Postembryonic epigenesis of Vasa-positive germ cells from aggregated hemoblasts in the colonial ascidian, Botryllus primigenus

Authors


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

Abstract

We investigated whether Vasa was a germline-specific marker in the colonial ascidian Botryllus primigenus, and whether it was inducible epigenetically in the adult life span. We cloned a Botryllus Vasa homologue (BpVas). The deduced open reading frame encoded 687 amino acid residues. It was expressed specifically by germline cells such as the loose cell mass, oogonia and juvenile oocytes in the ovary, and the primordial testis (compact cell mass), spermatogonia and juvenile spermatocytes in the testis. The loose cell mass, the most primitive germline cells, showed an ultrastructure of undifferentiated cells known as hemoblasts. The hemoblasts did not contain electron-dense materials or a mitochondrial assembly in the cytoplasm. These organelles appeared later in the oogonia and oocytes. When the loose cell mass and developing germ cells were eliminated by extirpating all zooids and buds from the colonies, BpVas transcripts disappeared completely from the vascularized colonies. After 14 days, when the colonies regenerated by vascular budding, BpVas-positive cells reappeared in some cases, and in 30 day colonies, BpVas-positive germ cells were observed in all the regenerated colonies. These results show that in B. primigenus, germ cells are inducible de novo from the Vasa-negative cells even at postembryonic stages.

Introduction

Germ cells are specialized cells that transfer genomic information from one generation to the next. Numerous studies have focused on the cellular and molecular mechanisms underlying the segregation of germ cells from somatic cells. In some cases, germ cell determination occurs early in embryogenesis due to localized maternal determinants, while in other cases it is based on the regulative interaction between embryonic cells. The former is termed ‘preformation’ and occurs, for example, in nematodes, fruit flies, anuran amphibians and chickens, while the latter is termed ‘epigenesis’ and occurs in urodele amphibians and mammals (Matova & Cooley 2001; Extavour & Akam 2003; Matsui & Okamura 2005).

The Vasa gene encodes an ATP-dependent RNA helicase of the DEAD box family of proteins, and the helicase is specifically expressed in the germ cell lineage of vertebrates (Fujiwara et al. 1994; Komiya et al. 1994; Olsen et al. 1997; Yoon et al. 1997; Castrillon et al. 2000; Tsunekawa et al. 2000), ascidians (Fujimura & Takamura 2000), insects (Nakao 1999; Chang et al. 2002), brachiopods (Sagawa et al. 2005), oyster (Fabioux et al. 2004), nematodes (Roussell & Bennett 1993; Gruidl et al. 1996; Kuznicki et al. 2000), planaria (Shibata et al. 1999; Mochizuki et al. 2001), cnidarians (Mochizuki et al. 2001; Extavour et al. 2005) and sponges (Mochizuki et al. 2001). In Drosophila, Vasa expression is essential for polar granule formation and germ cell development (Hay et al. 1988a,b; Lasko & Ashburner 1988; Liang et al. 1994). Therefore, it is generally accepted that the gene product of Vasa is one of the most reliable molecular markers for germline cells.

In ascidian embryos, the maternal mRNA of Vasa homologues is segregated into the posterior-most blastomeres during the early cleavage stages and is finally localized on the blastomeres known as B7.6 (Fujimura & Takamura 2000). The B7.6 cells contain an organelle known as the centrosome-attracting body (CAB; Hibino et al. 1998; Nishikata et al. 1999). The ultrastructure of CAB (Iseto & Nishida 1999) resembles the germplasm in Drosophila melanogaster (Mahowald 1962), Caenorhabditis elegans (Wolf et al. 1983) and Xenopus laevis (Czolowska 1969).

In contrast with solitary ascidians such as Ciona intestinalis, colonial ascidians reconstruct gonads during every asexual reproduction cycle. In Botryllus primigenus, the gonads and gametes originate from cell aggregates of undifferentiated cells known as hemoblasts (Mukai & Watanabe 1976; Mukai 1977). Hemoblasts are a type of mesenchymal cells that are present in the coelomic space of the body and tunic vessels. They aggregate to form a loose cell mass in the gonadal space located between the epidermis and the atrial epithelium (Fig. 1). A part of the loose cell mass generates oogonia and primary follicular cells (PFC) that surround an oogonium, and the remaining part develops a compact cell aggregate that is the primordial testis and ovary (Fig. 1). The ovary contains the oviduct (follicle stalk) and the secondary follicle cells (outer follicles; Mukai & Watanabe 1976). These observations have suggested that in B. primigenus, germ cells may continue to be generated from multipotent cells via the regulative interaction at the postembryonic stages. However, it is possible to assume that in B. primigenus, germline hemoblasts have already been established early during embryogenesis, similar to the process observed in solitary ascidians, and these are maintained within the colony for an extended period. In order to solve this problem, we need reliable molecular markers that can trace the germline cells of colonial ascidians.

Figure 1.

Diagram of gonadal and germ cell development in Botryllus primigenus. Dorsal view of a developing bud. Hermaphrodite gonads are formed in the gonadal space located between the atrial epithelium and epidermis. Primordial and mature gonads are shown on the right and left sides of the body, respectively. The primordial gonad appears as loose cell mass in the gonadal space. A part of them differentiates into the ovum, and the remaining part forms the testis and somatic parts of the ovary (Mukai & Watanabe 1976). a.e, atrial epithelium; a.f, atrial fold; c.c.m, compact cell mass; ep, epidermis; g.r, gut rudiment; l.c.m, loose cell mass; n.c., neural complex; oc, oocyte; og, oogonium; te, testis; ov, ovum; t.v, tunic vessel.

In this study, we isolated a B. primigenus Vasa homologue (BpVas) and examined the spatiotemporal expression of BpVas mRNA. BpVas was observed to be strongly expressed in loose cell masses. On the basis of the ultrastructure, we characterized the loose cell mass and oocytes with special reference to electron-dense materials and the mitochondria. Next, we investigated whether or not BpVas could be expressed de novo in BpVas-negative colonies by the vascularization technique. Our results are the first to indicate that germline cells can be induced in an epigenetic manner in colonial organisms.

Materials and methods

Animals

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

Vascularization

Zooids and buds were extirpated from colonies with razor blades. The remaining peripheral common vascular system was allowed to form vascular buds (Oka & Watanabe 1959; Sabbadin et al. 1975). Regenerating colonies were cut into halves 3, 14 and 30 days after the vascularization. One was fixed for in situ hybridization and RNA extraction, and the other was allowed to develop further.

RNA extraction and cDNA pool

Total RNA was extracted from sexually matured colonies by the AGPC method (Chomczynski & Sacchi 1987). Poly(A)+ RNA was then purified using Oligotex-dT30 Super (Takara Bio, Shiga, Japan). Complementary DNA was synthesized from approximately 200 ng of Poly(A)+ RNA using dT17(CAG) primer. Moloney Murine Leukemia Virus reverse transcriptase was used for the reaction (Invitrogen, Carlsbad, CA, USA).

Degenerate polymerase chain reaction

cDNA fragments of Botryllus Vasa homologue were obtained by polymerase chain reaction (PCR), using the cDNA pool as template. According to Shibata et al. (1999), the following degenerate primers were used 5′-CGGGATCCA(AG)AC(TCAG)GG(TCAG) (TA)(CG)(TCAG)GG(TCAG)AA(AG)AC-3′ and 5′-CCCAAGCTT(AG)AA(TCAG)CCCAT(AG)TC(TCAG)A(AG)CAT-3′. Each primer was designed to have cleavage sites for the restriction enzymes BamHI and HindIII. The conditions of degenerate PCR were according to Fujimura and Takamura (2000).

5′- and 3′-rapid amplification of cDNA ends

According to Frohman et al. (1988), 5′- and 3′-ends of cDNA were elongated. Five kinds of primers for 5′-rapid amplification of cDNA ends (RACE) were as follows: GSP1, 5′-CCGACAACAATTGCCTGCGGAGTTTGC-3′; GSP2, 5′-TCCAGCAAGGACAGGTAGCA-3′; GSP3, 5′-TTCATCTTCAGGAGGTGGTGGT-3′; GSP4, 5′-GAAACAGCCTTTTGGACGAG-3′; GSP5, 5′-CGAGAACTGCCTCCACCATT-3′.

Two specific primers were designed for 3′-RACE: GSP6, 5′-CAACCTGTAGTTGCGTATGGTG-3′; GSP7, 5′-GTGCAAAGTCAGCTTCGAGATT-3′.

Two kinds of adapter primers and adapter-dT17 primers were as follows: AP1, 5′-GACTCGAGTCGACATCG-3′; and AP1-dT17, 5′-GACTCGAGTCGACATCGAT17-3′ for 5′-RACE; AP2, 5′-AATCTAGAGCTCACTAGTAG-3′; and AP2-dT17, 5′-AATCTAGAGCTCACTAGTAGT17-3′ for 3′-RACE.

DNA sequencing

Polymerase chain reaction products were subcloned to a TA cloning vector (pGEM-T; Promega, Madison, WI, USA). The nucleotide sequences of cDNA inserts were determined with 373 A sequencing system and ABI PRISM 3100-Avant genetic analyzer system (Applied Biosystems, Foster, CA, USA). For the cycle sequencing reaction, Thermo Sequence Dye-terminator Cycle Sequencing kit (Amersham Biosciences, Piscataway, NJ, USA) and BigDye terminator v3.1 cycle sequencing kit (Applied Biosystems) were used.

Reverse transcription–polymerase chain reaction analysis

Total RNA were extracted from vascularized colonies. Before the reverse transcriptional reaction, they were treated with DNase I (Roche Diagnostics, Penzberg, Germany). The following primers were used: BpVas1, 5′-GATGCCATTCCTGTTGAGGTG-3′; BpVas2, 5′-TTCAAAGCCCATATCCAACAT-3′; BpVas3, 5′-GAATTCATGTTTGAGGATGACAACTG-3′; and GSP2 (mentioned above).

Whole mount in situ hybridization

Specimens were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C for 10 h. The fixed specimens were rinsed in PBST (PBS containing 0.1% Tween-20), and were dehydrated in graded series of methanol and stored in 75% methanol at −20°C. In situ hybridization was performed according to Hisata et al. (1998) with some modifications. A digoxigenin (DIG)-labeled antisense RNA probe was used for hybridization. To prevent non-specific hybridization, dehydrated specimens were treated with xylene at room temperature for 30 min before proteinase K treatment. After hybridization they were treated with an anti-DIG antibody labeled with alkaline phophatase. To avoid non-specific staining, they were pretreated with blocking solution containing 2% skim milk on ice for 15–18 h before antibody treatment. Specimens stained were embedded in JB-4 plastic resin (Polyscience, Warrington, PA, USA), and serial sectioned at a thickness of 2 µm.

Transmission electron microscopy

Specimens were fixed for 2 h on ice with 2.5% glutaraldehyde solution containing 0.45 m sucrose buffered with 0.1 m cacodylate at pH 7.4. They were then rinsed with the same buffer and postfixed for 1.5 h with 1% osmium tetroxide in the same buffer without sucrose. After dehydration, specimens were cleared with n-butyl glycidyl ether and embedded in low-viscosity epoxy resins. Thin sections were double-stained with uranyl acetate and lead citrate and examined with JEM-1010 transmission electron microscope (JEOL, Tokyo, Japan) at 80 kV.

Results

Isolation and characterization of cDNA of Botryllus primigenus Vasa homologue

We isolated PCR fragments that were approximately 400 bp in length from the B. primigenus cDNA pool by using degenerate primers designed from the conserved Vasa sequences of various animals (Fig. 2A, lane 1). Two different types of DEAD box genes were cloned: one of the genes was BpDEAD1, a putative Vasa homologue; the other was BpDEAD2, which was similar to the PL10 gene in zebrafish (Olsen et al. 1997). We focused mainly on BpDEAD1 for further analysis.

Figure 2.

Isolation of cDNA and primary structure of deduced protein of BpDEAD1(BpVas). (A) cDNA fragments of BpDEAD1 amplified by polymerase chain reaction (PCR). Lane 1, approximately 0.4 kb product after degenerate PCR. Lane 2, the longest cDNA of BpDEAD1, approximately 2.3 kb in length. Lanes 3 and 4, RT–PCR of the 5′ region of BpDEAD1. Two products were identified (Lane 3). No products were amplified without reverse transcriptase (Lane 4). (B) Deduced amino acid sequence of BpDEAD1. Eight conserved motifs are indicated by shaded boxes. Allows show five tandem repeats that begin with ‘G’. Each repeat contains a zinc-finger motif (cf. Fig. 2D). (C) Multiple alignment of BpDEAD1 and other Vasa proteins. Conserved amino acids (shaded boxes) are located at the C-terminal two-thirds region of total proteins. Acidic amino acid-enriched N- and C-terminus are indicated by hatched boxes. RGG boxes of Vasa proteins are underlined. (D) Alignment of five tandem repeats having zinc-finger motifs in BpDEAD1. Conserved C, C, H and C are indicated by hatched boxes. Identical amino acid residues are shaded.

In order to obtain longer cDNA fragments, we performed 5′- and 3′-RACE. On the basis of a 5′ sequence of approximately 1.1 kb and a 3′ sequence of approximately 1.2 kb, a single cDNA of 2.3 kb was obtained (GenBank accession number, AB235177; Fig. 2A, lane 2). It encoded 687 amino acid residues followed by approximately 150 bp of a 3′-untranslated region. The deduced polypeptide (BpDEAD1) showed a high similarity to other VASA proteins at two-thirds of its C-terminal portion (Fig. 2C, shaded). It contained all eight motifs, similar to the other DEAD box proteins (Linder et al. 1989; Fig. 2B, shaded). The N-terminal region was rich in glycine (23% of 1–180 amino acids), and GGG regions were often observed instead of RGG, unlike other DEAD box proteins (Fig. 2C, underlined). In addition, five tandem repeats of a zinc-finger motif CCHC were located at the N-terminal region (Fig. 2B). When RT–PCR was performed to amplify the 5′-coding region, two bands that were 877 and 952 bp in length became visible (Fig. 2A, lane 3). The longer one had a duplication of zinc finger in the repeat 3 (Fig. 2D), suggesting that there might be two transcripts in BpDEAD1.

A similarity search using GenBank and FASTA databases showed that BpDEAD1 shared maximum similarity with the Ciona Vasa homologues CiVH and Cs-DEAD1. A phylogenetic analysis of the full-length amino acid sequence by UPGMA methods indicated that BpDEAD1 formed a cluster along with members of the Vasa subfamily (Fig. 3). The phylogenic tree by the neighbor-joining (NJ) method also showed an identical branching pattern (not shown). Based on the primary structure and phylogenetic analysis, we consider that BpDEAD1 is a B. primigenus Vasa homologue (BpVas).

Expression of BpVas in the process of germ cell formation

In B. primigenus, germ-cell formation begins in the gonadal space with loose cell masses that consist of hemoblasts (Mukai & Watanabe 1976). The loose cell mass and developing germ cells specifically expressed BpVas, but the somatic cells and tissues such as the epidermis, pharynx, atrial epithelium, endostyle, heart and coelomic cells did not (Fig. 4A). It is noteworthy that every cell in the cell mass was stained at this stage, although accessory cells such as the PFC developed later from this cell mass (Fig. 4B). The cell mass of hemoblasts also appeared during the process of vascular budding (Fig. 4C, arrow). No BpVas was expressed by this mass (Fig. 4C, arrow), which indicated that cell aggregation by itself did not cause BpVas expression. The oocytes expressed BpVas mRNA, while the PFC surrounding them did not (Figs 4D,5D). The spot-like signals were often observed in the cytoplasm of developing oocytes with a prominent germinal vesicle (Fig. 4E). Thereafter, at the vitellogenesis stage, the signal became weak and disappeared from the oocytes (Fig. 4B,D,H). The compact cell mass, a precursor of the testis, showed a strong BpVas signal (Fig. 4F). It differentiated into the peripherally situated testicular epithelium (Fig. 4G, arrow) and the spermatogonia and spermatocytes internally situated. Only the spermatogonia and spermatocytes retained a strong signal (Fig. 4G). The signal became weak in the developing testis (Fig. 4H). Finally, no signal was detected in the well-developed gonad (cf. Fig. 4I).

Figure 4.

Expression of BpVas mRNA. (A) A developing bud. It should be noted that the signal was detected only in the germline cells located in the gonadal space (arrowhead). Bar, 100 µm. (B) Loose cell masses in the gonadal space (arrowheads), developing bud. Bar, 30 µm. (C) Vascular bud in the tunic vessel. Aggregating hemoblasts emitted no signals (arrow). Bar, 50 µm (C′) Toluidine blue staining of the aggregate (arrow). (D) Developing oogonia (arrowhead). It should be noted that primary follicle cells (PFC) attached to oogonia were not stained at all (arrow). Bar, 20 µm. (E) Developing, juvenile oocytes. They emitted spot-like signals (arrowheads), and later the signal diminished (cf. Fig. 4D). Bar, 20 µm. (F) Compact cell mass (arrowhead). Bar, 20 µm. (G) Developing testis. Germ cells located inside kept the signal, but the testicular epithelium did not (arrow). Newly formed loose cell masses were seen beside the testis (arrowhead). Bar, 30 µm. (H) Germ cells of various developmental stages. Loose cell masses had strong signals. In the testis, only peripheral germ cells emitted weak signals. Oocyte at vitellogenesis stage had no signal. Bar, 50 µm. (I) A well-developed gonad. A semithin section of the ovary and testis stained with toluidine blue. Bar, 100 µm. a.e, atrial epithelium; c.c, coelomic cells; en, endostyle; ep, epidermis; h, heart; p, pharynx; oc, oocyte; ov, ovum; te, testis; tu, tunic.

Figure 5.

Ultrastructures of a loose cell mass and oogonia. (A) Loose cell mass (arrowheads) located in the gonadal space. It was associated with the atrial epithelium (arrows) of a developing bud. Bar, 5 µm. (B) A hemoblast circulating in the vascular network. Bar, 1 µm. (C) Focal junctions (arrowhead) between adjacent hemoblasts in the loose cell mass. Bar, 0.5 µm. (D) An oogonium partially wrapped by primary follicular cells (PFC). Both oogonium and PFC have the large nucleus and prominent mitochondria (asterisks). Arrow shows cell junction between the oogonium and PFC. Bar, 2 µm. (E) Electron-dense material (arrowhead) in the vicinity of a mitochondrion (asterisk) in an oogonium. It was not associated with any membrane structures. Bar, 1 µm.

Ultrastructure of BpVas-expressing cells in the ovary

Cells of the loose cell mass were similar in shape, having a large nucleus, a prominent nucleolus and few organelles, including polysomes in the cytoplasm (Fig. 5A). It should be noted that at this stage, the cells did not contain electron-dense materials or localized mitochondria in the cytoplasm. They shared the undifferentiated features with free circulating hemoblasts (Fig. 5B). Cell aggregates had focal junctions in places (Fig. 5C).

At the next stage, oogonia with a diameter of 5–7 µm appeared in the loose cell mass. They were surrounded by PFC that had a large nucleus and were without a nucleolus (Fig. 5D). They contained small patches of dense material in the cytoplasm near the nuclear membrane (Fig. 5E). Junctional complexes were observed between the oogonium and the PFC (Fig. 5D, arrow). Young oocytes with a diameter of 10–12 µm were associated with each other (Fig. 6A). In the nucleus, synaptonemal complexes were occasionally observed (Fig. 6B, arrowhead), thereby indicating that the oocyte was in the zygoten–pachyten phase in meiosis. The oocyte possessed mitochondrial clusters (Fig. 6C, asterisks) and electron-dense materials (Fig. 6C, arrowhead) near the nuclear membrane. The PFC gradually flattened and surrounded the oocyte completely (Fig. 6A,D). In the developing oocytes that had a diameter of 50–70 µm, mitochondria increased in number (Fig. 6D). The electron-dense materials became prominent in the vicinity of the nucleus (Fig. 6E, arrowheads). In some cases, they were associated with multimembranous bodies (Fig. 6F, arrowhead and arrows).

Figure 6.

Ultrastructures of oocytes. (A) Young oocytes (oc) with a multinucleolar nucleus. They were enclosed by primary follicular cells. Bar, 5 µm. (B) Synaptonemal complexes found in the nucleus (arrowhead). Bar, 1 µm. (C) Electron-dense material (arrowhead) associated with mitochondria (asterisks). Bar, 1 µm. (D) Growing oocyte with a large number of mitochondria scattered in the cytoplasm. Bar, 10 µm. (E) Electron-dense materials (arrowheads) located close to the nuclear membrane. Bar, 1 µm. (F) A large number of multimembranous bodies (arrows) associated with electron-dense materials (arrowhead). Bar, 2 µm.

Disappearance and reappearance of BpVas in vascularized colonies

As already shown, the BpVas-positive cells exist only in the gonadal space of the buds and developing zooids. Therefore, we examined whether the BpVas-positive cells disappeared completely in the vascularized colonies from which the zooids and buds were extirpated (Fig. 7A). In the vascularized colonies, zooids and buds have the ability to regenerate by vascular budding (Oka & Watanabe 1957; Oka & Watanabe 1959; Milkman 1967; Sabbadin et al. 1975; Fig. 7B, arrowheads). Thus, our next question was whether the BpVas-positive cells could be generated de novo in the vascularized colonies.

Figure 7.

Disappearance and reappearance of BpVas in vascularized colonies. (A) A specimen immediately after vascularization. All zooids and buds were removed from colony. (B) A specimen, 3 days after vascularization. New zooids appeared by vascular budding (arrowheads). Bar, 1 mm. (C) Expression of BpVas in vascularized colonies examined by reverse transcription–polymerase chain reaction (RT–PCR). No signal was detected from 3 day regenerating colonies (left), whereas it became detectable in 30 day colonies (right) only in the presence (+) of reverse transcriptase (RT). (D) Expression of BpVas in vascularized colonies examined by in situ hybridization. BpVas-positive cells were not detected in 3 day regenerating colonies. Arrows indicate vascular buds. Bar, 100 µm. (Inset) In one exceptional case, an oogonium was found in the vascular network. Bar, 10 µm. (E) BpVas-positive germline cells in a 30 day colony (arrowhead). Bar, 50 µm.

A total of 25 specimens (vascularized pieces) were prepared from three independent colonies (A, B and C; Table 1), and they were cut into halves at 3, 14 and 30 days, respectively, after the vascularization. One of the colonies was allowed to grow naturally in a culture box. We confirmed that every specimen regenerated zooids and buds. The other was fixed for in situ hybridization or stored at −80°C for RT–PCR.

Table 1. BpVas-positive cells in vascularized colonies examined by in situ hybridization
SampleDays after vascularization
3 days14 days30 days
  1. —, not examined.

A1NoNoYes
A2NoNoYes
A3YesYes
A4NoYes
A5NoYes
B1NoYes
B2NoYes
B3NoYes
B4NoYes
B5NoNoYes
C1NoYes
C2NoYes
C3NoYes
C4NoYes
C5NoYes

The total RNA was extracted from 10 specimens of colonies A and B. Following RT–PCR, the bands remained undetectable in the 3 day specimens (Fig. 7C, left). On the other hand, a single clear signal was detected in the 30 day specimens in the presence of reverse transcriptase (Fig. 7C, right). On sequencing, the band was confirmed to be a fragment of BpVas cDNA (not shown). Unfortunately, we could not extract total RNA from 14 day specimens, but the data of in situ hybridization was available.

Five specimens from respective colonies A, B and C were used for in situ hybridization (Table 1). In 14 of the 15 3 day specimens, no BpVas-expressing cells could be detected (Table 1, Fig. 7D). In an exceptional case (A3), a single large cell emitted the signal (Fig. 7D, inset), and the cell might have been a migrating oogonium with PFC, as reported by Mukai and Watanabe (1976). With respect to 14 day specimens, the results varied across the colonies. The specimens that originated from colonies A and B did not emit any signals, while those from colony C contained BpVas-positive cells in the gonadal space (Table 1). All 10 specimens of the 30 day regenerating colonies possessed BpVas-expressing loose cell masses, oocytes and developing testes (Fig. 7E). These results suggested that BpVas-positive germ cells might have reappeared approximately 2 weeks after the onset of colonial regeneration.

Discussion

BpVas is a homologue of Vasa

Vasa homologues and Vasa-related genes belong to the DEAD box family. Proteins of this family share eight conserved motifs including the ATPase A motif (QTGSGKT), ATPase B motif (DEAD), RNA-biding motif (HRIGRTGR) and RNA-unwinding motif (SAT) (Pause & Sonenberg 1992; Pause et al. 1993). BpVas, like the Vasa subfamily, was characterized additionally by a glycine-rich sequence at the N-terminal region and acidic amino acids near the N- and C-terminus (cf. Mochizuki et al. 2001), but in BpVas, GGG appeared repeatedly at the N-terminal region. The RGG-box is known to bind to RNA (Kiledjian & Dreyfuss 1992).

BpVas contained five repeats of the zinc-finger motifs (CCHC) at the N-terminal region. In retroviruses, the CCHC zinc finger binds to nucleic acids (Rajavashisth et al. 1989). This type of motif has been reported in C. elegans (Gruidl et al. 1996; Kuznicki et al. 2000), hydra (Mochizuki et al. 2001), brachiopods (Sagawa et al. 2005), oyster (Fabioux et al. 2004) and ascidians (Fujimura & Takamura 2000).

Phylogenetic analyses have indicated that BpVas is a member of the Vasa subfamily, but the molecular phylogeny is inconsistent with the evolutional branching order. That is, the BpVas and Vasa homologues of other ascidians showed a close relationship to those of hydra, oyster and water fleas rather than vertebrates (cf. Fig. 3). This may be due to the presence of multiple CCHC motifs in their N-terminal regions. It is possible that the number and position of CCHC motifs are not affected by the evolution of Vasa genes.

Figure 3.

Phylogenetic tree of BpDEAD1 (underlined) and several DEAD-box gene proteins of other animals. The tree was drawn by the UPGMA method. The bootstrap values are shown on each branch.

The present study has suggested that there may be two types of BpVas mRNA that differ from each other with respect to their length. The shorter one was identical to the cDNA (GenBank accession number AB235177) reported here, while the longer one had a redundancy in the third CCHC motif (cf. Fig. 2D). Two types of Vasa mRNA and/or protein have been reported in teleost fishes, tilapia and zebrafish (Olsen et al. 1997; Yoon et al. 1997; Kobayashi et al. 2000; Kobayashi et al. 2002). In tilapia, the full length Vasa is expressed predominantly during spermatogenesis, while Vasa lacking 24 amino acid residues at the N-terminal region (Vas-s) is expressed predominantly during oogenesis. Whether the longer BpVas has an expression pattern and function that is distinct from the shorter one remains to be elucidated.

Molecular and cellular characterization of loose cell mass in B. primigenus

According to Mukai and Watanabe (1976), all germ cells and their accessory cells in B. primigenus are derived from the loose cell mass, and the loose cell mass from hemoblasts, which are undifferentiated cells present in the hemocoele. In this paper, we examined their light microscopic observation by electron microscopy, and confirmed that in B. primigenus, the cell mass was composed of similar, undifferentiated cells that could not be distinguished from hemoblasts in the hemocoele. In Botryllus schlosseri, putative germ cells in the gonadal space have a nucleus that is slightly larger than that of the hemoblasts (Sabbadin & Zaniolo 1979). We observed the same type of cells in B. primigenus and considered them as oogonia (and probably spermatogonia) because they possess a few electron-dense materials in the cytoplasm, unlike the loose cell mass.

In B. primigenus, somatic cells and tissues never expressed BpVas. The loose cell mass was the most primitive tissue that expressed BpVas. Expression was detected in oogonia and young oocytes in the ovary and by the compact cell mass, spermatogonia and young spermatocytes in the testis (Fig. 8). Gene expression disappeared permanently from accessory cells such as PFC and the testicular epithelium. These results have shown that in B. primigenus, BpVas is a reliable marker for germline cells.

Figure 8.

Schematic representation of BpVas expression with reference to germ line specification in B. primigenus. Hemoblasts aggregate in the gonadal space to form the loose cell mass. The loose cell mass is the germ line precursor that expresses first and strongly BpVas (blue). Accessory cells and epithelial cells derived from loose cell mass or compact cell mass lose BpVas signals (black), so that the expression is localized strictly to germ line cells (blue). Meanwhile, hemoblasts also aggregate in the vascular network to form the vascular bud. It is uncertain at present whether single hemoblasts have bipotency giving rise to germ line cells and somatic line cells or discrete hemoblasts are destined to soma and germ, respectively.

Electron-dense materials have been known to be another marker that characterizes germline cells (Eddy 1975; Extavour & Akam 2003). They are associated with mitochondria in the vicinity of the nuclear membrane (Inoue & Shirai 1991; Flores & Burns 1993). Some researchers insist that the electron-dense materials contain a germ cell determinant (Illmensee & Mahowald 1974; Ikenishi 1998). It is noteworthy that thus far in B. primigenus, electron-dense materials were not observed in the BpVas-positive loose cell mass, but it appeared during the later stages in the oogonia. Consistent with this finding, the nuage or electron-dense materials have been reported to be present in the oogonia in B. schlosseri (Manni et al. 1994) and Ciona savignyi (Sugino et al. 1990); however, in C. intestinalis, they appear in more primitive germ cells located in the gonad rudiment (Yamamoto & Okada 1999). These observations appear to be inconsistent with the idea that electron-dense materials appear at the earliest stage of germ cell formation, and they play a decisive role in germ cell determination.

Epigenetic appearance of BpVas-positive germline cells

In Botryllid colonial ascidians, when all the zooids and buds are removed from the colony (vascularization), new buds are generated by vascular budding from tunic vessels, thereby resulting in regeneration of the colony (Oka & Watanabe 1957, 1959; Milkman 1967; Sabbadin et al. 1975). Utilizing this feature of Botryllus, we investigated the regenerative potential of germ cells in vascularized colonies. Results of both RT–PCR and in situ hybridization showed that BpVas transcripts were depleted completely from the colonies after vascularization, and that they reappeared 14–30 days later. This is the first quantitative evidence that in B. primigenus the germline cells are reproducible epigenetically at postembryonic stages.

The epigenetic mode of germ cell specification has been suggested in the solitary ascidian C. intestinalis (Takamura et al. 2002). In this species, primordial germ cells derived from B7.6 blastomeres are distributed in the larval tail (Fujimura & Takamura 2000), but the gonad and germ cells are formed in the future zooids even although the larvae were allowed to metamorphose after the larval tail was cut off (Takamura et al. 2002). Although the mechanism by which the germ cell population differentiates is unclear, zooids in C. intestinalis have the ability to generate germ cells by means of both predeterminative and regulative pathways. In B. primigenus, the regulative pathway continues throughout the adult (asexual) life cycle. Some epigenetic manner of induction may occur in the gonadal space, as the gonadal space is the only place where BpVas-positive cells appear.

Possible regulative interaction of hemoblasts for BpVas expression

It is clear that in B. primigenus, some BpVas-negative cells have changed to BpVas-positive cells after vascularization. On the basis of ultrastructure, the most primitive BpVas-positive cells could be identified as hemoblasts, although it is uncertain at present if germline hemoblasts are identical to somatic hemoblasts that generate vascular buds (Fig. 8). Hydras and planarians have multipotent stem cells such as interstitial cells and neoblasts (Baguna 1981; Bosch & David 1987; Bode 1996). In both the cases, Vasa-related genes are expressed not only by the germline cells but also by the somatic stem cells (Shibata et al. 1999; Mochizuki et al. 2001). In the lower metazoans, Vasa-related genes may be required for the maintenance of the undifferentiated state of multipotent cells (Mochizuki et al. 2001). On the other hand, in vertebrates, Vasa homologues and Vasa-related genes are involved specifically in germ line specification, although the Xenopus Vasa-related gene is expressed in the somatic cells of the embryo (Ikenishi & Tanaka 2000).

In B. primigenus, BpVas-positive germline cell formation was initiated in the loose cell mass. The loose cell mass accompanied intercellular junctional complexes. It is unlikely that the process of cell adhesion and/or adhesion apparatus induce BpVas by themselves because hemoblasts that aggregated beneath the vascular epithelium (vascular buds) did not express any BpVas signals (Figs 4C,8). In Polyandrocarpa misakiensis, a related species of B. primigenus, the formation of gonads was also initiated in cell aggregates. It enters the gonadal pocket formed by the atrial epithelium that expresses p68 RNA helicase, another type of DEAD box gene (our unpubl. data, 2005). Cell interactions between coelomic hemoblasts and the atrial epithelium of the gonadal space or gonadal pocket may be important for epigenetic germ line determination in colonial ascidians.

Acknowledgements

This work is dedicated to the late Professor Hideo Mukai, Gunma University, with deep respect. We express special thanks to Dr Shigeki Fujiwara of Kochi University for encouragement, discussion and technical instruction throughout this study. We thank Dr Norito Shibata and Dr Maki Shirae-Kurabayashi for their advice. We are also grateful to the stuff of the Usa Marine Biological Institute of Kochi University and the Shimoda Marine Research Center, University of Tsukuba (SMRC) for their assistance and hospitality. This study is contribution no. 715 from SMRC.

Ancillary