Piwi-expressing hemoblasts serve as germline stem cells during postembryonic germ cell specification in colonial ascidian, Botryllus primigenus


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


Animals that propagate asexually are exciting models to investigate the cellular system, which produces germline cells constitutively throughout life. The present research investigated whether piwi was a germline-specific marker in the colonial ascidian Botryllus primigenus. An approximately 2.8 kb long cDNA fragment was cloned and termed BpPiwi, since the obtained amino acid sequence (874 aa) contained PAZ and PIWI domains. BpPiwi was expressed specifically by germline cells such as the loose cell mass (germline precursor cells), oocytes, spermatogonia, and spermatocytes. In addition, BpPiwi transcripts were also detected in some coelomic cells in the hemocoel and tunic vessels. BpPiwi+ coelomic cells possessed similar morphological features to hemoblasts (stem cells). The concentration of BpPiwi+ cells was found to be significantly lower than that obtained for hemoblasts suggesting that BpPiwi+ cells comprise a fraction of hemoblasts. Further, the ability of BpPiwi+ cells to serve as somatic stem cells was examined. No BpPiwi signals were detected from somatic hemoblasts forming vascular buds. The genetic knockdown of BpPiwi induced by siRNA injection resulted in the formation of a defective germline precursor. These results suggest that BpPiwi+ hemoblasts reside in the hemocoel and tunic vessels and function as germline stem cells in the postembryonic colony. Based on the findings of the characterization of three effective germline genes piwi, vasa, and nanos, we propose that germline stem cells reside as BpPiwi+/BpVas/BpNos+ hemoblasts in B. primigenus.


The separation of germline cells from somatic tissues is a crucial step in the development of all sexually reproducing animals. In the fruit fly, nematode worm, and zebrafish, germline cells are segregated during early embryogenesis by the inheritance of a cytoplasmic determinant. Germ cells are induced via regulatory interaction among embryonic cells in mice and urodele amphibians (Extavour & Akam 2003). Primordial germ cells (PGCs) in most sexually reproductive animals migrate to somatic gonadal tissue where they are preserved as germline stem cells for sexual maturation. However, in primitive metazoans such as hydras and planarians, germ cell specification occurs during adulthood. Thus, in these animals, the division of somatic and germline cells appears to be unclear. Undifferentiated cells such as the interstitial cells in hydras and neoblasts in planarians are the source of germ cells (Shibata et al. 1999; Mochizuki et al. 2001).

Colonial ascidians belong to Urochordata and propagate asexually. In botryllid ascidians, all tissues and organs, including the gonads are reconstructed during the asexual reproductive cycle (Berril 1941; Watanabe 1953; Mukai & Watanabe 1976). This raises the question of whether germline separation from the soma has been established within the postembryonic asexual colony. Recently, several species have been used to examine the origin of germ cells in colonial ascidians. In Botryllus, the gonads and gametes arise from hemoblasts (Mukai & Watanabe 1976; Sabbadin & Zaniolo 1979; Sunanaga et al. 2006); hemoblasts are a type of coelomic cell present in the hemocoel and tunic vessels. In Botryllus primigenus, germ cell formation begins in the gonadal space with loose cell masses (germline precursor cells) comprising hemoblasts (Mukai & Watanabe 1976; Sunanaga et al. 2006). Some cells of the loose cell masses develop into female germ cells, while the remaining form compact cell masses (testicular primordia). We previously showed that the loose cell masses are characterized by the expression of the vasa homologue BpVas. Germline cells can reappear even after BpVas-positive cells are completely extirpated from the colonies (Sunanaga et al. 2006). These observations suggest that in B. primigenus, germline cells are recruited from non-germline cells in the postembryonic stage. In addition, it has been reported that in the asexual reproduction of botryllid ascidians, hemoblasts serve as somatic stem cells (Oka & Watanabe 1957; Kawmaura & Nakauchi 1991). In vascular budding, most somatic organs in a new bud, except the epidermis, originate from an aggregate of hemoblasts that are formed in tunic vessels (Oka & Watanabe 1957). In a related budding ascidian species; namely Polyandrocarpa misakiensis, we have shown that germ cells are derived from hemoblast-like coelomic cells (Sunanaga et al. 2007). The asexually generated buds do not inherit vasa-expressing germline cells from their parents, and, therefore, it is probable that hemoblasts are reserved as totipotent stem cells in adulthood (Sunanaga et al. 2007, 2008). In contrast to this hypothesis, genetic and cell transplantation experiments in Botryllus schlosseri suggest that germline and somatic stem cells are separate lineages of cells (Laird et al. 2005). The candidate for germline stem cells was proposed in B. schlosseri and Botrylloides violaceus. It was characterized as a vasa-expressing cell that was found to be circulating in tunic vessels and located around the gonads (Brown & Swalla 2007; Brown et al. 2009; Rosner et al. 2009). These findings indicate that colonial ascidians have remarkable systems, which allow for assessment of postembryonic germ cell specification in metazoans. In order to further elucidate the cellular system structure that produce germline cells in colonial ascidians, further molecular characterization of the germline cells and stem cells (hemoblasts) residing in the colonies is warranted.

Argonaute proteins are characterized by two major motifs the Piwi/Argonaute/Zwille (PAZ) domain and PIWI domain. They represent a large protein family whose members are found in various organisms (Tolia & Joshua-Tor 2007). The Piwi subfamily proteins are expressed in germline cells, and are considered as a key regulator of gene expression at the post-transcriptional level (Lin 2007). Recently, interaction between the Piwi proteins and a class of small RNAs (piRNAs) has been discovered in mammalian germline cells and Drosophila melanogaster (Vagin et al. 2006). Piwi is required for the self-renewal and maintenance of both female and male germline stem cells in D. melanogaster (Cox et al. 1998). In mice, Miwi and Mili are essential for spermatogenesis (Deng & Lin 2002; Kuramochi-Miyagawa et al. 2004). It has also been reported that homologues of piwi are expressed specifically by germline cells in zebrafish as Ziwi, in humans as Hiwi, and in silkworm as SIWI (Qiao et al. 2002; Tan et al. 2002; Kawaoka et al. 2008). Piwi-like genes have also been isolated in primitive metazoans. Planarian piwi homologues smedwi-1, -2, and -3 (or DjPiwiA, B, and C) are expressed in neoblasts that serve as adult somatic stem cells (Reddien et al. 2005; Palakodeti et al. 2008; Hayashi et al. 2010; Shibata et al. 2010). The piwi homologue Cniwi in a Cnidarian is expressed not only in germline stem cells but also in somatic cells involved in transdifferentiation and regeneration (Seipel et al. 2004). In addition to them, EfPiwiA and EfPiwiB were isolated in a freshwater sponge, Ephydatia fluviatilis. EfPiwiA-expressing archeocytes were suggested to be multipotent stem cells (Funayama 2010). Taken together, these results indicate that the Piwi subfamily proteins play a crucial role in germline cells and/or stem cells in a wide range of animals.

In the present study, we describe the structure and expression of the Botryllus piwi homologous gene, i.e. BpPiwi. The germline-specific expression of BpPiwi mRNA revealed that BpPiwi is a reliable marker of germline cells in B. primigenus. Interestingly, a number of coelomic cells expressed BpPiwi. Next, to investigate whether BpPiwi-expressing coelomic cells serve as a source of germline cells in the postembryonic colony, the genetic knockdown of BpPiwi was carried out using short interfering RNA (siRNA). Results indicate that a lack of BpPiwi mRNA suppressed the appearance of the germline precursor cells. Based on these data and the expression patterns of other germline genes, we propose a candidate for germline stem cells in B. primigenus.

Materials and methods


Colonies of B. primigenus were collected in the vicinity of the Usa Marine Biological Institute of Kochi University (Kochi Prefecture, Japan). They were allowed to grow on glass plates in culture boxes settled in the Uranouchi Inlet near the institute.

cDNA cloning of the BpPiwi gene

The piwi gene fragments were amplified by polymerase chain reaction (PCR) from a cDNA pool of whole colonies. The following degenerate primers were used: 5′-G(TG)CC(TCAG)GA(TCAG)TGGTA(TC)GA(TC)TT(TC)TT-3′, 5′-GT(TCAG)CC(TCAG)GGCCA(AG)TT(AG)TA(AG)TA-3′, 5′-AA(TC)TG(TC)AA(AG)(TCA)T(TCAG)GG(TCAG)GG(TC-AG)GA-3′, 5′-G(TG)GC(TG)GA(TC)TT(TC)(CA)A(TCAG)AC(TCAG)ATGAA-3′, 5′-TATAATAA(TC)AAGAC(TCAG)TA(TC)(CA)G(TCAG)(AG)T(TCAG)GA-3′, and 5′-TGG-TGTTT(TG)TA(TC)CA(AG)TA(TC)(TCA)(CG)(TCAG)GT-3′. Degenerate PCR was performed with the following conditions: 94°C for 30 s, 45–55°C for 30 s, 72°C for 1 min (annealing temperature was increased 5°C each 10 cycles, total 30 cycles). According to previous studies, the 5′ and 3′ ends of the cDNA were elongated (Sunanaga et al. 2006, 2007). There were five kinds of primers for 5′ rapid amplification of the cDNA ends (RACE): GSP1, 5′-TTCCTCTTCTCGCTCCTCTG-3′; GSP2, 5′-CTGGTATAGGCGCCATGATT-3′; GSP3, 5′-AGTTGGAAACCAATGACACGG-3′; GSP4, 5′-TGCA-GTACCTGGATCCCTTG-3′; and GSP5, 5′-CACCTGT-CCTCAAAGCACCA-3′. The two primers specific for 3′-RACE were as follows: GSP6, 5′-CACGGTTTCTCCCACTCACT-3′ and GSP7, 5′-AGCTTGCCTTCAAGCTGTGT-3′. As described by Sunanaga et al. (2006), two kinds of adapter primers and adapter-dT17 primers were designed. The PCR products were subcloned into a TA cloning vector. The nucleotide sequences of cDNA inserts were determined with the ABI PRISM3100-Avant genetic analyzer system (Applied Biosystems). For cycle sequence reaction, the BigDye terminator v3.1 cycle sequencing kit (Applied Biosystems) was used.

Whole-mount in situhybridization

Specimens were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C for 10 h. The fixed specimens were rinsed in PBS containing 0.1% Tween20 (PBST), and dehydrated in a graded series of methanol. The in situ hybridization was carried out as described by Sunanaga et al. (2006). Specimens were embedded in JB-4 plastic resin (Polyscience Inc., Warrington, PA, USA), and serially sectioned at a thickness of 2 μm.


Specimens were fixed in Zamboni’s fixative (Zamboni & DeMartino 1967) at 4°C for 30 min. After dehydration, samples were embedded in Technovit 8100 resin (Heraeus Kulzer GmbH, Hanau, Germany) and sectioned at 2 μm. All sections were mounted serially on cover slips. Then, sections were treated with the blocking solution (3% normal horse serum in PBS) for 30 min, and reacted with anti-BpVas monoclonal antibody at room temperature for 60 min. They were washed twice with PBST (10 min/wash), and stained with anti-mouse secondary antibody labeled with horseradish peroxidase (HRP). Specimens were colored by Trueblue (KPL).

Reverse transcriptase-PCR

Total RNA was extracted from the B. primigenus colonies by using the AGPC method (Chomczynski & Sacchi 1987). The following primers were used for amplification of BpPiwi. GSP8, 5′-ATGGCTG-AGCAAAAAGGAACTC-3′ and GSP2 (mentioned above). To detect GAPDH, GAPDH-F and GAPDH-R primers were used according to Sunanaga et al. (2008). RT–PCR was carried out with the following conditions: 94°C for 30 s, 55°C for 30 s, 72°C for 45 s. The number of cycle ranged from 15 to 35.

RNA interference

Three different siRNAs were designed from the BpPiwi mRNA and were obtained from SIGMA-PROLIGO. Their oligonucleotide sequences were as follows: siRNA-1, 5′-GACCGGAGCAUUUAUUAAACA-3′ and 5′-UUUAAUAAAUGCUCCGGUCUU-3′, siRNA-2; 5′-CACGUACCGUGUUGAUGAUAU-3′ and 5′-AUCAUC-AACACGGUACGUGUU-3′; siRNA-3, 5′-GCUCUACUUCCUGUAGUAUUG-3′ and 5′-AUACUACAGGAAGU-AGAGCUU-3′. The three siRNA oligonucleotides were mixed (final concentration of 40 μM each) in filtered seawater. As a control, siRNA corresponding to lacZ was prepared at a final concentration of 120 μM. In order to knock down the BpPiwi gene, 4 μL of the siRNA mixture was microinjected into an experimental B. primigenus colony through the peripheral ampullae. Following treatment, specimens were cultured in the sea or in the aquarium.


Zooids and buds were extirpated from colonies by using razor blades. The remaining peripheral common vascular system was allowed to form vascular buds (Sunanaga et al. 2006).


Isolation and characterization of the B. primigenus Piwi homologue

We isolated PCR fragments that were approximately 2.1 kb in length from the B. primigenus cDNA pool by using degenerate primers designed from the conserved regions on the PAZ and PIWI domains of various animals. In order to obtain longer cDNA fragments, 5′- and 3′-RACE was carried out. The total length of the determined sequence was approximately 2.8 kb (GenBank accession no. AB455103), and it encoded 874 amino acid residues followed by a 3′-untranslated region that was approximately 170 bp in length (Fig. 1a). A similarity search using the GenBank database revealed that the deduced polypeptide was most closely related to the zebrafish piwi homologous protein Ziwi; consequently, we termed the polypeptide ‘BpPiwi’. A phylogenetic analysis of the full-length amino acid sequence by the neighbour-joining method indicated that BpPiwi belonged to the PIWI subfamily (Fig. 1b). The predicted amino acid sequence of BpPiwi contained the PAZ domain (285–423 aa) and Piwi domain (567–860 aa). The PAZ domain is known to recognize the single-stranded 3′ end of small RNAs (Bernstein et al. 2001). The Piwi domain, which is structurally similar to RNaseH catalytic domain, contains the highly conserved motif Asp-Asp-His (D-D-H) that is required for endonucleolytic cleavage (Tolia & Joshua-Tor 2007). The D-D-H motif was observed in the Piwi domain of BpPiwi (Fig. 1a). This suggests that BpPiwi has functional activities that are conserved among the Argonaute protein family members. Using the expressed sequence tags (EST) databases of other ascidian species, i.e. Ciona intestinalis and Halocynthia roretzi, the cDNA of Piwi subfamily genes in both species was identified. These findings imply that a gene control system mediated by small non-coding RNAs functions in ascidians.

Figure 1.

 Characterization of BpPiwi. (a) Deduced amino acid sequence and structure of BpPiwi. BpPiwi has the Piwi/Argonaute/Zwille (PAZ) domain (green) and the PIWI domain (red). The amino acid residues comprising the catalytic D-D-H motif are shown in boxes. (b) Phylogenetic tree of BpPiwi (in dotted box) and several Argonaute family proteins in other organisms. Their accession numbers are given in Table S1. The tree was drawn by the Neighbor-joining method. The bootstrap values are shown on each branch.

Expression of BpPiwiin the process of germ cell formation

The expression of BpPiwi mRNA was examined by whole-mount in situ hybridization (WISH). In B. primigenus, germ cell formation begins in the gonadal space with a loose cell mass that consists of hemoblasts. The loose cell mass and developing germ cells specifically expressed BpPiwi, but no expression was observed in somatic tissues and organs such as the epidermis, pharynx, atrial epithelium, and digestive tract (Fig. 2a). Every cell in the loose cell mass emitted a strong signal (Fig. 2b). In the developing testis, spermatogonia and spermatocytes were enveloped by testicular epithelium. It is important to note that only the germline cells were stained (Fig. 2a,b), and this signal became weaker as the testis developed. The oocytes expressed BpPiwi mRNA, while the accessory cells surrounding them did not (Fig. 2c). The signal became weaker at the vitellogenesis stage (Fig. 2c). Finally, no signal was detected in the developed gonad (Fig. 2d). In B. primigenus, somatic tissues failed to express BpPiwi. The loose cell mass (germline precursor) differentiates into not only germ cells but also the somatic parts of the gonad. BpPiwi expression disappeared permanently from those somatic cells. These results indicate that expression of BpPiwi mRNA is a reliable marker for germline cells in B. primigenus.

Figure 2.

 Expression of BpPiwi mRNA. (a) A developing bud. It should be noted that the signal was detected only in the germline cells such as the loose cell mass (arrowhead), juvenile oocyte (arrow), developing oocyte (oc), and developing testis (te). Bar, 100 μm. ae, atrial epithelium; cc, coelomic cell; dt, digestive tract; ep, epidermis; ph, pharynx. (b) Loose cell mass in gonadal space of developing bud emitted a strong signal (arrowheads). Bar, 30 μm. (c) Juvenile oocytes were stained (arrowheads). The signal became weak in developing oocyte (oc) enveloped by accessory cells. Accessory cell layer is outlined with dotted line. Bar, 30 μm. ac, accessory cell. (d) A well-developed gonad. No signal was detected from the ovary (ov) and testis. Bar, 100 μm. (e) There were BpPiwi-expressing coelomic cells (arrowheads) in the hemocoel of the pharynx. Pharyngeal walls are traced by dotted lines. Bar, 30 μm. (f) There were BpPiwi-expressing coelomic cells (arrowheads) in the hemocoel close to the endostyle (en). Endostyle is outlined with solid line. Bar, 30 μm. (g) There were BpPiwi-expressing coelomic cells (arrowheads) close to the digestive tract (dt). Wall of the digestive tract is outlined with solid lines. Bar, 30 μm. (h) There were BpPiwi-expressing coelomic cells (arrowheads) in the tunic vessel (tv). Dotted lines indicate epidermis of the tunic vessel. Bar, 20 μm. (i) Part of BpPiwi-negative coelomic cells (white arrowheads) and BpPiwi-expressing cells (black arrowheads) showed similar morphological features. Bar, 20 μm.

Piwi-expressing coelomic cells in the hemocoel and tunic vessels

Notably, some coelomic cells expressed BpPiwi in the hemocoel and tunic vessels. BpPiwi-expressing cells in the hemocoel close to the pharynx, endostyle, and intestinal tube were observed (Fig. 2e–g). Similarly, BpPiwi-expressing cells were also observed in the tunic vessels (Fig. 2h). Examination of these cells under a light microscope revealed that these cells were circular and lacked the prominent organelles; hemoblasts exhibit these morphological features (Hirose et al. 2003; Sunanaga et al. 2006). In an attempt to calculate the amount of BpPiwi-expressing coelomic cells, three independent counts showed that 9.75% of coelomic cells emitted the BpPiwi signal (Table 1). Consistent with the result obtained by Taneda & Watanabe (1982), in B. primigenus, hemoblasts comprise approximately 30% of the coelomic cells (Table 1). This implies that BpPiwi-expressing coelomic cells comprise a fraction of hemoblasts. In addition, BpPiwi-negative coelomic cells with identical morphology to hemoblasts were often observed (Fig. 2i). To examine whether the expression of BpPiwi is detected specially in sexual colonies, RT–PCR was carried out using mRNA extracted from both asexual and sexual colonies of B. primigenus. Results indicate that both colonies expressed BpPiwi mRNA (Fig. 3).

Table 1.   Concentrations of stem cells and BpPiwi-expressing cells
Stem cells (Taneda & Watanabe 1982)34.5%
Hemoblasts in tunic vessels (This study)29.3 ± 0.47 %
BpPiwi-expressing cells in tunic vessels (This study)9.75 ± 0.65 %
Figure 3.

 Reverse transcription–polymerase chain reaction (RT–PCR) analysis of BpPiwi in the sexual and asexual colonies. Expression level of BpPiwi mRNA in asexual colonies was weaker than that in sexual colonies.

Piwi-expressing cells in juvenile colonies

In order to delineate the embryonic origin of BpPiwi-expressing cells, WISH was performed on the most juvenile colonies. They contained one oozooid generated from a tadpole larva via metamorphosis, first bud, and tunic vessels (Fig. 4a). To assess the efficiency of WISH on juvenile colonies, the riboprobe against trypsin mRNA was prepared. The developing intestinal tube emitted the signal suggesting that the technique adopted in the analysis was active (Fig. 4b). No BpPiwi-expressing cell was detected in the hemocoel and tunic vessel of the juvenile colonies (Fig. 4c,d).

Figure 4.

 Expression of BpPiwi mRNA on juvenile colony. (a) A juvenile colony. It contained an oozooid, the first bud, and tunic vessels. The end of the vessel is ampullae (am). Bar, 500 μm. (b) In situ hybridization using trypsin probe. The digestive tract (dt) of a bud expressed trypsin mRNA (allowheads). Bar, 10 μm. (c), (d) In situ hybridization using BpPiwi probe. (c) Coelomic cells in hemocoel of endostyle (en) did not emit any signals. Endostyle is outlined with dotted line. (d) Coelomic cells in tunic vessel (tv) did not emit any signals. Bar in (c) indicates 10 μm; and is adopted in (d).

Expression of BpPiwi in the process of vascular budding

It is apparent that hemoblasts serve as somatic stem cells in the vascular budding of B. primigenus (Oka & Watanabe 1957; Sunanaga et al. 2006). In order to examine whether BpPiwi is involved in the budding process, the expression of BpPiwi mRNA was determined by WISH. Vascular budding was initiated by the gathering of hemoblasts under the epidermis of the tunic vessel (Fig. 5a’). The hemoblasts forming an aggregate are homogenous and similar to circulating hemoblasts in their fine structure (Kawamura & Sunanaga 2010). BpPiwi was not expressed by the aggregating hemoblasts (Fig. 5a). Interestingly, a cavity appeared in the center of the aggregate (Fig. 5b’); the size of the cavity increased, resulting in the bud consisting of two types of cell layers. The cell layers included an outer epidermis layer that originated from the tunic vessels and an inner multipotent epithelium showing a blastula-like structure formed from the aggregate (Fig. 5c’). BpPiwi expression in the buds during their development was not detected (Fig. 5b,c). The evagination and invagination of the inner epithelium resulted in the formation of the rudiments of somatic organs such as the pharynx, digestive tract, and endostyle (Berril 1941; Watanabe 1953).

Figure 5.

 Expression of BpPiwi mRNA in vascular budding. (a’)–(c’) Sections were stained with toluidine blue. (a)–(c) In situ hybridization using BpPiwi probe. (a’), (a) Aggregating hemoblasts (arrow heads) under the tunic vessel (tv). They did not emit any BpPiwi signals. (b’), (b) A vascular bud (arrowhead) with a small cavity (arrow). It was enveloped by the tunic vessel. (c’), (c) A vascular bud at double-layered stage. Arrowhead indicates the inner multipotent epithelium. Bar in (a’) indicates 50 μm; and is adopted in all the panels.

Genetic knockdown of BpPiwi by siRNA injection

In order to examine whether BpPiwi-expressing coelomic cells function as a cellular source of germline cells, the genetic knockdown of BpPiwi was induced using siRNAs. Three types of siRNAs targeting BpPiwi mRNA were mixed, and an siRNA targeting lacZ was used as the control. In B. primigenus colony, three successive generations, a feeding zooid of the first generation, primary buds of second generation, and secondary buds of third generation form a unit (Fig. 6a) (Watanabe 1953). This unit represents four different asexual reproductive phases (A–D) by combinations of the developmental stages of the three generations (Watanabe 1953). For this knockdown analysis, the colonies from which germline cells were eliminated by the vascularization technique were prepared according to our previous studies (Fig. 6b) (Sunanaga et al. 2006, 2008). The siRNA(BpPiwi) was injected into the regenerating colonies 3 days after vascularization. Results indicate that new buds appeared by vascular budding within these colonies (Fig. 6c’,d’). Immunohistochemistry using anti-BpVas antibody showed that no germline cells resided in them (Fig. 6c,d). To assess the effect of siRNA on BpPiwi transcripts, WISH was conducted 10 days after the injection. The siRNA(BpPiwi)-treated colonies did not show BpPiwi signal (Fig. 6e,f). On the other hand, BpPiwi-expressing coelomic cells were observed in the hemocoel and tunic vessels of siRNA(lacZ)-treated colonies (Fig. 6h,i). These results suggest that siRNA(BpPiwi) treatment successfully reduced BpPiwi mRNA to undetectable level with WISH. Immunohistochemical analysis showed that siRNA(BpPiwi)-treated colonies did not have Vasa-positive cells (Fig. 6g). These observations indicate that the loose cell mass (germline precursor) was not generated in the siRNA(BpPiwi)-treated colonies (Table 2). In contrast with siRNA(BpPiwi), siRNA(lacZ)-treated specimens contained the loose cell mass and developing oocytes (Fig. 6j). Somatic tissues and organs developed normally in all specimens treated with both siRNAs (Fig. 6e–j). These results indicate that loss of BpPiwi function led to defective germline precursor formation. In additional studies, BpPiwi knockdown was carried out on the sexual colonies. We did not detect any abnormality in the development of germline and gonadal cells (Fig. S1).

Figure 6.

 Effects of BpPiwi siRNA on germline precursor formation. (a) A colony used in the knockout analysis. Botryllus colonies contain three asexual generations: feeding zooid primary bud (pb), and secondary bud. Because the secondary buds are too small, they are invisible on the panel. The terminal of tunic vessel forms ampule-like structure (ampullae; am) on the edge of colony. Dotted line outlines the colony. (b) A specimen immediately after vascularization. All zooids and buds were removed from colony. siRNA solution was injected into the specimen through the tip of ampullae (arrowheads). Bar in (a) indicates 1 mm; and is adopted in (b). (c), (d) Immunohistostaining of regenerating colony with anti-BpVas monoclonal antibody. BpVas-expressing cells were not detected in 3-day regenerating colonies. (c’), (d’) Counter stained section of (c), (d). Arrowheads indicate new zooids generated by vascular budding. Bar in (c’) indicates 50 μm; and is adopted in (c), (d), (d’). (e), (f), (h), (i) In situ hybridization on siRNA-treated specimens using BpPiwi probe. (g), (j) Immunohistostaining of siRNA-treated specimens with anti-BpVas monoclonal antibody. (e), (f) BpPiwi mRNA was not detected from coelomic cells in hemocoel close to the digestive tract (dt) and tunic vessels in siRNA(BpPiwi)-treated specimen. Wall of the digestive tract is outlined with dotted lines. Bar in (e) indicates 10 μm; and is adopted in (f), (h), (i). (g) Germline precursor cells did not reappear in siRNA(BpPiwi)-treated specimen. A primary bud (pb) and a secondary bud (sb) did not have germline precursor cells. Bar, 50 μm. (h), (i) Coelomic cells in hemocoel close to the endostyle (en) and tunic vessels emitted signals distinctly. (j) Germline precursor cells reappeared in siRNA(lacZ)-treated specimen (arrowheads). White arrowhead shows the cluster of juvenile oocytes. Bar, 20 μm.

Table 2.   Appearance of loose cell mass in siRNA-treated colonies
Sample IDLoose cell massSample IDLoose cell mass


Animals that propagate asexually reconstruct their gonads and germline cells in every asexual generation. In the present study, in order to delineate the cellular system that produces germline cells in B. primigenus, we tracked the germline cells using the molecular marker BpPiwi.

Piwi-expressing hemoblasts produce germline cells in postembryonic colonies

In B. primigenus, it has been reported that the vasa homologue BpVas is a useful marker for germline cells. In previous studies, BpVas-expressing coelomic cells in the hemocoel and tunic vessels except for migrating oogonia and/or oocytes were not observed (Sunanaga et al. 2006, 2008). Therefore, the type of cell recruited to the germline remains to be determined. BpPiwi was expressed by coelomic cells in the hemocoel and tunic vessels as well as germline cells (Fig. 2). In vascular budding, the aggregating hemoblasts did not contain any BpPiwi-expressing cells (Fig. 5), which strongly suggests that hemoblasts serving as somatic stem cells are separated from BpPiwi-expressing cells. Further, the developmental potency of BpPiwi-expressing cells appears to be limited to germline and gonadal cells. It suggests that germline cells are already established prior to the formation of the loose cell mass (germline precursor). Treatment with siRNA-targeting BpPiwi caused defective germline precursor formation (Fig. 6 and Table 2) suggesting that the BpPiwi-expressing hemoblasts function as germline producing cells. It is likely that they gather and differentiate into loose cell masses. Currently, the molecular function of BpPiwi remains unclear in postembryonic germline formation in B. primigenus. The disruption of BpPiwi transcripts did not induce any abnormality in the development of germ cells (Fig. S1). Therefore, it is possible that the role of BpPiwi in germline precursor formation is different from the role in germ cell development.

Molecular characterization of germline stem cells in B. primigenus

In addition to BpPiwi, it is well established that BpVas and nanos homologue (BpNos) are useful marker genes for germline (Sunanaga et al. 2006, 2008; Brown & Swalla 2007; Brown et al. 2009; Rosner et al. 2009). Their expression patterns in adulthood were examined in B. primigenus and are summarized in Figure 7. The loose cell mass was observed to express all of the marker genes. BpPiwi and BpVas transcripts were readily detected by in situ hybridization, and BpNos was weakly expressed in the loose cell mass. In previous studies, we reported the presence of BpNos-expressing hemoblasts in the hemocoel and tunic vessels (Sunanaga et al. 2008). BpNos is expressed in germline cells, especially in male germ cells. The genetic knockdown of BpNos induced disappearance of loose cell mass and a delay in germ cell regeneration (Sunanaga et al. 2008). In the present study, BpPiwi was expressed specifically in germline cells (Fig. 2a–d). In addition, BpPiwi-expressing coelomic cells were observed (Fig. 2e–i). Both appeared to share their morphological features with BpNos-expressing coelomic cells. Further, BpPiwi- and BpNos-expressing cells showed similar localization in the hemocoel and tunic vessels (Fig. 2e–i) (Sunanaga et al. 2008). These data led to a germline-producing cells hypothesis: BpPiwi+/BpVas/BpNos+ hemoblasts reside in the hemocoel and tunic vessels and serve as germline stem cells. RT–PCR analysis showed that BpPiwi transcripts were detected in asexual colonies, as well as in sexual colonies (Fig. 3). It suggests that BpPiwi-expressing hemoblasts are preserved during the asexual period of colonies in B. primigenus. In B. violaceus and B. schlosseri, vasa-positive circulating cells have been proposed to be the source of germline cells within the colony (Brown & Swalla 2007; Brown et al. 2009; Rosner et al. 2009). In a previous study, BpVas-positive circulating hemoblasts in B. primigenus was not observed (Sunanaga et al. 2006). These inconsistent results between closely related species can be explained by a highly sensitive analysis, which uses the fluorescence in situ hybridization method that may be effective in B. primigenus. However, we cannot rule out the possibility that their cellular systems, which produce germline cells during postembryonic stage are different from each other. Characterization of germline stem cells using molecular markers have been carried out in many animals containing non-model organisms like colonial ascidians. For example, an enchytraeid worm, Enchytraeus japonensis, is thought to reserve germline stem cells as Ej-piwi-positive cells in adulthood (Tadokoro et al. 2006). In addition to Ej-piwi, the germline stem cells express Ej-vasa-like genes; however, Ej-vasa-like genes are also expressed in somatic cells (Sugio et al. 2008; Yoshida-Noro and Tochinai 2010). Cnidarian germline stem cells express vasa-related genes and nanos-related genes in Hydra magnipapillata (Mochizuki et al. 2000, 2001). However, the Podocorye carnea piwi homologue Cniwi is both a germline and somatic stem cell gene in jellyfish (Seipel et al. 2004). Together these observations indicate that optimal marker genes for germline cells vary according to the animal species and that accurate understanding of a cellular profile requires multiple indicator molecules. To our knowledge, this is the first study to explore germline stem cells using reasonable multiple markers in colonial ascidians.

Figure 7.

 Schematic representation of probable germline producing system in postembryonic stage in B. primi-genus. Germline stem cells are thought to be present as BpPiwi+/BpVas/BpNos+ hemoblasts in a colony.

Possible origins of germline lineage in colonial ascidians

On the basis of the above findings, we propose that BpPiwi+/BpVas/BpNos+ hemoblasts function as germline stem cells in B. primigenus adult colonies. To explore their ancestral cells, WISH targeting BpPiwi mRNA was conducted on juvenile colonies. Results show that the serial sections did not emit any signal (Fig. 4). The findings imply that the establishment of germline stem cells is subsequent to the asexual development of the colonies. Recent studies have reported that in B. schlosseri, the germline precursors were vasa-positive cells and segregated from somatic lineage during embryonic development, and that this contributed to gametogenesis throughout the life of the colony (Brown et al. 2009). We never deny the pathway in which adult colonies inherit germline-committed cells from their embryonic stages in B. primigenus. The present data highlight the potential to regenerate the germline cells in ascidian species. In the solitary ascidian C. intestinalis, primordial germ cells specified early in embryogenesis were distributed in the larval tail. Thereafter, they migrated into the primordial gonad of the juvenile zooid and differentiated into germ cells (Fujimura & Takamura 2000; Shirae-Kurabayashi et al. 2006). However the germ cells appeared to form in the future zooid even though the larvae were allowed to metamorphose after the larval tail was cut off (Takamura et al. 2002). At present, the underlying mechanism responsible for the formation of regenerative germ cells remains to be fully determined. Piwi may be a key gene to solving this question. Together, these findings suggest that the flexible cellular systems developed in ascidian species, resulting in them being able to specify germline cells constitutively even if the embryonic germline cells are disrupted. Finally, the stem cell system of colonial ascidians delineated from these studies and others does not exclude the possibility of the presence of totipotent stem cells in the colony (Fig. 7). In order to fully understand the cellular system that supplies germline and somatic cells in colonial ascidians, the identification of additional markers, especially in the somatic stem cell genes, is necessary. In vivo and in vitro clonal analyses using a clonal and genetically labeled cell line derived from hemoblasts are also required.


We express our special thanks to the staff of the Usa Marine Biological Institute of Kochi University, Japan, for providing facilities for the culturing of animals. This work was partly supported by KAKENHI on Innovative Areas, ‘Regulatory Mechanism of Gamete Stem Cells’ (21 116 507) to TS, and by KAKENHI 19 570 208 to KK.