*Author to whom all correspondence should be addressed. Present address: Department of Biology, Graduate School of Science, Osaka University, 1-1 Machikaneyama-cho, Toyonaka, Osaka 560-0043, Japan. Email: firstname.lastname@example.org
Ascidian eggs develop into tadpole larvae. They have a simple central nervous system (CNS) at the dorsal midline. The CNS is formed through neural tube formation at the neurula stage, as in vertebrates. The total number of cells in the CNS is approximately 300. In Halocynthia roretzi, the anterior part of the CNS, which consists of the brain (sensory vesicle) and the visceral ganglion, is exclusively derived from 10 blastomeres at the 110-cell stage. The anterior CNS is relatively complex and shows remarkable left–right asymmetry, with the lumen of the sensory vesicle, the otolith, and the ocellus on the right side, and the presumed hydrostatic pressure organ on the left side. We labeled these 10 precursor blastomeres – six in the animal hemisphere (a-line) and four in the vegetal hemisphere (A-line) – with lineage tracer, and examined the fates in swimming larvae. The clonal organization of the anterior CNS is essentially invariant among individuals, although slight variation in the clonal boundary was observed. There was no extensive mixing between descendants of each precursor. We observed no evidence of cell migration except for two neuronal cells derived from a8.25 blastomeres. The eventual fates of the bilateral blastomeres produced extensive left–right asymmetry. The results suggest that the anterior neural tube rotates in a clockwise direction when viewed from the posterior pole. Staged observation indicated that this rotation takes place during the last 5 h of embryogenesis. We describe detailed positions of descendants of each precursor blastomere. In particular, specific cells of sensory structures were identified by their morphology and staining with specific antibodies and probes. The otolith and ocellus pigment cells were derived from left and right a8.25 blastomeres. Lens cells of the ocellus have a right A8.8 origin, and most of the photoreceptor cells originated from the right A8.7. The presumed pressure organ was formed by descendants of left and right a8.19 and left a8.17 blastomeres. The description of cell lineages of the CNS would facilitate future research to analyze the mechanisms of development of the simple CNS of ascidian tadpole larvae.
Ascidian eggs develop into tadpole larvae with a dorsal central nervous system (CNS). The CNS comprises a hollow tube that is formed through neural tube formation at the neurula stage. Brain formation from animal blastomeres requires inductive cell interactions with vegetal cells as in vertebrates. A number of recent studies have focused on the mechanisms of the formation and patterning of the ascidian CNS because of its structural simplicity (reviewed by Okamura et al. 1993; Meinertzhagen & Okamura 2001; Lemaire et al. 2002). Recently, genome sequencing of the ascidian Ciona intestinalis was completed (Dehal et al. 2002). Various genes are expressed in all or in some CNS cells (e.g. Mochizuki et al. 2003). Therefore, detailed description of the process of formation of the CNS would provide substantial information for future research on the development of the ascidian CNS.
Cell lineages and developmental fates of the blastomeres in ascidians have been well documented during embryogenesis (Conklin 1905; Ortolani 1955; Nishida 1986, 1987; Nicol & Meinertzhagen 1988a, 1988b) and during metamorphosis (Hirano & Nishida 1997, 2000). Ascidian embryos exhibit a determinative cleavage pattern, and the cell lineage is essentially invariant. The invariance provides profound advantages for the study of the mechanisms of cell fate specification, because the developmental fate of each embryonic cell is predictable. Therefore, for example, gene expression in specific cells in early embryos can be precisely related to future cell fates. In the CNS, the early cell division history up to 10th generation (counted from the first cleavage) before neural tube closure has been described from observations of embryos using a scanning electron microscope (Nishida 1986; Nicol & Meinertzhagen 1988a). Nicol & Meinertzhagen (1988b) have documented further cell divisions in the CNS up to the 11th and 12th generations until completion of neural tube closure by reconstruction of serial semi-thin sections. They have also reported cell counts and maps in the CNS of hatched swimming larvae (Nicol & Meinertzhagen 1991). However, it is supposed that a few rounds of cell division after the neural-tube-closure stage are required to attain the final cell number of the larval CNS (approximately 300), and therefore detailed and eventual fates of the cells at the neural-tube-closure stage are still unknown.
Developmental fates of blastomeres of 110-cell embryos (late blastula) of H. roretzi have been traced by intracellular injection of horseradish peroxidase (HRP) as a lineage tracer molecule (Nishida 1987). At that stage, fates of most blastomeres are restricted to give rise to a single tissue type. That work showed that the larval CNS is derived from 16 blastomeres of the 110-cell embryo. Bilateral a8.17, 19, and 25 cell pairs (anterior-animal cells) exclusively contribute to most parts of the brain (or sensory vesicle). Posterior brain, visceral ganglion and tail nerve cord originate form bilateral A8.7, 8, 15, 16 (anterior-vegetal cells) and b8.19 cell pairs. By labeling of early blastomeres with HRP, cell fates can be traced at the larval stage, although the cell division history after labeling cannot be followed. Because cell division history has been well documented up to neural-tube-closure stage by other studies, the history can supplement the missing cell lineages in the cell labeling studies. Fate tracing by labeling blastomeres would be useful, as gene expression in specific cells in early embryos can be precisely related to terminal cell fates.
In our previous studies, descendant cells of labeled blastomeres were detected by observing whole-mounted larvae (Nishida 1987). Therefore, the detailed fates and positions of the labeled cells within the brain were obscure. Furthermore, the brain shows significant left–right asymmetry, whereas the other parts of the body are bilaterally symmetrical. In our previous studies, we have not paid much attention to this asymmetry of the brain. In addition, there are various sensory structures in the brain: an otolith cell senses gravity (Eakin & Kuda 1971; Tsuda et al. 2003a); the ocellus is a light sensor that consists of a pigment cell, lens cells and photoreceptor cells (Eakin & Kuda 1971; Kajiwara & Yoshida 1985; Kawakami et al. 2002; Tsuda et al. 2003a); and another has been presumed to sense hydrostatic water pressure (Eakin & Kuda 1971; Nicol & Meinertzhagen 1991). The origins of these cells, except for the otolith and ocellus pigment cells, remain unknown. Therefore, in this study we extended our previous study and labeled the CNS precursor cells at the 110-cell stage with HRP and then examined the fates and positions of the descendants in serial transverse sections of the hatched tadpole larvae. The sensory structures were identified by observing morphology, facilitated by staining with specific antibodies and RNA probes.
Observations in the previous studies and the present study will provide basic but useful information for research on the development of the ascidian CNS. Understanding of the basic and essential mechanisms revealed in the simple ascidian CNS would also contribute to the understanding of the more complicated vertebrate CNS.
Materials and Methods
Animals and embryos
Adults of the ascidian H. roretzi were purchased from fishermen near the Otsuchi Marine Research Center, Ocean Research Institute, University of Tokyo, Iwate, Japan, and near the Asamushi Marine Biological Station, Tohoku University, Aomori, Japan. Naturally spawned eggs with a diameter of 280 µm were fertilized with a suspension of non-self sperm. The relatively large size of the egg facilitates identification of each blastomere and intracellular injection of lineage tracer molecules. Embryos were cultured at 13°C. They developed into the 110-cell stage at 9 h, and tadpole larvae hatched after 35 h of development. The nomenclature of blastomeres is according to Conklin (1905). Detailed descriptions of the cell division pattern of Halocynthia at the cleavage and gastrula stages, observed by scanning electron microscopy, are available in Satoh (1979) and Nishida (1986).
Injection of HRP
Horseradish peroxidase was injected as described in Hirano & Nishida (1997, 2000). After fertilization, follicle cells were removed by treatment with 0.05% actinase E (Kaken, Tokyo, Japan) in Millipore-filtered seawater for about 5 min. After washing in seawater, defolliculated eggs within their vitelline membranes were attached to coverslips. They were cultured until the 2-, 8-, or 110-cell stage. Pressure injection and iontophoretic injection were used. Iontophoretic injection was carried out according to Nishida (1987). Microelectrodes were pulled from 1 mm outer diameter (o.d.) thin-walled capillary tubing with inner fibers (GDC-1, Narishige, Tokyo, Japan). Electrodes were filled with 4% HRP (Type 1-C, Toyobo, Osaka, Japan) and 0.5% lysine-fixable dextran tetramethylrhodamine (RDL; Fluoro Ruby, Molecular Probes, Eugene, OR, USA) dissolved in 0.2 m KCl. RDL was used for confirming by fluorescence microscope equipped with G-filters whether tracer molecules were injected into the correct blastomeres. Plasma membranes were penetrated by microelectrode using the capacitance compensation control of the amplifier to induce tip oscillation. Iontophoresis was conducted at about 13°C with a 2 s duration depolarizing current pulse of about 1.5 × 10−8 A at 1-Section intervals for 2 min.
Horseradish peroxidase histochemical staining
Horseradish peroxidase-injected larvae were fixed at 4 h after hatching with 1% glutaraldehyde in seawater for 3 h at room temperature. Fixed larvae were washed with 0.1 m Tris·HCl buffer (pH 7.6) for 5 min, transferred to 0.1 m phosphate buffer (pH 6.4), and stained for HRP with 0.06% diaminobenzidine tetrahydrochloride (Muller & McMahon 1976; Nishida 1987). Coloring was enhanced by pretreatment of fixed larvae with 0.5% CoCl2. Hydrogen peroxide was added to the solution at a final concentration of 0.01%, and the coloring reaction was stopped when enough staining was obtained (5–10 min). Specimens were cleared with xylene and photographed from dorsal, left and right sides. Then each larva was embedded in Technovit 7100 (Heraeus Kulzer GmbH, Bad Sachsa, Germany) and sectioned at 5 µm for detailed observations of the labeled cells. Sections were counterstained with 0.005% toluidine blue and observed with differential interference contrast (DIC) optics.
Immunostaining and in situ hybridization
The hydrostatic pressure organ was stained with the monoclonal antibody Hpr-1 (Darras & Nishida 2001; Akanuma et al. 2002). Larvae were fixed in methanol at -20°C. Indirect immunostaining was carried out by standard methods using HRP-conjugated secondary antibody (MAX-PO, Nichirei, Tokyo). Photoreceptor cells were detected with Hrarr RNA probe (Akanuma & Nishida 2003). Whole-mount in situ hybridization was performed by using digoxigenin (DIG)-labeled antisense probes, as described previously (Miya et al. 1997).
The CNS precursor blastomeres were labeled with HRP at the 110-cell stage. The whole larval CNS of Halocynthia is derived from 16 blastomeres of the 110-cell embryo (Fig. 1) (Nishida 1987). The cleavage pattern of ascidian embryos is bilaterally symmetrical. Bilateral a8.17, 19, and 25 cell pairs (anterior marginal cells of the animal hemisphere) exclusively give rise to most parts of the brain. Posterior brain, visceral ganglion and tail nerve cord are derived from bilateral A8.7, 8, 15, 16 (anterior marginal cells of the vegetal hemisphere) and b8.19 cell pairs. In this study, most of these cells were labeled except for the A8.15, 16, and b8.19 cell pairs (Fig. 1). This is because these cell pairs contribute only to the tail nerve cord. The tail nerve cord is very simple in its structure, just four cells in its transverse section, and mostly consists of ependymal-glial cells. It has already been shown that A8.15 and 16 cells give rise only to the lateral row, and b8.19 to the dorsal row of the tail nerve cord. Therefore, these cells were excluded from the present analysis. Figure 1 (lower part) is a schematic illustration of the dorsal view of the neural plate cells with color coding of relevant blastomeres. Left and right blastomeres are distinguished by underlining of the right blastomeres (Conklin 1905), and by the paler colors of the right blastomeres. We focused on these 10 cells that form the complex anterior part of the CNS in cranial and trunk regions. During neural tube formation, a8.19 and A8.7 cell pairs form the ventral floor of the neural tube, and the a8.25 cell pair meets at the dorsal midline from the left and right sides.
Most results are presented in the same manner, representing typical specimens (Figs 2,5,6,7 and 9). The position of labeled cells in the neural plate is indicated in the upper-right corner of each panel. The left column of each panel shows right, dorsal and left views of whole mounted larvae to show the position of labeled cells in whole CNS. The middle and right columns show sections of the same larvae. We prepared serial transverse sections, but only sections of a selected level along the anterior–posterior axis are shown. The position of each level is indicated in Fig. 2(A) (arrowheads). Level 1 (Lv.1) is the most anterior part of the brain. Lv.2 corresponds to the beginning of the cavity of the brain sensory vesicle. The cavity is invariably positioned on the right, and thus the larval CNS has remarkable left–right asymmetry. At Lv.3, an otolith pigment cell is present in the cavity. There is a pouch on the left side of the brain at this level. The wall of the pouch consists of hydrostatic pressure organ cells. Lv.4 is between the otolith and ocellus, showing the posterior part of the brain cavity. Lv.5 is the position of the ocellus showing the posterior wall of the brain cavity. Lv.6 is a more posterior region that represents the visceral ganglion of Halocynthia larvae. Unlike in Ciona, there is no narrow neck region between the brain and visceral ganglion in Halocynthia. In some figures (Figs 7 and 9), the more posterior region of the visceral ganglion is shown as Lv.7. Numbers of examined larvae are shown in Table 1. In the labeling of blastomeres of the 110-cell embryos, we examined 8 to 17-sectioned specimens for each blastomere.
Table 1. Number of specimens examined in which blastomeres were labeled with HRP at three stages
(a) Listed blastomeres were labeled with HRP at the 110-cell stage, and the descendants were detected in swimming larvae. (b) One blastomere was labeled at the 2-cell stage, then the descendants were detected at indicated stage of embryogenesis. (c) Listed blastomeres were labeled at the 8-cell stage, and the descendants were detected in swimming larvae.
In general, descendant cells of the 10 blastomeres did not mix during embryogenesis, and each blastomere gave rise to a continuous region within the larval CNS, except for the cases mentioned later. There is no indication of cell migration except for a single neuronal cell derived from a8.25 blastomere, as mentioned in the following. The relative positions and fates of descendant cells were essentially invariant among examined specimens, although there was a slight variation in the positions of the clonal boundary in some blastomeres.
Fate of each CNS precursor blastomere
All descendant cells of the left a8.19 (a8.19L) blastomeres lay on the left side of the larval brain (Fig. 2A). The blastomere gave rise to the left-ventral part of the anterior protrusion of the brain at Lv.1, and the left part of the brain vesicle at Lv.3–5 (Fig. 2B). There was no staining behind Lv.6. At Lv.3, the labeled cells formed the left wall of the pouch where the hydrostatic pressure receptor is supposed to be present. The organ has been presumed to sense hydrostatic water pressure (Eakin & Kuda 1971; Nicol & Meinertzhagen 1991), although no functional analysis has been carried out. Various authors have reported that specialized globular structures of the pressure receptor cells protrude into the brain lumen and are connected to the wall by a short stalk (Dilly 1969; Eakin & Kuda 1971; Reverberi 1979; Ohtsuki 1991). In fine morphology, this structure is almost identical to that of coronet cells of the succus vasculosus in fishes (Jansen & Flight 1969; von Harrach 1970; Rossi & Palombi 1976), although the function of coronet cells is also controversial. Nicol & Meinertzhagen (1991) reported that 19 pressure receptor cells are present in C. intestinalis. In Halocynthia, the globular structures are also present (Fig. 3C,D, arrowheads), and the cell number is estimated at 16–21 (Darras & Nishida 2001; Akanuma et al. 2002).
To confirm that a8.19L gives rise to part of the hydrostatic pressure organ, we immunostained larvae with the monoclonal antibody Hpr-1, which is specific to the specialized globular structure of the pressure organ (Darras & Nishida 2001; Akanuma et al. 2002). In whole mount, dotty staining was visible along the entire wall of a pouch on the left side of the brain vesicle (Fig. 3A, arrowheads), and in sections positive structures were observed in the pouch mainly at Lv.3, partly at Lv.4 (Fig. 3B, arrowheads). Therefore, it is plausible that a8.19L gives rise to the lateral part of the hydrostatic pressure organ.
Right a8.19 (a8.19R) blastomeres gave rise to the right-ventral part of the anterior protrusion of the brain at Lv.1, and to the left part of the brain vesicle at Lv.3–5 (Fig. 2). There was no staining behind Lv.5. At Lv.3, the descendants formed the ventral wall of the pouch in all specimens, and likely developed into pressure organ cells. The clonal organization of the anterior CNS, on the basis of the present observation of multiple specimens, is summarized in Figure 4 with color coding. The boundary between a8.19L and a8.19R descendants (red and pink areas) represents the original ventral midline of the neural tube. These results suggest that the neural tube is rotated during embryogenesis approximately 45° clockwise when viewed from the posterior.
a8.17L (Fig. 5) cells developed into cells in the anterior brain at Lv.1, into the left region of the brain vesicle at Lv.2–5, and into the middle and left regions of the anterior visceral ganglion at Lv.6 (Fig. 5A,B). At Lv.3 they invariably gave rise to the dorsal wall of the pouch, suggesting that the pressure organ is also derived from the a8.17L blastomere. The position of the edge and the size of the labeled area varied a bit among individuals at Lv.5 and 6, and hence the edge of the region is indicated with white broken lines in Figure 4.
Descendant cells of the a8.17R blastomeres gave rise to the anterior brain at Lv.1. At Lv.2 and 3, their descendants were separated into dorsal and ventral regions of the brain vesicle (Fig. 5C.D), and at Lv.4, labeled cells were observed in the ventral wall of the brain vesicle. We scarcely detected labeled cells behind Lv.5. At Lv.4, descendant cells of a8.17R were mixed with those of A8.7R (see Fig. 7D), and the position of each descendant cell varied among specimens. Hence, the region is represented with stripes of pale orange and light blue in Figure 4. This indicates that precise positions of the descendant cells are not deterministic at the clonal boundary between a8.17R and A8.7R, and that intercalation occurs between a8.17R and A8.7R derivatives to some extent.
Descendants of a8.25L and a8.25R (Fig. 6) cells meet at the dorsal midline when the neural tube closes. However, their descendants mostly formed the right side of the anterior CNS of the larvae. This result supports the view that the neural tube is rotated during later embryogenesis in a clockwise direction when viewed from the posterior.
The otolith cell senses gravity. The ocellus is a light sensor that contains a pigment cell (Eakin & Kuda 1971). The otolith and the ocellus pigment cells were, respectively, derived either from a8.25L or a8.25R blastomeres. There was no relationship between left–right origin and fate, namely otolith and ocellus pigment cells, among individuals. This observation is consistent with previous studies (Nishida 1987; Nishida & Satoh 1989). We never observed labeling of photoreceptor cells or lens cells in the ocellus, as these cells are derived from other blastomeres, as mentioned later. At Lv.1, both a8.25L and a8.25R gave rise to a cell that dorsally capped the anterior brain, and at Lv.2, both cells formed the left wall of the brain vesicle together, as indicated by stripes of dark and light yellow in Figure 4.
In addition, both a8.25L and a8.25R developed into right, anterior, posterior, dorsal and right-ventral regions of the thin wall of the brain vesicle at Lv.2–5. a8.25R tended to form the anterior part of the region, while a8.25L contributed to the posterior part, with the boundary present at Lv.3. This is especially evident when whole-mount specimens shown in Figure 6(A) and (C) are compared. Two neurons that sense movements of the otolith pigment cell are present close to the stalk of the pigment cell in the ventral floor of the sensory vesicle of several ascidian larvae (Torrence 1986; Ohtsuki 1990, 1991). These sensory neurons are also likely derived from a8.25L and/or a8.25R.
We rarely observed evidence of cell migration in this study. The only exception was descendant cells of a8.25L and a8.25R. In every case, we observed a single labeled cell present at Lv.6 that lay apart from an anterior cluster of labeled cells (Fig. 6A–D, arrows). Under heavy staining, an axon-like extension, likely to be neuronal cell, was observed posteriorly. These two left and right special cells had been previously reported (Nishida 1987), and the position of the cells gradually posteriorized as development proceeded, indicating that these cells migrate posteriorly (Nishida, unpubl. obs.).
There were no descendant cells of the A-line blastomeres in the CNS from Lv.1 to Lv.3, thus confirming that A-line cells contribute to posterior brain, visceral ganglion, and tail nerve cord (Nishida 1987) (Fig. 7). Descendant cells of the A8.7L blastomeres were distributed to the left side of the posterior brain and visceral ganglion, although the distribution varied a little among specimens. A shift of the ventral midline of the neural tube toward the left side was observed at Lv.5 and 6, as observed in the a-line.
A8.7R blastomeres gave rise to right and ventral regions from Lv.4–6. In this region, photoreceptor cells are expected to be present. The ocellus consists of a pigment cell, lens cells, and photoreceptor cells (Dilly 1961, 1964; Barnes 1971; Eakin & Kuda 1971; Kajiwara & Yoshida 1985; Ohtsuki 1991). In Ciona, there are approximately 20 photoreceptor cells (Nicol & Meinertzhagen 1991) that express arrestin mRNA and protein (Horie et al. 2002; Nakagawa et al. 2002; Tsuda et al. 2003b). These cells extend their outer segments into the space inside the cup-shaped ocellus pigment cell; the outer segments consist of photosensitive membranes and express opsin protein (Ohkuma & Tsuda 2000; Inada et al. 2003; Tsuda et al. 2003b). To visualize the photoreceptor cells, we carried out in situ hybridization using Halocynthia arrestin (Hrarr) probe (Fig. 8) (Akanuma & Nishida 2003). Hrarr-positive photoreceptor cells were present ventral to the ocellus pigment cells and surrounded the lens cells. The inner space of the cup-shaped pigment cell (containing the outer segments) was also stained. The Hrarr-positive region almost precisely matched the region derived from A8.7R blastomeres in both whole mounts and sections (compare Figs 7C,D and 8A,B). However, a clear difference was observed at Lv.7, where the Hrarr-positive cells were present in a more posterior region than the region of A8.7R descendants. We have no idea whether these cells are photoreceptor cells or not. We also confirmed the position of the photoreceptor cells using antibody of Ciona arrestin. The antibody well cross-reacted with Halocynthia arrestin. The immunostaining gave similar pattern to Hrarr in situ hybridization (data not shown). Thus, most photoreceptor cells are derived from A8.7R cells. Interestingly, lens cells were surrounded by A8.7R descendants, but never labeled.
A8.8L blastomeres developed into the left wall of the brain vesicle at Lv.4, and into the dorsal part of posterior brain and the central part of the visceral ganglion from Lv.5–7 (Fig. 9). We observed slight variability in the position at Lv.6.
A8.8R blastomeres gave rise to lens cells at Lv.5. They also formed the right side of the posterior brain and visceral ganglion with small variability. In Ciona, three oval lens cells are aligned in a single row (Eakin & Kuda 1971; Nicol & Meinertzhagen 1991). In Halocynthia, we observed two or three spherical lens cells depending on the individual. In most cases, the largest cell was present ventrally and close to pigment cells. Additionally, one or two smaller cells were observed in a more ventral position, but not aligned in a single row. In every case where A8.8R blastomeres were labeled, the cytoplasm of every lens cell was stained. This is consistent with the fact that we never observed labeling of lens cells when the blastomeres other than A8.8R were labeled.
Rotation of midline of the anterior CNS
The results mentioned in the preceding suggest that the anterior part of the neural tube rotated clockwise during later embryogenesis. To confirm this and to examine when it happens, we injected HRP into blastomeres at the 2-cell stage, and fixed embryos at various stages from 20 to 39 h to visualize the boundaries of the descendants of the left and right blastomeres (Table 1b). Outside the CNS, the boundaries always corresponded to the midline of the body, even in cranial epidermis. At 29 h, pigmentation of ocellus starts. Before and at 29 h, the boundary lay precisely at the midline in the anterior CNS (Fig. 10 left). At and after 34 h, however, the boundary had rotated clockwise through the Lv.2–6 (Fig. 10 right). Thus, rotation of the anterior CNS takes place between 29 and 34 h, just before hatching at 35 h.
Clonal boundary between a- and A-line cells along the A–P axis
To globally visualize the clonal boundary between a-line cells (derived from anterior-animal blastomeres) and A-line cells (derived from anterior-vegetal blastomeres) along the anterior–posterior axis, we labeled a4.2 and A4.1 blastomeres of 8-cell embryos (Fig. 11). Hatched larvae were fixed and sagittally sectioned so that we could easily recognize the A–P boundary (Table 1c). A couple of examples are shown in Figure 11. In general, the boundary lay at Lv.4 between otolith and ocellus pigment cells, confirming the results shown in Figure 4. However, as mentioned in the preceding a8.17R section, we observed some variability between specimens.
Figure 4 and other photos of sectioned larvae at Lv.1 show a tissue that dorsally covers the anterior protrusion of the brain at Lv.1. There are clear boundaries between this tissue and epidermis, endoderm and brain. We do not know what this tissue is or what its functions are. The presence of this tissue has been noticed also in transmission electronmicrographs of larvae (Nishida, unpubl. obs.). We saw no labeling in this area when the 10 precursor blastomeres of the anterior CNS were labeled at the 110-cell stage, or when a4.2 of the 8-cell embryos was labeled (Fig. 11 left, arrow). In contrast, this unknown tissue was stained when A4.1 blastomeres were labeled (Fig. 11 right, arrow). Therefore, this tissue originated from vegetal A-line cells other than A8.7 and 8.
Tadpole larvae of ascidians have a dorsal CNS that is formed by neural tube formation. The total number of cells in the CNS is approximately 300 (Nicol & Meinertzhagen 1991). The ascidian CNS provides a simple model for the study of the development of the nervous system in chordates (Okamura et al. 1993; Meinertzhagen & Okamura 2001; Lemaire et al. 2002). In this study, we describe a detailed analysis of the fates of the precursor cells of the anterior CNS, six in the animal hemisphere (a-line) and four in the vegetal hemisphere (A-line), of Halocynthia larvae. The anterior part of the CNS consists of the brain (sensory vesicle) and the visceral ganglion. Ten blastomeres of the 110-cell embryos were labeled, and the eventual positions and fates of their descendants were detected in swimming tadpoles (Fig. 4). The cleavage pattern is bilaterally symmetrical. However, the anterior CNS is relatively complex and shows remarkable left–right asymmetry, with the lumen of the sensory vesicle, the otolith, and the ocellus on the right side, and the presumed hydrostatic pressure organ on the left side. Therefore, we labeled each bilateral blastomere and analyzed them separately.
Position of descendants of each CNS precursor
Generally, each blastomere gave rise to a continuous region within the anterior CNS. Therefore, there is no extensive mixing of their descendants during embryogenesis. The relative positions and fates of descendant cells were essentially invariant among examined specimens, although there was a slight variation in the positions of the clonal boundary (shown by white broken lines in Fig. 4). The variation was probably caused by intercalation and slight mixing of cells that are close to clonal boundaries. There was no indication of long-distance cell migration except for two neuronal cells derived from the left and right a8.25 blastomeres.
It appears complicated at first glance to understand how the spatial arrangement of each clonal descendant shown in Figure 4 is established. However, it can be explained relatively easily if we take account of how the neural tube closes. We also have to take account of the clockwise rotation of the anterior neural tube when viewed from posterior pole, which we confirmed by labeling blastomeres at the 2-cell stage. The staged observation indicated that this rotation takes place quite late, during the last 5 h of embryogenesis. The rotation results in a shift of the original ventral midline of the neural tube, corresponding to the boundary between left and right a8.19 and A8.7, toward the left. Both a8.25 blastomeres meet at the dorsal midline when the neural tube closes, but they eventually give rise to the right portion of the brain vesicle (Fig. 4). The spatial arrangement of each clonal descendant mostly reflects the rolling up movement of the neural plate and the subsequent rotation of the neural tube. There is no intensive replacement between clonal descendants during embryogenesis. Relative positions of clones are conserved during these movements. Thus, the development of the ascidian CNS is very simple, without extensive mixing or migration of constituent cells.
The boundary between left and right descendants is shown in Figure 10. In the right panel of Figure 10 (34 h of development), it is obvious that labeled and non-labeled cells constitute opposite sides of most parts of the brain vesicle. Therefore, it is likely that the origin of the lumen of the brain vesicle is an initial hollow of the neural tube. Probably, the narrow hollow of the neural tube expands in the anterior region to form the brain vesicle.
Along the anterior–posterior axis, descendants of right and left a8.25 cells tended to occupy anterior and posterior walls, respectively, of the brain vesicle. The descendants first meet at the dorsal midline when the neural tube closes. We do not know when or how this displacement along the A–P axis takes place. The boundary between a-line and A-line cells along the A–P axis lies approximately between otolith and ocellus (Nishida 1987). This was confirmed by labeling each blastomere of the 110-cell embryos and a4.2 and A4.1 blastomeres of the 8-cell embryos. Only a8.17L violated this boundary and had some descendants behind the ocellus (Fig. 4). We observed slight variability in the position of the boundary between a-line and A-line cells within the region. The variation could be attributable to the slight mixing of cells that are close to clonal boundaries, and the manner of mixing may vary between individuals. There is no clear morphological or anatomical boundary between the region derived from a- and A-line cells. We do not know whether in this transient region of a- and A-lines, cell types formed by a- and A-lines are different, or both give rise to the same cell types together. We also do not know whether there is variability of cell fates among individual larvae, or cell fates of a- and A-line cells are constant irrespective of the ambiguity of cell positions in this region, because we do not have specific cell markers except for sensory organs.
Fates of descendants of each CNS precursor
We could not distinguish neuronal cells and glial cells. Neither could we count the number of descendant cells that were derived from each blastomere. The total number of cells in the anterior CNS is approximately 266 in Ciona (Nicol & Meinertzhagen 1991). Therefore, we estimate the average number of descendants of each blastomere of the 110-cell embryos as fewer than 30 cells. There are four or five rounds of cell divisions after the eighth generation at the 110-cell stage. Three or four rounds of cell divisions after the 110-cell stage have been described by Nishida (1986) and Nicol & Meinertzhagen (1988a, 1988b).
The eventual fates of bilateral blastomeres also showed extensive left–right asymmetry, as did the structure of the anterior CNS. Cells of sensory structures were identified by their morphology and staining with specific antibodies and probe. We confirmed that otolith and ocellus pigment cells are complementarily derived from left and right a8.25 blastomeres (Nishida 1987; Nishida & Satoh 1989). The ocellus consists of a pigment cell, lens cells and photoreceptor cells. In Ciona, there are approximately 20 photoreceptor cells (Nicol & Meinertzhagen 1991) that express arrestin (Tsuda et al. 2003b). We revealed that lens cells of the ocellus have a right A8.8 origin, and most photoreceptor cells originate from the right A8.7. The presumed pressure organ is present in the pouch of the brain vesicle on the left side and comprises approximately 20 cells in Halocynthia (Darras & Nishida 2001; Akanuma et al. 2002). The organ was formed by descendants of left and right a8.19 and left a8.17 blastomeres.
There are three motor neurons on each side of the nerve cord at the transition of trunk and tail, and these cells express Hrlim and TuNa2 genes. These motor neurons are supposed to originate from A5.2 blastomeres of the 16-cell embryo (Okada et al. 2002). Descendants of the A5.2 cells that contribute to the CNS are A8.15 and 16 cells at the 110-cell stage (Fig. 1). These cells were excluded from the present analysis. Therefore, we could not trace the lineage that gives rise to motor neurons. We noticed an unknown tissue that dorsally covered the anterior protrusion of the brain at Lv.1. There were clear boundaries between this tissue and epidermis, endoderm and brain. The present results indicate that these cells have an A-line origin, although we do not know what this tissue is or what its function is.
The cleavage and cell division patterns are bilaterally symmetrical until at least gastrulation. However, the present analysis revealed that the fates of bilateral blastomeres show left–right asymmetry. For example, lens cells and most arrestin-positive photoreceptor cells are derived only from the right blastomeres. It will be interesting to analyze how this asymmetry is established. When cleavages are permanently arrested at the 110-cell stage, the arrestin gene (Hrarr) is eventually expressed in all of the A8.7 L, A8.8 L, A8.7R, and A8.8R blastomeres (Akanuma & Nishida 2003). Therefore, cell interactions at later embryogenesis are likely involved in restriction of only A8.7R to photoreceptor cell fate. Another clue to left–right asymmetry is left-sided Nodal and Pitx expression, as in vertebrates. At the tailbud stage, HrNodal and HrPitx are expressed in the left epidermis in Halocynthia (Morokuma et al. 2002). The asymmetric expression would do something to establish asymmetry in the CNS. Thus, the description of cell lineages of the CNS would facilitate future studies that analyze the mechanisms of not only L–R asymmetry, but also of the development and patterning of the entire but relatively simple CNS of ascidian tadpole larvae.
The authors thank members of the Asamushi Marine Biological Station and the Otsuchi Marine Research Center for help in collecting live ascidian adults, and members of the Misaki Marine Biological Laboratory for help in maintaining them. The authors are also grateful to Dr G. J. Kim for Hpr-1 antibody and Dr T. Akanuma for Hrarr cDNA. This work was supported by Grants-in-Aid from MEXT and JSPS (13480245 and 13044003).