Regeneration of the gut requires retinoic acid in the budding ascidian Polyandrocarpa misakiensis

Authors


Author to whom all correspondence should be addressed.
Email: tatataa@kochi-u.ac.jp

Abstract

The protochordate ascidian Polyandrocarpa misakiensis has a striking ability to regenerate. When the posterior half of the adult body is amputated, the anterior half completely loses the esophagus, stomach and intestine. These organs are reconstituted in a week. Histological observation revealed that the regeneration involves transdifferentiation of the atrial epithelium near the cut surface. The morphological features of the gut primordium were similar to those observed in the developing bud of this species. Inhibitors of the synthesis of retinoic acid (RA) suppressed the formation of the gut. 13-cis RA rescued the regenerates from the inhibitor-induced hypoplasia. These results suggest that RA is required for the regeneration of the gut. A gene encoding the RA receptor (Pm-RAR) and its target gene, TRAMP, were expressed in and around the regenerating gut. Pm-RAR-specific and TRAMP-specific double-stranded RNA molecules inhibited the regeneration of the gut, indicating that the RA signal is mediated at least in part by Pm-RAR and TRAMP. These results suggested that RA triggers the transdifferentiation of the atrial epithelium into the gut in regenerating animals, as it does during asexual reproduction.

Introduction

The budding ascidian Polyandrocarpa misakiensis belongs to the phylum Chordata. It has a remarkable ability to regenerate. A half fragment of the adult body regenerates a complete body in one week (Fig. 1A). Even one-tenth body fragments can survive and regenerate. Although vertebrates also belong to the phylum Chordata, their ability to regenerate organs is restricted (reviewed in Stoick-Cooper et al. 2007). Some urodele amphibian species are known to regenerate their limbs, tail, lenses and many other organs (reviewed in Roy & Lévesque 2006). The lens can be reconstituted by transdifferentiation of the iris. The pigmented epithelial cells of the iris and retina possess the ability to transdifferentiate in vitro into lens cells in many non-amphibian vertebrate species (Kodama & Eguchi 1995). The capability of transdifferentiation is, however, somehow suppressed in most of these species. Ascidians have a well-developed digestive tract, heart, and central nervous system, comprising a chordate-specific architecture of the body (Fig. 1B). All of these organs can be reconstructed even when they are completely lost (Fig. 1A).

Figure 1.

 Regeneration and bud development in Polyandrocarpa misakiensis. (A) The left panel shows an adult just after it was cut into two pieces. The dorsal side is shown, with the animal’s anterior oriented toward the top. The gut is in orange. The middle panel shows the regenerating halves one day after the amputation. The right panel shows the regenerates 7 days after the amputation. An orange gut was seen in both halves. (B) The left panel shows an adult that has two growing buds (gb). The right panel shows a diagram of the individual. ae, atrial epithelium; bs, branchial sac; ep, epidermis; he, hemocoel; in, intestine; st, stomach. (C) Schematic diagrams showing the process of bud development. The growing bud (gb) is an outgrowth of the adult body wall, consisting of the epidermis (ep) and atrial epithelium (ae). The space between these two epithelia is the hemocoel. The development of the bud proceeds in the direction of the arrows. After separation from the parent, cell differentiation and morphogenesis start. The gut primordium (gp) is formed from foldings of the atrial epithelium at the proximal end of the bud (the site of separation from the parent). Within 3 days after the separation, the pharyngeal primordium (pp) develops.

Among ascidian species, a marked regenerative capacity is observed in those that can reproduce asexually (e.g. Oka & Watanabe 1959; Rinkevich et al. 1995). Ciona intestinalis is a solitary ascidian that does not reproduce asexually. Adult C. intestinalis individuals can regenerate the entire central nervous system (Dahlberg et al. 2009). However, they cannot survive when they are cut in two. Polyandrocarpa misakiensis reproduces asexually by palleal budding (Kawamura & Watanabe 1982; Fig. 1C). The bud appears as an outgrowth of the parental body wall (Fig. 1C). The bud separates from the parent, and starts cell differentiation and morphogenesis (Fig. 1C). During the bud’s development, the atrial epithelium transdifferentiates into the digestive tract (Fujiwara & Kawamura 1992; Kawamura & Fujiwara 1994). When the posterior half of the adult body is amputated, the anterior half does not contain the esophagus, stomach and intestine (Fig. 1A). These organs are formed from the atrial epithelium near the cut surface through a morphallactic process (Katsuyama 1992). Therefore, both bud development and regeneration may employ similar molecular mechanisms for the formation of the gut.

Retinoic acid (RA) can induce an ectopic gut formation, resulting in the complete duplication of bud organs (Hara et al. 1992; Kawamura et al. 1993). RA-treated mesenchyme cells can also induce the duplication of bud organs (Kenji Hara, pers. comm., 1992). The enzymatic activity of aldehyde dehydrogenase (ALDH), a candidate RA synthase, is detected at the proximal end (morphogenetic region) of the developing bud (Kawamura et al. 1993). A homologue of the RA receptor (RAR) was identified in this species, as the first invertebrate RAR (Hisata et al. 1998). Its gene, named Pm-RAR, was revealed to function as an RA-dependent transcriptional activator (Kamimura et al. 2000). The Pm-RAR mRNA is expressed in mesenchyme cells during budding (Hisata et al. 1998). RA induces the expression of a serine protease, named tunicate retinoic acid-inducible modular protease (TRAMP), in the mesenchyme cells (Ohashi et al. 1999). A recombinant protein containing the catalytic domain of TRAMP stimulated the proliferation (and possibly dedifferentiation) of a cell line derived from the atrial epithelium (Ohashi et al. 1999). These observations suggest that Pm-RAR and TRAMP play an important role in the transdifferentiation of the atrial epithelium in the bud (Kawamura & Fujiwara 2000; Fujiwara et al. 2001). It is of interest whether the formation of the gut in regenerates involves the transdifferentiation of the atrial epithelium.

In the present study, the adult body of P. misakiensis was cut antero-posteriorly into two pieces, and the process of the gut’s regeneration in the anterior half was observed using histological sections. Regenerates were treated with RA or inhibitors of RA synthesis. Inhibition of RA synthesis suppressed the regeneration, and this was rescued by RA treatment. The expression of the Pm-RAR and TRAMP mRNAs was observed in the atrial epithelium and some cell types around the gut-forming field of the regenerates. Treatment with Pm-RAR-specific or TRAMP-specific double-stranded RNA (dsRNA) caused inhibition of the gut’s regeneration. These results suggest that endogenous RA is required for the gut to regenerate, and that the RA signal is mediated by Pm-RAR and TRAMP.

Materials and methods

Animals

Polyandrocarpa misakiensis was cultivated on glass slides in boxes settled in Tosa Bay close to the Usa Marine Biological Institute of Kochi University. Growing buds were surgically separated from the parent with a razor blade. The buds were allowed to develop to functional zooids for 2 weeks in artificial seawater, “Super Marine Art SF1” (Tomita Pharmaceuticals). After the branchial and atrial siphons had opened, the animals were provided with 0.25% Sun-Culture diatoms (Yamaha, Shizuoka, Japan) as food. Adult individuals were used for regeneration experiments before they started budding. Six hours before the experiments, the animals were moved into foodless artificial seawater, so that the digestive tract became empty.

The posterior half of the animals was amputated with a razor blade (Fig. 1A). A new razor blade was used for each individual. The anterior halves were allowed to regenerate in the Super Marine Art SF1 at 20°C for 0–10 days. The artificial seawater was prepared using non-sterilized distilled water, and did not contain antibiotics. The artificial seawater was changed every 2 or 3 days.

Drug administration

13-cis RA, 3,7-dimethyl-2,6-octadienal (citral) and 4-diethylaminobenzaldehyde (DEAB) were purchased from Sigma-Aldrich. A stock solution of 13-cis RA was prepared in dimethylsulfoxide (DMSO) at a concentration of 100 mmol/L. Citral and DEAB were dissolved in DMSO at 20–120 mmol/L and 10–500 mmol/L, respectively. When the animals were to be treated with 1 μmol/L 13-cis RA, the stock solution was once diluted in DMSO to make a 1000× (1 mmol/L) concentrated solution. The 1000× solution was then diluted in the artificial seawater to make a working solution. The other drugs were diluted in a similar way. Control animals were treated with 0.1% DMSO. The regenerates were soaked in the drug-containing artificial seawater just after they were cut. They were kept shaded by aluminum foil and incubated at 20°C for 10 days. The citral- or DEAB-containing seawater was changed every 2 days.

Detection of the activity of β-gal

Regenerating animals were fixed with 5% formalin at room temperature for 30 min. The samples were washed three times with phosphate-buffered saline (PBS) (137 mmol/L NaCl, 2.7 mmol/L KCl, 10 mmol/L Na2HPO4, and 2 mmol/L KH2PO4, pH 7.4) for 5 min. They were stained at 37°C for 24 h in a solution containing 0.5 mg/mL of X-gal (Promega), 40 mmol/L sodium phosphate buffer (pH 7.2), 5 mmol/L potassium ferrocyanide, 5 mmol/L potassium ferricyanide, 150 mmol/L NaCl, 2 mmol/L MgCl2 and 0.1% Tween20. The specimens were sequentially immersed in 50%, 70%, 90% and 100% ethanol, and observed from the ventral side in 100% ethanol.

RNA interference

Three different regions of the Pm-RAR cDNA (RAR-1i, RAR-2i and RAR-3i) were amplified by polymerase chain reaction (PCR), using pCMX-(h+Pm) RAR (Kamimura et al. 2000) as a template (Supplementary Fig. S1). Three different regions of the TRAMP cDNA (TRAMP-1i, TRAMP-2i and TRAMP-3i) were amplified by PCR, using the cDNA clone obtained by Ohashi et al. (1999) as a template (Supplementary Fig. S1b). Primers used for the PCR are listed in Supplementary Table S1. Products of the PCR were inserted into the recognition site of EcoRV in pBluescript II SK+ (Stratagene). Each cDNA fragment was then amplified by PCR using M13-forward and M13-reverse primers. Using the products of the PCR as templates, single-stranded RNA was synthesized using T7 RNA polymerase (Takara) or T3 RNA polymerase (Promega). The RNA synthesis reaction was performed in a 100 μL solution at 37°C for 3 h. The RNA products were then incubated with 0.2 units/μL of DNaseI (Roche) at 37°C for 1 h. After phenol/chloroform extraction and ethanol precipitation, the samples were dissolved in 24 μL of sterile RNase-free water. The single-stranded RNA samples were denatured at 65°C for 5 min. Complementary strands were mixed and further incubated at 65°C for 15 min. After 16 μL of 20× standard saline citrate (SSC) [3 mol NaCl and 0.3 mol sodium citrate (pH 7.0)] was added, the RNA mixture was slowly cooled to form dsRNA. The dsRNA samples were then incubated with 15 μg/mL of RNaseA at room temperature for 1 h. The formation of double strands was confirmed by an agarose gel electrophoresis. The RAR-1i, RAR-2i and RAR-3i dsRNA samples were mixed, as were the TRAMP-1i, TRAMP-2i and TRAMP-3i dsRNA samples. After the quantity of dsRNA was determined, the samples were diluted with the artificial seawater at a concentration of 0.6 μg/μL, and used for treating regenerates.

The animals were cut into two pieces as described above, and the anterior halves were used. Water around the animals was wiped off. The animals were then immersed in 10 μL of the diluted dsRNA solution at 20°C for 1 h. During the dsRNA treatment, the animals were kept in a moist chamber. The animals were allowed to regenerate in the artificial seawater at 20°C for a week. Regeneration of the gut was examined with X-gal staining as described above.

Reverse transcription–polymerase chain reaction

Total RNA was extracted from regenerating animals as described by Fujiwara et al. (1993). After the quantity was determined, 30 μg of the RNA samples were dissolved in 50 μL of a solution containing 100 mmol/L sodium acetate (pH 5.2), 5 mmol/L MgSO4, 0.8 units/μL of RNase inhibitor (Wako) and 0.2 units/μL of DNaseI (Roche). The solution was incubated at 37°C for 30 min. DNaseI was inactivated by heating at 80°C for 10 min, and purified by phenol/chloroform extraction and ethanol precipitation. Purified RNA samples were dissolved in sterile RNase-free water. Reverse transcription (RT) was carried out using 1 μg of the RNA sample as a template, the primer [5′-T17(A/C/G)-3′], and MMLV reverse transcriptase (Promega). After the enzyme was inactivated by heating at 95°C for 5 min, cDNA products were diluted with 60 μL of Tris-EDTA (TE) (10 mmol/L Tris-HCl, 1 mmol/L ethylenediaminetetraacetic acid [EDTA], pH 7.5). Polymerase chain reaction (PCR) was carried out using 2 μL of diluted cDNA as a template, and primers listed in Supplementary Table S2. Products of the PCR were fractioned on an agarose gel, and stained with ethidium bromide. Signals of the PCR products were quantified using Image J software (http://rsbweb.nih.gov/ij/). The ratio of the fluorescent intensity of Pm-RAR-derived to β-actin-derived bands (R:A) was calculated. Similarly, the ratio of the intensity of TRAMP-derived to β-actin-derived bands (T:A) was calculated. The R:A and T:A values were further relativized, taking the values at day 0 of regeneration (just after the animals were cut) to be 1.0.

In situ hybridization and histological sections

Antisense and sense RNA probes for Pm-RAR and TRAMP were labeled with digoxigenin (DIG), according to the protocol supplied by Roche. The templates used for the synthesis of probes were those used for the production of dsRNA (Supplementary Fig. S1). Regenerating animals on glass slides were fixed with 4% paraformaldehyde in PBS at 4°C for 14 h. In situ hybridization was carried out as described by Sunanaga et al. (2006). The left or right edge or the anterior tip of regenerating animals was cut off before treatment with proteinase K, to facilitate the infiltration of RNA probes. After 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.

Regenerating animals were also embedded in Technovit 8100 resin (Heraeus Kulzer). In this case, the animals were fixed with Zamboni’s solution (Zamboni & de Martino 1967) on ice for 30 min. The samples were dehydrated by ethanol, and embedded in Technovit 8100, according to the protocol supplied by the manufacturer. The samples were sectioned at a thickness of 2.0–2.5 μm, and stained with Giemsa’s solution (Muto Pharmaceuticals).

Results

Regeneration of the gut in the anterior half of the adult P. misakiensis body

The adult bodies of P. misakiensis were cut antero-posteriorly into two pieces, and the process of regeneration of the anterior halves was histologically examined. Since the pharynx of ascidians occupies almost the entire atrial cavity to form a large branchial sac, the anterior half does not contain the posterior components of the gut (the esophagus, stomach and intestine) (see Fig. 1A,B). Soon after the amputation, the cut surface rapidly healed (see Fig. 1A), and the atrial epithelium became continuous (Fig. 2A). Within the first 2 days, a small region of the atrial epithelium near the cut surface invaginated into the mesenchyme space to form the gut primordium (Fig. 2A,B). The invaginating epithelial cells were cuboidal, while the other parts of the atrial epithelium were squamous (Fig. 2B). The density of mesenchyme cells around the primordium was low, and the extracellular matrix (ECM) developed to form a web of filaments (Fig. 2B). The overall morphology around the gut primordium was similar to that observed in the developing bud (Kawamura et al. 2008). The stomach and intestine became differentiated 3–4 days after the amputation (Fig. 2C–F). Looping of the intestine was observed (Fig. 2E,F). A week after the amputation, the gastric epithelium gradually thickened and the pyloric caecum became prominent (Fig. 2G). The esophageal epithelium became folded (Fig. 2H).

Figure 2.

 Histological sections of regenerating anterior halves of Polyandrocarpa misakiensis. All panels show frontal sections. The cut surface is oriented to the left, and the anterior is to the right. Scale bars indicate 50 μm. ae, atrial epithelium; bs, branchial sac; ECM, extracellular matrix; en, endostyle; ep, epidermis; es, esophagus; gp, gut primordium; in, intestine; pc, pyloric caecum; pe, perivisceral epithelium; st, stomach; tu, tunic. (A) A gut primordium (gp) is formed 2 days after the amputation. (B) A high magnification view of the boxed region in (A). (C) A gut primordium 3 days after the amputation. (D) A high magnification view of the boxed region in (C). (E) A regenerate 4 days after the amputation. (F) A high magnification view of the boxed region in (E). (G) A regionalized digestive tract 7 days after the amputation. The esophagus (es), stomach (st), intestine (in) and pyloric caecum (pc) are well developed. (H) An esophagus 7 days after the amputation.

The effect of inhibitors of RA synthesis on the regeneration of the gut

Since RA regulates the formation of the gut during bud development, it was examined whether RA was required for the regeneration of the gut. Citral and DEAB, known as inhibitors of RA synthesizing ALDH in vertebrates (Connor & Smit 1987; Russo et al. 1988), were used. The anterior halves of the P. misakiensis adults were allowed to regenerate in the presence of citral (20–120 μmol/L) or DEAB (10–500 μmol/L) for 10 days. Differentiated stomach and intestine show β-galactosidase (β-gal) activity (K. Kawamura, unpubl. data, 2003). The experimental animals were stained with X-gal to examine whether they regenerated these organs. Both drugs caused a series of phenotypes (Fig. 3), which were categorized into four types (I–IV). Type I looked normal, with a spherical stomach and looping intestine heavily stained by X-gal (Fig. 3A). In the Type II animals, the staining was weak (Fig. 3B). Both the stomach and intestine looked small (Fig. 3B). The Type III animals did not show β-gal activity, although they were apparently alive, with their cut surface completely healed (Fig. 3C). The Type IV animals were dead, with their cut surface not healed (Fig. 3D). Table 1 summarizes the results of the drug treatment. Of 23 control regenerates treated with DMSO, 20 (87%) showed the Type I phenotype. The formation of the gut was inhibited (Type III) in eight (62%) of the 13 regenerates treated with 60 μmol/L citral, and in all of the 15 regenerates treated with 120 μmol/L citral (Table 1, Fig. 4). None of them showed the Type IV phenotype, indicating that citral did not kill the animals at a concentration of no more than 120 μmol/L. The gut’s formation was suppressed (Type III) in 24 (80%) of the 30 animals treated with 100 μmol/L DEAB. DEAB did not kill the animals at a concentration of less than 250 μmol/L. Figure 4 represents the ratio of animals that developed a gut even though it was small (Types I and II). It shows that citral and DEAB affected the regeneration of the gut in a dose-dependent manner.

Figure 3.

 Phenotypes of regenerates, obtained by drug treatments. The adult individuals were cut antero-posteriorly into two fragments. The anterior halves were allowed to regenerate for 10 days. On the 10th day, the activity of β-gal was examined as a differentiation marker of the stomach and intestine. in, intestine; st, stomach. (A) The stomach and intestine are clearly stained in indigo blue in normally regenerating animals (Type I). (B) The stomach and intestine are seen, even though the staining is weak and the size is small (Type II). (C) No staining was observed, although the cut surface was healed (Type III). (D) The anterior halves cannot survive in the presence of excessive doses of drugs (Type IV).

Table 1.   Effect of inhibitors of RA synthesis on the regeneration of the gut
InhibitorsConcentration (μmol/L)Number of casesPhenotypes (%)
Type IType IIType IIIType IV
  1. †Control animals were treated with 0.1% dimethylsulfoxide (DMSO).

Control† 2387490
 Citral2014860140
4016446500
6013318620
12015001000
 DEAB1010800200
10030137800
25026001000
Figure 4.

 Effect of citral and 4-diethylaminobenzaldehyde (DEAB) on the regeneration of the gut. The ratio of regenerates that formed a gut (Types I and II in Fig. 3) is expressed as a percentage. Control animals were treated with 0.1% dimethylsulfoxide (DMSO) (lane 1). Regenerates treated with 20, 40, 60 or 120 μmol/L citral (lanes 2–5, respectively). Regenerates treated with 10, 100 or 250 μmol/L DEAB (lanes 6–8, respectively).

Rescue by RA from the impaired gut formation

To assess whether the suppression of the gut’s regeneration resulted only from specific inhibition of RA synthesis by the drugs, we attempted to rescue the regenerating animals through treatment with RA. We used 13-cis RA because it induced an ectopic gut to form without obvious toxicity, when applied to the developing P. misakiensis bud (Kawamura et al. 1993). The anterior halves of the P. misakiensis adults were allowed to regenerate in the presence of 90 μmol/L citral or 75 μmol/L DEAB for 10 days. The animals were treated with 0.1–2.0 μmol/L 13-cis RA during the first 4 h. The formation of the gut was suppressed in 11 (33%) of the 33 regenerates treated with 90 μmol/L citral (Table 2, Fig. 5). Treatment with 13-cis RA rescued the gut-forming activity of regenerates in a dose-dependent manner (Table 2, Fig. 5). All of the animals simultaneously treated with 90 μmol/L citral and 1 μmol/L 13-cis RA developed a gut (Types I and II; Table 2, Fig. 5). 13-cis RA seemed to have a toxic effect at 2 μmol/L (Table 2, Fig. 5). The formation of the gut was completely suppressed (Type III) in 5 (83%) of the six animals treated with 75 μmol/L DEAB (Table 2, Fig. 5). Simultaneous treatment with 75 μmol/L DEAB and 2 μmol/L 13-cis RA resulted in a gut (Type I or Type II) in five (63%) of eight animals (Table 2, Fig. 5). These results indicated that RA can rescue the defects caused by citral and DEAB. This, in turn, suggests that endogenous RA is involved in the formation of the gut during the regeneration of P. misakiensis.

Table 2.   Regeneration of the gut in the animals simultaneously treated with inhibitors of retinoic acid (RA) synthesis and 13-cis RA
InhibitorsConcentration of 13-cis RA (μmol/L)Number of casesPhenotypes (%)
Type IType IIType IIIType IV
  1. †Control animals were treated with 0.1% dimethylsulfoxide (DMSO).

Control†03791350
 90  μmol/L Citral0335512330
0.1255624200
0.512583380
1.07435700
2.0175324240
 75  μmol/L DEAB06017830
2.083825380
Figure 5.

 Rescue of the gut’s formation by 13-cis RA from citral- or 4-diethylaminobenzaldehyde (DEAB)-induced hypoplasia. The ratio of animals that formed a gut (Types I and II in Fig. 3) is expressed as a percentage. Control regenerates were treated with 0.1% dimethylsulfoxide (DMSO) (lane 1). Regenerates treated with 90 μmol/L citral and 0, 0.1, 0.5, 1.0 or 2.0 μmol/L 13-cis RA (lanes 2–6, respectively). Regenerates treated with 75 μmol/L DEAB and 0 or 2.0 μmol/L 13-cis RA (lanes 7–8, respectively).

Expression of Pm-RAR and TRAMP during regeneration

Expression of Pm-RAR was examined by in situ hybridization (Fig. 6). The mRNA was expressed in the perivisceral epithelium in the intact adult (Fig. 6A). The perivisceral and atrial epithelia are continuous, comprising a topologically single epithelial sheet (Fig. 6A). The atrial epithelium in the most posterior region weakly expressed the Pm-RAR mRNA (Fig. 6A). The signal was also detected in the branchial epithelium (Fig. 6B). A control sense RNA probe did not give signals other than non-specific staining of the glomerulocytes and tunic (Fig. 6C). It is reasonable that the glomerulocytes showed non-specific signals, because this hemocyte synthesizes components of the tunic (Kimura & Itoh 1995). One day after the amputation of the posterior half, the expression of Pm-RAR was observed in the branchial and atrial epithelia near the cut surface in the anterior half (Fig. 6D–F). Weak signals were also detected in the epidermis, and some mesenchyme cells that formed aggregates near the cut surface (Fig. 6E). The hybridization signal became strong in the perivisceral and atrial epithelia around the gut primordium 2 days after the amputation (Fig. 6G,H). The gut primordium itself was strongly stained (Fig. 6G,H). The perivisceral epithelium at the pharynx level did not express Pm-RAR (Fig. 6G,I). The epidermis also expressed the Pm-RAR mRNA near the gut primordium, but not on the opposite side of the regenerate (Fig. 6G–I). The level of expression of Pm-RAR became weaker on the third day of regeneration (Fig. 6J,K).

Figure 6.

 Expression of the Pm-RAR mRNA during regeneration. The posterior half of adult animals was amputated, and the anterior half was allowed to regenerate. All panels, except for (C), show the expression of Pm-RAR, examined by in situ hybridization using specific antisense probes. A control hybridization with a sense probe is shown in (C). Red arrowheads indicate non-specific staining of glomerulocytes. All panels show frontal sections. The cut surface is oriented to the left or upper left, and the anterior is to the right or lower right. Scale bars indicate 50 μm. ae, atrial epithelium; be, branchial epithelium; bs, branchial sac; en, endostyle; ep, epidermis; gp, gut primordium; mc, mesenchyme cells; pe, perivisceral epithelium; tu, tunic. (A) The most posterior region of the body wall of an intact adult. (B) The branchial sac of an intact adult. (C) The most posterior region of the body wall of an intact adult. (D) A regenerate, one day after the amputation. (E) A high magnification view of the posterior body wall shown in (D). (F) A high magnification view of the regenerating branchial epithelium (be) shown in (D). (G) A regenerate, 2 days after the amputation. The hybridization signals are localized to the gut-forming region. (H) A high magnification view of the boxed region in (G). (I) The most anterior region of a regenerate, 2 days after the amputation. (J) A regenerating gut, 3 days after the amputation. (K) A more dorsal level of the regenerating gut shown in (J).

Expression of TRAMP was examined by in situ hybridization (Fig. 7). The TRAMP mRNA was not detected in the adult (data not shown). A weak signal was detected in the atrial epithelium one day after the amputation (Fig. 7A,B). The branchial epithelium also expressed the TRAMP mRNA (Fig. 7A,C). On the second day of regeneration, the atrial and perivisceral epithelia around the gut primordium strongly expressed the TRAMP mRNA (Fig. 7D,E). The differentiating gut also expressed the TRAMP mRNA (Fig. 7D,E). The level of expression of TRAMP gradually decreased from the third day of regeneration (Fig. 7F). Weak signals were observed only in the perivisceral epithelium and differentiating gut (Fig. 7F).

Figure 7.

 Expression of the TRAMP mRNA during regeneration. The posterior half of adult animals was amputated, and the anterior half was allowed to regenerate. The expression of TRAMP was examined by in situ hybridization using specific antisense probes. Red arrowheads indicate non-specific staining of glomerulocytes. All panels show frontal sections. The cut surface is oriented to the upper left in (A–C), while it is oriented to the bottom in (D–F). Scale bars indicate 50 μm. ae, atrial epithelium; be, branchial epithelium; ep, epidermis; gp, gut primordium; pe, perivisceral epithelium. (A) The posterior region of a regenerate, 1 day after the amputation. (B) The posterior body wall of the regenerate, shown by a box in (A). (C) The branchial epithelium of a regenerate 1 day after the amputation. (D) A regenerate, 2 days after the amputation. The hybridization signals are localized to the gut-forming region. (E) A high magnification view of the boxed region in (D). (F) A regenerating gut, 3 days after the amputation.

Disruption of the function of Pm-RAR and TRAMP by RNA interference

To examine whether Pm-RAR is essential for regeneration of the gut, a mixture of Pm-RAR-specific dsRNA molecules was prepared, corresponding to three different regions of the Pm-RAR mRNA (Supplementary Fig. S1A). After the amputation of the posterior halves, the anterior halves were incubated in 10 μL of seawater containing 0.6 μg/μL of the dsRNA for 1 h. They were allowed to regenerate for 7 days. The regeneration of the gut was examined by X-gal staining. The gut regenerated normally in 33 (97%) of the 34 control animals treated with lacZ-specific dsRNA (Table 3). In contrast, no gut formed in 20 (50%) of the 40 animals treated with Pm-RAR-specific dsRNA (Table 3, Fig. 8A). In our preliminary experiments, three dsRNA molecules were separately applied. These experiments caused a similar phenotype but the effect was weak (data not shown).

Table 3.   Effect of double-stranded RNA (dsRNA) on the regeneration of the gut
dsRNANumber of casesPhenotypes (%)
Type IType IIType IIIType IV
lacZ3497300
Pm-RAR40483500
TRAMP50702280
Figure 8.

 Effect of dsRNA on the regeneration of the gut. (A) The ratio of regenerates that formed a gut (Types I and II in Fig. 3) is expressed as a percentage. Control regenerates were treated with lacZ-specific dsRNA (lane 1). Experimental regenerates were treated with Pm-RAR-specific dsRNA (lane 2) or TRAMP-specific dsRNA (lane 3). (B) Total RNA samples were purified from the regenerates treated with lacZ-specific dsRNA or Pm-RAR-specific dsRNA. The RNA samples were collected from the regenerates 0, 1, 2 or 3 days after the amputation. Reverse transcription–polymerase chain reaction (RT–PCR) was carried out to amplify the cDNA fragments corresponding to β-actin, Pm-RAR or TRAMP. The relative intensity of the RT–PCR signals of Pm-RAR (left) and TRAMP (right) was calculated, using the signal of β-actin as a reference.

The amount of Pm-RAR mRNA was examined by RT–PCR in the regenerating animals treated with Pm-RAR-specific dsRNA. The relative amount was calculated for each sample, using the signal of β-actin as a reference. Control animals, treated with lacZ-specific dsRNA, showed an increase in the relative intensity of RT–PCR signals for Pm-RAR during the first 2 days of regeneration (Fig. 8B). Treatment with Pm-RAR-specific dsRNA resulted in a slowdown of the rise in the relative amount of Pm-RAR mRNA (Fig. 8B). The relative amount of TRAMP mRNA was also estimated in the regenerates treated with Pm-RAR-specific dsRNA. Control animals, treated with lacZ-specific dsRNA, showed an increase in the relative intensity of RT–PCR signals for TRAMP during the first 2 days of regeneration (Fig. 8B). Treatment with Pm-RAR-specific dsRNA resulted in a slowdown of the rise in the relative amount of RT–PCR signals for TRAMP (Fig. 8B).

The regenerating anterior half animals were also treated with TRAMP-specific dsRNA. A mixture of TRAMP-specific dsRNA molecules was prepared, corresponding to three different regions of the TRAMP mRNA (Supplementary Fig. S1B). After the amputation of the posterior halves, the anterior halves were incubated in seawater containing 0.6 μg/μL of the dsRNA for 1 h. No X-gal staining was observed in 14 (28%) of the 50 animals treated with TRAMP-specific dsRNA (Table 3, Fig. 8A).

Discussion

Regeneration of the gut through transdifferentiation of the atrial epithelium

When the posterior half was amputated, the anterior half of the adult P. misakiensis formed an esophagus, stomach and intestine from the atrial epithelium near the cut surface. The atrial epithelium is a differentiated epithelium that expresses specific antigens (Fujiwara & Kawamura 1992; Kawamura & Fujiwara 1994). Most cells in the atrial epithelium are thought to be in the G0 phase of the cell cycle, and enter the cell cycle only when the epithelium forms the organ primordia (Kawamura & Nakauchi 1986). Therefore, the formation of the gut in the regenerates can be regarded as transdifferentiation, just as it is in developing buds (Fujiwara & Kawamura 1992; Kawamura & Fujiwara 1994). This does not exclude the possibility that the differentiated gut epithelium in adults contains specialized stem cell populations that were “determined” to produce the gut epithelial cells only. However, considering that the entire esophagus, stomach and intestine were removed, no such stem cells could have contributed to the regeneration observed in the present study.

Most ascidian species possess undifferentiated multipotent mesenchymal stem cells, called hemoblasts (reviewed in Kawamura et al. 2008). Hemoblasts are found everywhere in the mesenchymal space (hemocoel). Hemoblasts can form an aggregate and differentiate into the atrial or gonadal epithelium in botryllid ascidians (Oka & Watanabe 1957, 1959; Mukai & Watanabe 1976). The gonadal epithelium is also formed from hemoblasts in P. misakiensis (Sunanaga et al. 2007). Hemoblasts in P. misakiensis are integrated into the atrial epithelium during budding (Kawamura et al. 1991). However, the contribution of epithelialized hemoblasts to the formation of organs seems limited. Most of the gut cells in the bud of P. misakiensis are thought to derive from the atrial epithelium, because differentiating gut cells contained traces of atrial epithelium-specific antigens (Fujiwara & Kawamura 1992; Kawamura & Fujiwara 1994). In the present study, we observed aggregates of mesenchyme cells in the hemocoel around the regenerating gut. They expressed Pm-RAR. These cells were relatively small, and seemed to have a large nucleus, which are characteristics of hemoblasts. Although the main source of the regenerating gut is probably the atrial epithelium, mesenchymal stem cells may contribute to the formation of the gut. Further ultrastructural, immunohistochemical and molecular analyses are important for evaluating the contribution of mesenchymal stem cells to the regeneration of the gut.

Regeneration of the gut requires RA

Citral and DEAB suppressed the formation of the gut in regenerating animals, and exogenously administered 13-cis RA rescued the animals from defects in the gut. The Pm-RAR mRNA was expressed around the differentiating gut during regeneration. The Pm-RAR-specific dsRNA suppressed the formation of the gut. In this case, the amount of Pm-RAR mRNA did not increase to the level equivalent to that in control regenerates. The Pm-RAR-specific dsRNA also affected the upregulation of TRAMP mRNA expression during regeneration. These observations suggest that RA is necessary for the gut’s regeneration. The RA signal seems to be mediated at least in part by Pm-RAR and TRAMP.

Retinoic acid is also involved in regeneration in another ascidian, Botrylloides leachi (Rinkevich et al. 2007). In this species, asexually proliferating individuals, called zooids, are all connected with extracorporeal blood vessels (tunic vessels). When all of the zooids were surgically removed from a Botrylloides colony, mesenchyme cells (also called blood cells in ascidians) formed an aggregate within the tunic vessels (Oka & Watanabe 1959; Rinkevich et al. 1995). The aggregate becomes a hollow sphere that gives rise to the atrial epithelium, from which all the other tissues derive (Oka & Watanabe 1959; Rinkevich et al. 1995). This process is called whole body regeneration (WBR; Rinkevich et al. 1995). An RAR-encoding gene is expressed in almost all of the differentiating cells during WBR (Rinkevich et al. 2007). Citral and DEAB inhibited WBR, and RAR-specific siRNA molecules caused malformation, suggesting that RA is required for WBR in B. leachi (Rinkevich et al. 2007). RA may be commonly used for the initiation of regeneration in ascidians.

Role of RA in the regeneration of the gut

In B. leachi, exogenously administered RA induced excessive development of regenerates throughout the tunic vessels (Rinkevich et al. 2007). Therefore, RA is thought to trigger the initiation of an ontogenetic process in the adult body (Katsuyama & Saiga 1998; Rinkevich et al. 2007). In vertebrates, RA is involved in the regeneration of the central nervous system, limbs, alveoli and many other tissues, where it is thought to trigger a re-initiation of the developmental programs that are executed primarily during embryonic development (reviewed in Maden & Hind 2003). In the developing bud of P. misakiensis, RA is thought to trigger the dedifferentiation of the atrial epithelium to re-start the development of the gut (reviewed in Kawamura & Fujiwara 1995). The role of RA during regeneration of the gut in the adult body fragment may be similar. RA may trigger the re-initiation of the developmental program primarily used for the formation of the gut during budding.

In vertebrates, RA affects the differentiation of the intestinal epithelium by modifying the activity of intestinal mesenchyme cells (Plateroti et al. 1997). Hox genes show a co-linear pattern of expression in the gastro-intestinal wall epithelia (Sekimoto et al. 1998). The enhancer region of the murine Hoxb-1 gene contains an RA-response element that is required for activation in the foregut (Huang et al. 1998). A complementary pattern of the expression of RA-synthesizing ALDH (RALDH2) and RA-degrading enzyme (CYP26A1) regulate the left-right asymmetric morphogenesis of the digestive tract (Lipscomb et al. 2006). These observations suggest that RA regulates the antero-posterior regionalization of the digestive tract through the transcriptional regulation of Hox genes. Exogenously administered RA inhibited the morphogenesis of pharyngeal gill slits in ascidian juvenile adults (Hinman & Degnan 1998, 2000). Similarly, RA perturbs the formation of gill slits in amphioxus embryos by expanding the anterior limit of the expression of Hox1 (Schubert et al. 2004). These results can be explained as a posterior (homeotic) transformation of the positional identity of the digestive tract. Thus, the role of RA in the pattern formation of the digestive tract is thought to be common to all of the chordate groups. However, in P. misakiensis, RA induces the formation of a completely organized gut, irrespective of its dose (Hara et al. 1992, and the present study). These results suggest that RA is not involved in the pattern formation but determines the site of regeneration of the gut.

In the present study, accumulation of the TRAMP mRNA was suppressed by treatment of the regenerates with Pm-RAR-specific dsRNA. The result suggests that Pm-RAR activates the expression of TRAMP in the regenerating animals, as is the case with the developing bud (Ohashi et al. 1999). TRAMP is a secreted serine protease that is thought to cause cells of the atrial epithelium to dedifferentiate and undergo proliferation (Ohashi et al. 1999). Since the morphological features of the regenerating gut cells were similar to those of the differentiating gut cells in the bud, TRAMP may induce dedifferentiation of the atrial epithelium during regeneration. During bud development, Pm-RAR and TRAMP were expressed in mesenchyme cells but not in the atrial epithelium (Hisata et al. 1998; Ohashi et al. 1999). In contrast, Pm-RAR and TRAMP were expressed in the atrial epithelium during regeneration. Therefore, the genetic machinery that induces the expression of Pm-RAR and TRAMP in regenerating individuals may be different from that in developing buds. However, in both cases, RA eventually induces the transdifferentiation of the atrial epithelium into the gut. This implies that the regeneration process is driven at least in part (downstream of TRAMP?) by the genetic circuitry used for asexual bud development. Asexual reproduction is a re-activation of the genetic circuitry that drives the ontogeny in the adult body. Therefore, it may be reasonable that colonial ascidian species show stronger regenerative activity than solitary ascidian species.

Acknowledgment

We thank Baku Takahashi, for constructing plasmids used for producing TRAMP-specific dsRNA. We are grateful to Z. Imoto at the Usa Marine Biological Institute of Kochi University for taking care of the culture system of animals. We also thank T. Ishii at Akita University for providing us with animals, and K. Hirayama and N. Satoh for a generous gift of Sun-Culture. We are grateful to T. Sunanaga, H. Kashihara, T. Isozaki, S. Nishioka and other members of our laboratory for valuable discussions. This work was supported in part by the Japan Society for Promotion of Science (Grant number 19570220 to S. F.).

Ancillary