generally, Myc’s function can be regulated by external signals and interaction with other proteins (Pelengaris & Khan 2003). Mitogen-activated protein kinase phosphorylated the MbI motif (Lutterbach & Hann 1994). The stress-responsive kinase Pak2 phosphorylated the bHLH domain and suppressed dimerization and DNA-binding of Myc (Huang et al. 2004). Mad family proteins, Mnt, and some other proteins regulate the activity of Myc by competing for Max as a partner of dimerization (Zhou & Hurlin 2001; Walker et al. 2005). The possible involvement of these factors should be examined to understand the regulation of Pm-Myc in the growing bud.
Retinoic acid is synthesized in the proximal region of the developing bud several hours after separation from the parent (Kawamura et al. 1993). In response to retinoic acid, mesenchyme cells secrete soluble factor(s) that directly induce the atrial epithelium to dedifferentiate (Hara et al. 1992; Ohashi et al. 1999). If the growing bud is cut at both proximal and distal ends, retinoic acid synthase becomes activated only in the proximal region (Kawamura et al. 1993). This result indicated the existence of proximal-distal polarity in the growing buds. The proximal-distal concentration gradient of the Pm-myc mRNA is a new indication of the polarity, although its functional relevance is unclear. In the developing bud, retinoic acid and its downstream factors may contribute to the activation of the Pm-Myc protein or maintenance of the transcription of Pm-myc. However, these factors are not responsible for the polarity of the growing bud and initial transcriptional activation of Pm-myc.
Requirement of Pm-myc for the organogenesis in the bud
Treatment of the bud with Pm-myc-specific dsRNA caused defects in the formation of the gut. In some cases, the epithelial cells did not change their shape from squamous to cuboidal. They did not invaginate into the mesenchymal space. No such effect was observed in control buds, suggesting that the dsRNA specifically suppressed dedifferentiation of the atrial epithelium. However, degradation of the Pm-myc mRNA by treatment with dsRNA was not effective in the present study. The intensity of the signals obtained by in situ hybridization was almost at the same level in the atrial epithelium between the experimental and control buds. Results of the semi-quantitative RT–PCR supported the persistence of the Pm-myc mRNA. The amount of the Pm-myc mRNA seemed slightly increased by treatment with dsRNA. At present, it is not clear whether the dsRNA somehow stimulated upregulation of Pm-myc. Given that the gut formation was prevented, it is possible that the dsRNA inhibited the translation of the Pm-Myc protein. However, a more plausible explanation is that suppression of Pm-myc in the fibroblast-like mesenchyme cells resulted in the impaired organogenesis. Indeed, the Pm-myc-positive mesenchyme cells were not observed in the experimental bud. Tissue specificity of the sensitivity to RNAi has been reported. Neuronal tissues are less sensitive to RNAi in Caenorhabditis elegans (Timmons et al. 2001; Kamath & Ahringer 2003). Many possibilities have been proposed, including differences in the efficiency of uptake and metabolism of dsRNA (Timmons et al. 2001). Mesenchyme cells in P. misakiensis may be more sensitive to RNAi than the atrial epithelium.
Kawamura et al. (1991) showed that the gut primordium was not formed when the buds were injected with an antibody against the TC14-1 lectin. TC14-1 is a component of the extracellular matrix formed in the mesenchymal space around the gut primordium (Kawamura et al. 1991). The antibody affected the formation of pseudopods of the mesenchyme cells and their integration into the atrial epithelium (Kawamura et al. 1991). These observations suggested that the integration of mesenchyme cells is an essential step for the formation of the gut primordium. During the mesenchymal-epithelial transition, the fibroblast-like cells also enter the cell cycle (Kawamura et al. 1988, 1995). The dsRNA may have inhibited this process and affected the epithelialization of these mesenchyme cells. Given that the dsRNA prevented dedifferentiation without affecting the expression of Pm-myc in the atrial epithelium, the Pm-myc-positive mesenchyme cells may have a crucial role in the induction of dedifferentiation.
We thank Zenji Imoto at the Usa Marine Biological Institute of Kochi University for maintenance of the aquarium. We also thank Takeshi Sunanaga and the other members of our laboratory for helpful advice and discussions. This work was supported in part by the Japan Society for the Promotion of Science (Grant number 19570220).