INTRODUCTION

In addition to the usual developmental processes involved in forming a tailed (tadpole or urodele) larva, a dozen or so ascidian species, exhibit other modes of embryonic development or the formation of tailless (anural) larvae (Lacaze-Duthiers, 1874; Berrill, 1931; Millar, 1954; Whittaker, 1979; Swalla and Jeffery, 1990, 1992; Bates and Mallett, 1991; Bates, 1995). These tailless (or anural) species belong to the families Molgulidae and Styelidae and include Molgula bleizi, M. occulta, M. arenata, M. pacifica, M. provisionalis, M. robusta, M. retortiformis, Bostrichobranchus pilularis, B. digonas, Eugyra arenosa, Pelonaia corrugata and Polycarpa tinctor (reviewed by Jeffery and Swalla, 1990, 1992; Bates, 1993). Anural species may be subdivided into those that exhibit indirect (incomplete) or direct (complete) development. In species with indirect anural development, such as M. arenata and M. occulta, a tailless, slug-like larva hatches from the chorion before metamorphosis (Whittaker, 1979; Swalla and Jeffery, 1990). Species with direct anural development, such as M. provisionalis and M. pacifica (Bates and Mallett, 1991; Bates, 1995), have completely eliminated the larval stage. In these species, hatching occurs after the beginning of metamorphosis, which is initiated within the chorion. A recent molecular phylogenetic analysis suggests that anural species have been derived at least five different times from ancestors with urodele larvae (Hadfield et al., 1995).

The tadpole larva has a head containing a dorsal brain and pigmented sensory organs (otolith and ocellus) and a tail containing a notochord, spinal cord, and striated muscle cells. These chordate features are lacking in anural larvae, although undifferentiated precursor cells are frequently observed (Whittaker, 1979; Swalla and Jeffery, 1990). Therefore, ascidian species with these two modes of development provide an opportunity to study molecular mechanisms involved in an evolutionary change in development (Swalla and Jeffery, 1990, 1996; Swalla et al., 1993; Satoh and Jeffery, 1995; Kusakabe et al., 1996).

No ascidian species with anural development have been previously reported from the coast of Japan. Recently, an ascidian species belonging to the family Molgulidae has appeared in large numbers off the coast of northern Japan. This new species has been described by Nishikawa (1991) as Molgula tectiformis. Here we report that M. tectiformis develops without a larval stage and is another example of an anural (direct) developer.

MATERIALS AND METHODS

RESULTS

Embryogenesis of M. tectiformis

Figure 1 shows the embryonic development of M. tectiformis. The egg was orange in color and about 150 μm in diameter (Fig. 1A). It is surrounded by numerous, large test cells which vigorously interact with the surface of the embryo during development. M. tectiformis is self-fertile: eggs fertilized with sperm of the same individual develop normally. In this study, however, we followed embryogenesis of eggs fertilized with sperm of another individual. Fertilization initiates ooplasmic segregation. After ooplasmic segregation, the animal hemisphere became transparent whereas the vegetal hemisphere became orange-brown (compare Fig. 1B with 1A). The transparent, animal cytoplasm was segregated mainly into the epidermis of the embryo, while the orange-brown vegetal cytoplasm was partitioned mainly into the endoderm of the embryo (Fig. 1). Whether the myoplasm, an egg cytoplasmic region containing tail-muscle cell determinants (cf. Satoh, 1994), was present in M. tectiformis eggs could not be determined.

Fig. 1

Normal embryogenesis of the ascidian Molgula tectiformis. (A) An unfertilized egg. (B) A 2-cell embryo, lateral view. The animal pole is up. (C) A 4-cell embryo. (D) An 8-cell embryo, posterior view. (E) A 16-cell embryo, vegetal pole view. The anterior is up. (F) An embryo at about 64-cell stage, vegetal pole view. (G) An embryo at about 100-cell stage, lateral view. The vegetal side is flattened. (H) A middle gastrula (5-hr embryo), lateral view. The animal pole is up. Invagination of the vegetal cells is seen (arrowhead). (I) An 11-hr embryo. Neurulation is evident by an appearance of a slit (arrowhead). (J) A 12-hr embryo. Formation of a stolon is evident (arrowhead). (K) A 13-hr embryo immediately before hatching. Arrowheads indicate stolons. (L) A hatched juvenile. Three adhesive stolons develop (arrowheads). Scale bar represents 50 μm for all panels.

As in the case of ascidians with urodele larvae, cleavage of M. tectiformis eggs was bilaterally symmetrical. The first and second cleavages were meridional, dividing the egg into two (Fig. 1B) or four cells (Fig. 1C), usually of equal sizes. The third cleavage was latitudinal, and it usually produced animal and vegetal cells of unequal sizes. In 8-cell embryos, the four animal cells were transparent and small, whereas the four vegetal cells were intensely in orange color and large (Fig. 1D). Therefore, the future dorsal (vegetal)-ventral (animal) axis could be discerned by the position of colored blastomeres at the 8-cell stage. At the fourth cleavage of ascidian eggs, the division pattern of the animal tier cells is reversed relative to the vegetal tier cells (cf. Satoh, 1994). This division pattern reversal was also observed at the fourth cleavage in M. tectiformis embryos (Fig. 1E). The fifth cleavage occurred about 20 min later to form a 32-cell embryo. The subsequent cleavages increased the number of embryonic cells, although precise division pattern could not been followed (Fig. 1F, G). Gastrulation occurred at about 4.5 hr after fertilization (18°C). At first, the vegetal side of the embryo flattened (Fig. 1G). Then the vegetal cells invaginated (Fig. 1H). Gastrulation appeared to be accompanied by epibolic movement of the animal cells. At about 11 hr after insemination and after the completion of gastrulation, a transparent stripe became visible on the posterior-vegetal side of the embryo. This transparent stripe may correspond to the position of the neural tube, which forms on the dorsal side of ascidian embryos during neurulation (cf. Satoh, 1994). The morphogenetic processes of gastrulation and neurulation were completed within a few hours.

Later during embryogenesis, the outer surface of the embryo became transparent, indicating the formation of the epidermis (Fig. 1J). No other conspicuous changes in the embryonic morphology were evident during later phases of embryogenesis. About 13 hr after fertilization, M. tectiformis embryos hatched from the chorion. A portion of the chorion became thin, a hole was eventually formed in this region, and the juvenile passed through the hole in the chorion (Fig. 1J). The hatched larva was ovoid. Three transparent ampullae appeared while the embryo was still in the chorion (Fig. 1J, K). Following hatching and settlement, they form a firm attachment to the substratum (Fig. 1L).

Therefore, M. tectiformis does not develop a conventional tadpole larva. Because juvenile formation took place during late embryogenesis, this species is also a direct developer. Other direct-developing ascidian species are B. digonas (Swalla and Jeffery, 1992) and M. pacifica (Young et al., 1988; Bates and Mallett, 1991): there was no tail region with apparent differentiation of the notochord and muscle (Fig. 1). No otolith and ocellus pigment cells were observable in the embryo (Fig. 1). In addition, no larval adhesive organ was evident (Fig. 1).

After hatching, the formation of young adults proceeded rapidly (Fig. 2). Juveniles soon lost the ampullae. About 3 or 4 days after hatching, the development of various internal organs was evident in the young adult (Fig. 2B).

Fig. 2

Development of Molgula tectiformis juveniles. (A) A juvenile at several hr after hatching showing three ampullae. (B) A young adult 4 days after hatching, showing the development of various internal organs. Scale bar, 50 μm for A and 200 μm for B.

Muscle cell differentiation in M. tectiformis embryos

During indirect anural development in M. occulta, neither the larval muscle actin nor the myosin heavy chain genes are expressed but the muscle-lineage cells develop the muscle-specific enzyme AChE (Jeffery and Swalla, 1991; Kusakabe et al., 1996). As described above, M. tectiformis is a direct developer, suggesting that differentiation of muscle cells is suppressed. We examined this issue by detection of AChE by histochemistry and muscle actin gene expression by in situ hybridization.

As shown in Fig. 3A, the control C. savignyi larva developed AChE activity in the tail muscle cells. However, AChE activity was not detected in M. tectiformis embryos (Fig. 3B). This was confirmed by examining more than 30 embryos at different stages of later embryogenesis (at 10 and 12 hr of development). AChE activity, however, was detected in the body-wall muscle cells of juveniles 4 days after hatching (data not shown). The results show that M. tectiformis embryos do not express AChE.

Fig. 3

Histochemical detection of tissue-specific enzymatic activities. (A, B) Detection of muscle-specific AChE. (A) Ciona savignyi larva (control) showing AChE activity in muscle cells of the tail (arrow). (B) AChE activity is not detected in Molgula tectiformis embryo at 13.5 hr after fertilization. (C, D) Histochemistry of endoderm-specific AP. (C) M. tectiformis juvenile at 13.5 hr of development does not show AP activity. (D) AP activity is evident in the endoderm of a M. tectiformis juvenile at 4 days after hatching (arrowhead) as well as C. savignyi larva (arrow; control). (E, F) Histochemical detection of pigment cell-specific tyrosinase. (E) C. savignyi tailbud embryo showing tyrosinase activity in two pigment cells (arrows). (F) Tyrosinase activity is not detected in M. tectiformis embryo at 13.5 hr of development. Scale bars represent 50 μm.

Sections of M. tectiformis embryos at 12 hr of development were hybridized in situ with the antisense MocuMA1 probe to determine whether muscle actin is expressed. The MocuMA1 probe, which contains most of the coding region of an M. oculata larval-muscle actin gene, has been shown to detect muscle actin transcripts in urodele ascidian embryos (Kusakabe et al., 1996). As shown in Fig. 4A, the MocuMA1 probe detected transcripts in tail muscle cells of M. occidentalis, the urodele developer used as a control in our experiments. In contrast, muscle actin transcripts could not be detected in M. tectiformis embryos, although small cells visible in the posterior region may be vestigial-muscle cell precursors (Fig. 4B). The results suggest that M. tectiformis embryos do not express muscle actin.

Fig. 4

Detection of muscle actin transcripts by in situ hybridization. (A) M. occidentalis embryo (control) at early tailbud stage showing transcripts in the tail muscle cells. (B) Transcripts are not detected in a 12 hr. M. tectiformis embryo, although small unlabeled cells are present in the posterior region (arrowheads). Scale bar represents 15 μm for all panels.

In summary, the lack of AChE and muscle actin expression suggests that M. tectiformis embryos lack differentiated muscle cells.

DISCUSSION

In this investigation, we have shown that M. tectiformis is an anural (direct) developer. Therefore, similar to other anural species, M. tectiformis does not exhibit a tadpole larva. Despite this modification in the conventional mode of embryogenesis, fertilized M. tectiformis eggs undergo ooplasmic segregation and cleavage similarly to those of urodele ascidian species, and they gastrulate and form a neural tube. After neurulation, however, they do not differentiate a pigmented neural sensory organ or exhibit the morphogenetic movements resulting in notochord and tail development. In addition, M. tectiformis embryos do not express AP, suggesting a deficiency in endoderm differentiation. Similar to a few other anural species, including M. bleizi, M. pacifica, M. retortiformis, and B. digonas (Berrill, 1931; Bates and Mallett, 1991; Swalla and Jeffery, 1992; Bates, 1995), M. tectiformis embryos begin metamorphosis prior to hatching. In most urodele ascidian species, metamorphosis, as marked by the extension of the ampullae, is initiated after hatching (cf. Satoh, 1994). Thus, there appears to be two major differences in development between M. tectiformis and conventional urodele ascidian embryos. The first difference, which is shared with all other anural species, is that M. tectiformis embryos lack the urodele neural sensory and tail structures. The second difference is that metamorphosis begins precociously before hatching. The form of anural development characteristic of M. tectiformis may have evolved through a combination of elimination and heterochronic acceleration of developmental processes.

The anural species that has been most thoroughly investigated is M. occulta, an indirect developer (Swalla and Jeffery, 1990, 1996; Jeffery and Swalla, 1991, 1992; Kusakabe et al., 1996). Although no tail appears during embryogenesis, M. occulta embryos contain vestigial notochord and muscle cells, which fail to undergo terminal differentiation. The vestigial muscle cells of M. occulta embryos express AChE but they do not contain muscle actin transcripts (Swalla and Jeffery, 1990; Kusakabe et al., 1996). AChE is also expressed in the vestigial muscle cells of another anural indirect developer, M. arenata (Whittaker, 1979). A recent study showed that lack of muscle actin expression in M. occulta was caused by structural deficiencies in the coding regions rather than the regulatory machinery of the larval muscle actin genes (Kusakabe et al., 1996). In M. tectiformis embryos, we have been unable to detect either the expression of muscle actin or AChE during embryonic development, suggesting that the program of muscle differentiation has degenerated even further than in M. occulta and M. arenata. It will be interesting to determine if larval muscle actin genes are still present in the M. tectiformis genome.

The extent to which urodele features are modified differs among anural ascidian species. All known anural species, with the notable exception of B. digonas (Swalla and Jeffery, 1992), have lost the pigmented neural sensory organ (Berrill, 1931; Jeffery and Swalla, 1990). The lack of pigment cell differentiation stems from a deficiency in the melanin synthesis pathway, as shown by the absence of tyrosinase activity in M. occulta embryos (Swalla and Jeffery, 1990). In the present study, we have shown that tyrosinase activity and pigmented neural sensory cells are also absent in M. tectiformis embryos. Differences in AChE expression have also been reported in anural ascidian species. The indirect anural developers M. arenata (Whittaker, 1979) and M. occulta (Jeffery and Swalla, 1991) express AChE, whereas the activity of this enzyme is reduced to very low levels or absent in the anural direct developers M. provisionalis, M. pacifica, and B. digonas (Bates and Mallett, 1991; Swalla and Jeffery, 1992; Bates, 1995). These comparisons suggest that loss of AChE expression is correlated with the direct mode of anural development, consistent with the suggestion that these anural species are older, in an evolutionary sense, than species with indirect anural development (Bates, 1993).

There is one feature of anural development we have observed in M. tectiformis eggs and embryos that is distinct from other anural species: the fact that they are surrounded by large and active test cells. It has been shown previously that M. pacifica and M. occulta embryos either lack test cells or contain reduced numbers of smaller test cells (Young et al., 1988; Bates and Mallett, 1991). The test cells have a secretory function in tunic synthesis and assembly during larval development (cf. Satoh, 1994). The reason why M. tectiformis has retained the test cells while other anural ascidian species have eliminated them is unclear, but may be related to fundamental differences in the structure of the larval tunic.

Many anural ascidian species, including M. arenata, M. occulta, B. pilularis, and B. digonas, are restricted to rather homogeneous habitats at the ocean bottom, where they lie partially buried in sand or mud. Berrill (1931) suggested that these anural species evolved from urodele ancestors because there was no selective advantage for a larval dispersal stage in these habitats. More recently, Hadfield et al. (1995) proposed that the tadpole larva may be under negative selective pressure in patchy bottom habitats because extensive larval dispersal would interfere with settling and feeding of the rapidly-developing juveniles. It is difficult to adapt either hypothesis to M. tectiformis, which are attached to firm substrates in Otsuchi Bay. Other anural ascidian species, including M. bleizi, M. pacifica, and M. provisionalis (Berrill, 1931; Young et al., 1988; Bates and Mallett, 1991; Bates, 1995), are also attached to substrates rather than buried in sand or mud flats. It is possible that anural developers initially evolved in homogeneous bottom habitats, then speciated and invaded the shoreline, as first suggested by Berrill (1931). If this is the case, the presumed bottom-dwelling ancestor of M. tectiformis must have become extinct because no other anural Molgula species have been reported in the coastal waters of northern Japan.

Anural species have been useful as models for elucidating the molecular basis of urodele ascidian development, the evolution of developmental mechanisms, and the regulatory genes that may have operated in the ancestral chordate (Jeffery and Swalla, 1992; Swalla et al., 1993; Satoh and Jeffery, 1995; Kusakabe et al., 1996; Swalla and Jeffery, 1996). The Japanese ascidian M. tectiformis has several features that are unique among anural species, including year round availability of gametes, relatively large eggs and embryos, and existence of colored cytoplasmic regions. These features will be important assets in future studies of the mechanisms of anural development.