The isolated right half (RH) or left half (LH) of Xenopus embryos can undergo regulation so as to form well-proportioned larvae. To assess how the combined actions of maternal determinants and cell–cell interactions contribute to form the well-proportioned larvae, we quantitatively compared four-cell stage blastomere fate between normal larvae and regulated larvae from RH embryos. In normal larvae, the clones of the right dorsal blastomere (RD) and right ventral blastomere (RV) were located unilaterally. In contrast, in regulated larvae: (i) the RD clone exclusively occupied the anterior endomesoderm (AE) derivatives, coinciding no RV progeny in those derivatives of normal larvae. The clone bilaterally populated tissues along the dorsal midline, which characteristically included the medial regions of both somites adjoining the notochord, with higher percentages on the right and anterior sides. (ii) The RV clone extensively compensated for the missing left side at the expense of its right side contribution, and bilaterally occupied the ventroposterior and also dorsal regions excluding the AE derivatives. This clone considerably populated, with altered orientations, the derivatives of the left half gastrocoel roof plate (GRP), the left half GRP being essential for laterality determination. These results show that the high cell-autonomy in the AE constitutes a mechanism common to both normal and regulative development. In regulated larvae, cell–cell interactions shifted the midlines on the dorsal side slightly and the ventral side to a greater extent. The cell lineage difference in the left half GRP could result in a different utilization of maternal determinants in that area.
Animal development entails the serial actions of maternal determinants, transmitted to specific blastomere(s), and of cell–cell interactions. Previous studies showed that some kinds of isolated animal embryo parts, particularly in deuterostomes, can develop into complete, albeit smaller larvae (Gilbert 2010). A critically important issue in developmental biology is how the combined actions of maternal determinants and cell–cell interactions govern the formation of the completely patterned larvae in not only normal but also regulative development.
To address this issue, cell lineage analysis during regulative development is highly informative. If the fate of a blastomere is cell-autonomously determined by maternal determinants, then the progeny would contribute to the same tissues as in a normal embryo. Alternately, if the fate of a blastomere is determined by cell–cell interactions, the fate would shift during regulative development.
Embryological experiments in Xenopus laevis, which reported regulative development in the greatest detail in vertebrates, have shown that the early embryo of this animal is more mosaic than had previously been thought (Kageura & Yamana 1983, 1984; Cooke & Webber 1985; Kageura & Yamana 1986; Yamana & Kageura 1987; Kageura 1995). While there are many reports on blastomere fate during normal development (Jacobson & Hirose 1978; Hirose & Jacobson 1979; Jacobson 1985; Masho & Kubota 1986; Dale & Slack 1987; Moody 1987a,b; Moody & Kline 1990; Bauer et al. 1994; Vodicka & Gerhart 1995), cell linage analysis during regulative development has been rarely reported, and the mosaic nature or cell-autonomy of the blastomere remains elusive.
Here, we describe blastomere fate in isolated four-cell stage right half (RH) embryos during regulative development, in comparison with normal embryos. The right dorsal blastomere (RD) inherits dorsal determinants (Moon & Kimelman 1998; Gerhart 2001; Sakai 2008), while the right ventral blastomere (RV) contains very little or no activity of dorsal determinants (Kageura & Yamana 1983; Cooke & Webber 1985; Yamana & Kageura 1987; Kageura 1997), and vegetally localized determinants such as VegT and Vg1 are inherited in both blastomeres (Moon & Kimelman 1998; Gerhart 2001; Sakai 2008). The RH embryo lacks the left half of the dorsal determinants and thus has a serious perturbation of its dorsoventral development (in this paper, the organizer/contraorganizer axis is called the dorsoventral axis, based on traditional terminology). However, it can still develop into a completely patterned half-volume larva, albeit having a much more slender appearance than a normal larva (Kageura & Yamana 1983). RH embryos were analyzed and derivatives of the gastrocoel roof plate (GRP) were included in the examination, because in the early RH but not LH embryos, reversed heart tube looping occurs in the newt (Spemann & Falkenberg 1919; Takano et al. 2007), and the function on the left side of the GRP is indispensable for unilateral gene expression, but that on the right is not in Xenopus (Vick et al. 2009).
The present results show a high degree of cell-autonomy in the anterior dorsal endomesoderm (anterior endomesoderm, AE), which is triggered by the actions of maternal determinants inherited in the dorsal blastomere, during both normal and regulative development. In regulated larvae, cell–cell interactions contributed to the formation of the more posterior dorsal endomesoderm including the GRP, through probably the ventralization of the RD progeny on the left, and the dorsalization of the RV progeny on the left side and at the midline. The RV progeny extensively compensated for the missing left side. The cell lineage difference in the left half GRP could result in the different utilization of maternal determinants in that half.
Materials and methods
Isolation of right half embryos and lineage tracing
The embryos were obtained by natural mating and four-cell embryos with regular and symmetrical patterns of cleavage and pigmentation were selected (Kageura 1997). The isolation of RH embryos at the four-cell stage and microinjection of a lineage tracer were performed as in the previous methods (Koga 1999) with certain modifications, as described below. A stereoscopic microscope equipped with visible and UV light sources and Modified amphibian Ringer's solution (MR) (Larabell et al. 1996) supplemented with 50 μg/mL of gentamicin sulfate were used for the experiments. In 60% MR, RD or RV at the four-cell stage (Fig. 1a,b), or both of its daughter blastomeres at the eight-cell stage was/were injected with 9.2 or 4.6 nl of Dextran Oregon Green 488 (Invitrogen, 0.1% in H2O) using a Nanoject II (Drummond Scientific) (Fig. 1c,d). The embryos with aberrant cleavage or tracer leakage from the blastomere were discarded by the 16-cell stage. The embryos with no abnormality were successively cultured until stages (stage; St.) 9–10 (Nieuwkoop & Faber 1967), at 14°C, which facilitates normal cleavage and regulation. The medium was gradually changed to 5% MR, and embryos were further cultured at 14–23°C. The clonal distribution in the ectoderm of the living embryos was photographed at St.13–18 (late gastrula/early neurula stages) (Fig. 2). When the controls reached St. 26/27 (tail bud stage), the embryos with a normal proportion (dorsoanterior index (DAI) 5, Kao & Elinson 1988; Fig. 3a–d) were fixed (Masho 1988). They were then processed for transverse paraffin sectioning at 10 μm (Koga 1999), to enable an examination of the tissues derived from the three germ layers.
Quantitative lineage analysis
The clonal distribution in each tissue section was investigated under epifluorescence or confocal laser microscopy. At least four batches of embryos were analyzed for each blastomere of each embryo type. Sections were selected at 11 levels along the anteroposterior axis of the larva (Fig. 3e). At each level in each sample a digital image was recorded (representative images of six levels are shown in Figure 4).
At each level, a standardized drawing was made, and tissues, excluding the hypochord, which is observed as a single cell, were subdivided by symmetrically placed grids (Vodicka & Gerhart 1995) (Fig. 5). The percentages of labeled cells were estimated from the image data, with at least six (a mean of 8.2) samples per grid, and the mean percentage and standard error were calculated for each grid. In 96.6% of the grids, the total of the mean percentages of the RD and RV clones were between 80% and 120%. To reduce any potential bias in estimation due to the shape of the cells, the cell numbers in the section or other factors, the mean percentages were standardized as follows: the total of the mean percentages of both clones in a grid was adjusted to 100 in regulated larvae, and in normal larvae, the total in two symmetrically located grids in the tissue was standardized to 100 (except for 50, only in the grid for the hypochord).
The RH and whole (normal) embryos in which RD or RV was tracer-labeled were designated as W-RD, W-RV, RH-RD and RH-RV, respectively (Fig. 1c,d). The gastrula/ neurula embryos (Fig. 2) or tailbud larvae (DAI 5, Kao & Elinson 1988; Fig. 3a–d) bearing a normal appearance were investigated. Without any tracer injection, these DAI5 larvae developed at 60% or sometimes >90%, and the other morphological traits were for the most part a small head (DAI 2–4) or incomplete invagination, while larvae exhibiting a large head (DAI > 5) or no head (DAI 0, 1) were comparatively rare.
Each of the right blastomere clones populates the ectoderm on both sides, with similar dorsoventral and anteroposterior patterns to normal embryos
In the ectoderm of regulated late gastrula/early neurula embryos, the RD clone was distributed in the dorsal embryonic regions, which comprised the head epidermis and a longitudinal tissue region in the neural plate, most of which are positioned on the right side, but which also extend to the left side, with wider areas on the anterior side (Fig. 2a,c). Complementarily, the RV clone occupied the ventral/lateral regions on both sides, which are composed of the entire epidermis, excluding the head area, and also the neural plate regions along the left and right margins, with wider areas on the left and posterior sides (Fig. 2e,g). These results indicate that during the early cleavage stages, the prospectively most ventral RV descendants had been juxtaposed on the left side of the prospectively most dorsal RD descendants.
In normal embryos, most of the RD and RV clones were confined to the right side. The RD clone occupied the dorsal regions, the head epidermis and the entire neural plate excluding the outer margin in the tail, on the right side, with wider areas on the anterior side (Fig. 2b,d). The RV clone was distributed in the ventral/lateral regions on the right side, which are complementary to the RD clone regions, and also in a narrow left area of the ventral epidermis (Fig. 2f,h).
Thus, in regulated embryos, both the RD and RV clones were distributed on both sides, differing from normal embryos, with wider areas on the right and left sides, respectively. However, both clones exhibited distribution patterns similar to normal embryos along the dorsoventral and anteroposterior axes: the RD clone populated wider areas on the dorsal and anterior sides, while the RV clone occupied wider areas on the ventral and posterior sides. The observed distribution of the clones does not fit either of the two predictable most extreme blastomere fate shift modes, i.e. that the RD progeny contribute to the entire dorsal half of the embryo while the RV progeny form the ventral half if the dorsal endomesodermal tissue formation is controlled by the cell-autonomous action of dorsal and vegetally localized determinants, or that RD becomes the right half of the embryo while RV gives rise to the left half if the dorsal endomesoderm on the left is induced by the Nieuwkoop center on the right (Nieuwkoop 1969a,b; Nieuwkoop & Ubbels 1972; Gimlich & Gerhart 1984).
The anterior part of the dorsal endomesoderm exclusively originates from the dorsal blastomere
The distribution of the clones in the transverse histological sections of the tailbud stage larvae (Figs 3, 4) was quantitatively measured and reconstructed three-dimensionally (Fig. 5). This enabled comparison along the dorsoventral, anteroposterior and left-right axes between the larva types. To make it easier to follow the results, we provide only an overview of the clonal distribution in the larvae in the text, and the detailed distribution patterns in each tissue are summarized in Table S1, with figures representing the tissue and the examined sample number. The blastomere fate in regulated larvae was reproducible, but had a slightly larger variability than in normal larvae, as indicated by the standard errors for the measured progeny percentages in the grids (Table S2).
In regulated larvae, it is noteworthy that the RD progeny exclusively populated only the endoderm anterior to the first visceral pouch (Figs 4a, 5b), the most ventral endoderm region at the first visceral pouch level (Fig. 5c), the liver (Figs 4c, 5e) and the anterior tip of the notochord (Fig. 5c), which are AE derivatives (see Hausen & Riebesell 1991). This was the most anterior appearance of the distribution pattern of the RD clone: most of the RD progeny were distributed in tissue areas along the dorsal midline in the three germ layers, with wider areas and higher percentages on the anterior compared to the posterior side (Figs 4a–f, 5a–k and Table S1). This was closely similar to the distribution pattern in normal larvae along the dorsoventral and anteroposterior axes. They were located on both sides, with wider areas and higher percentages on the right side than the left, differing from normal larvae, where they were confined to the right side or midline tissues. The above distribution pattern was observed in the endoderm, somites, lateral plate, central nervous system (CNS) and epidermis. In the notochord and hypochord, this distribution pattern was observed only anteroposteriorly. Bilateral labeling throughout the tissues along the anteroposterior axis was seen in the notochord, hypochord, the areas in both somites adjoining the notochord and the ventral part of the CNS, but in the notochord and hypochord, difference of the labeling percentages was not detected between both sides at most of the levels. These tissues are the major parts of the dorsal axial structures, and the first three contain derivatives of the GRP, i.e. the ventral part of the notochord, the hypochord, and the medial ventral cells of the somites (Shook et al. 2004; see the legend for Fig. 7e,f). The labeled areas in both somites adjoining the notochord were broader in the dorsal direction than the region into which superficial cells had ingressed (Fig. 4b–f, Fig. 4 in Shook et al. 2004). Although the left area was narrower and had a smaller cell mass than the right area at each level, the labeled areas were found in all nine larvae examined throughout or almost throughout the tissue. Labeling throughout the tissues was also seen on the right side, in the dorsal endoderm underlying the hypochord, the dorsal part of the CNS and the dorsal epidermis. Bilateral labeling was detected also in the head mesenchyme and the ventral part of the body, which consisted of the head and trunk endoderm, the anterior portion of the lateral plate and the heart anlage.
In normal larvae, high RD-percentages were detected on the right side in all the AE derivatives described above, which are the endoderm areas anterior to the first visceral pouch (Fig. 5b′), the most ventral endoderm region at the first visceral pouch level (Fig. 5c′), the liver (Figs 4o, 5e′) and the anterior notochord tip (Fig. 5c′). The RD progeny were also present to a certain extent on the left. Most of the RD descendants were distributed on the right side along the dorsal midline in the three germ layers (Figs 4m–r, 5a′–k′ and Table S1). The further along the anterior side we observed, the wider the RD progeny distribution and the higher the percentages. Furthermore, bilateral labeling was observed throughout the notochord and hypochord (Figs 4m–r, 5c′–k′), as well as the ventral areas of the forebrain and midbrain, and in both retinas. Labeling throughout the tissues was seen on the right side in the dorsal region of the endoderm, the area in the somite flanking the notochord (Figs 4n–r, 5e′–k′) and the ventral region of the CNS. The breadth of the labeled right somite area was similar to the right counterpart area in regulated larvae. Thus, for the GRP derivatives, bilateral labeling was seen in the hypochord and the ventral region of the notochord, while high percentages of right side labeling and almost no left side labeling were seen in the medial regions of the somites. Relatively high percentages of right side labeling were observed in the head mesenchyme, and the ventral part of the larvae which consisted of the liver, anterior endoderm, anterior lateral plate and heart anlage.
The ventral blastomere progeny populate most of the ventral and posterior tissues derived from all of the germ layers
In regulated larvae, no RV progeny were observed in the endoderm anterior to the first visceral pouch (Figs 4g, 5m), the most ventral endoderm region at the first visceral pouch level (Fig. 5n), the liver (Figs 4i, 5p) or the anterior notochord tip (Fig. 5n). The RV clonal distribution was complementary to that of RD in each tissue. Most of the RV progeny were distributed in areas other than those along the dorsal midline, with wider areas and higher percentages in the posterior compared to the anterior regions, as well as in normal larvae, on the left side than the right, differing from normal larvae (Figs 4g–l, 5l–v and Table S1). This distribution pattern was seen in the somites, CNS and epidermis, in areas in which labeling on both sides throughout the tissues was observed, as well as in the endoderm and lateral plate, in which the labeling throughout the tissues was not observed. Considerable percentages of the RV progeny were detected throughout the tissues at the dorsal midline, consisting of the notochord excluding the anterior tip, and the hypochord (Figs 5o–u, 6a–c), which contain derivatives from the medial region of the GRP (Shook et al. 2004), unlike normal larvae. In the medial regions of the somites adjoining the notochord, to the ventral regions of which the GRP-derived cells contribute (Shook et al. 2004), the RV percentages throughout the tissue were higher on the left side of regulated larvae than on the right side of both regulated and normal larvae. In the head mesenchyme, some RV progeny were observed (Fig. 5l–o), and these were the RV progeny that had been derived from the margins of the neural plate (Fig. 2e,g).
In normal larvae, it is noteworthy that no RV progeny were detected in the endoderm at the level of the first visceral pouch (Fig. 5n′) and further to the anterior (Figs 4s, 5m′), the liver (Figs 4u, 5p′) or the notochord (Figs 4t–x, 5n′–u′, 6e,f). The RV clone for the most part occupied the major regions of the ventral tissues and also the posterior regions of some of the dorsal tissues derived from all the germ layers on the right side (Figs 4s–x, 5l′–v′ and Table S1). These regions consist of the tissue areas other than the RD progeny-rich areas located along the dorsal midline. In tissues derived from all three germ layers, the further along the posterior side we observed, the wider the distribution areas and the higher the clone percentages. This distribution pattern was seen not only in the somite, pronephric anlage, lateral plate, CNS and epidermis, in which labeling throughout the tissues was observed, but also in the endoderm, in which this labeling was not observed. In the hypochord at the trunk levels, the RV cells were observed at very low percentages (Figs 5q′,s′,t′, 6e). Thus, the GRP derivatives had certain percentages of RV progeny in the medial region of the right somite, but with very low to no percentages in the hypochord, the ventral part of the notochord and the medial region of the left somite.
The shift in the clonal distribution patterns in the tailbud larvae coincided well with that in the late gastrula/early neurula embryos, suggesting that most of the regulation had been achieved by the late gastrula stage.
The anterior endomesoderm formation exhibits a high cell-autonomy
The present results show a higher degree of cell-autonomy in the formation of the AE than of the more posterior dorsal endomesoderm in not only normal but also regulative development. In regulated tailbud larvae, the RD progeny exclusively populated the AE derivatives, which contain the liver, the most ventral endoderm region at the first visceral pouch level, all the endoderm region anterior to the first visceral pouch and the anterior notochord tip (Figs 4a,c,g,i, 5b,c,e,m,n,p). However, the more posterior dorsal endomesoderm and CNS had some RV progeny (Figs 4g–l, 5l–v). Also in normal larvae, the AE had the RD but no RV progeny (Figs 4m,o,s,u, 5b′,c′,e′,m′,n′,p′). Thus, “a high degree of cell-autonomy’’ here means that the dorsal blastomere progeny interact with themselves but not with the ventral blastomere progeny. The AE is the first embryonic part that converges and extends anteriorly on the inner wall of the blastocoel, and within it the prospective tissues are aligned in order for the above described derivatives. The anterior portion of AE passes through the future mouth in the ectoderm, eventually reaching the ventral body region, where it differentiates into the liver (see Hausen & Riebesell 1991).
The exclusive RD and no RV progeny in the liver and some parts of the pharyngeal endoderm region have not been previously reported in the isolation or recombination of either blastomeres or embryonic pieces. The exclusive RD-origination of the anterior notochord tip suggests an exclusive RD-derivation of the prechordal plate, because the prechordal plate involution precedes that of the anterior notochord (Hausen & Riebesell 1991). The exclusive RD-origination in the AE may be derived from any or some combination of the following causes. (i) The cell-autonomous histodifferentiation occurred in the RD progeny. (ii) The convergent extension was either greater or started earlier in the RD than the RV progeny. (iii) The ventralization of the RD progeny by the neighboring RV progeny (discussed in the following subsection) ultimately led the RD progeny region, which had preserved the expression of AE-specific genes, to locate in an area distant from the RV progeny.
The high cell-autonomy in the AE coincides with the model that emphasizes the cell-autonomy of the dorsal endomesoderm; the first developmental phase of the AE is determined by the joint actions of maternal Wnt/β-catenin and transforming growth factor (TGF)-β signaling (Zorn et al. 1999), or there is cell autonomous formation of a part of the Nieuwkoop center (parts 1 and 4 in Fig. 5 in Gerhart (2001)) or the organizer in the gastrula (Nagano et al. 2000; Sakai 2008). However, the results of this study do not necessarily exclude meso-endoderm induction by the vegetal dorsal cells (Nieuwkoop 1969a,b, Nieuwkoop & Ubbels 1972; Gimlich & Gerhart 1984). The present results do not seem to be adequately predicted by any of these models alone. It is unclear whether the prechordal plate and the anterior notochord tip, or the anterior endoderm are the offspring of the objective cell-autonomous region in the model of Zorn et al. (1999), or Nagano et al. (2000) and Sakai (2008), respectively. Gerhart (2001) hypothesized that the cell–cell interactions mediated by the Xnr3 protein secreted from the superficial dorsal cells (parts 1, 2 and 4) are needed for the blastula deep yolky endoderm (part 3 in his fig. 5) to form the anterior endoderm derivatives. However, no RV progeny were seen in those derivatives. This seems to be very difficult to predict. A more detailed, comprehensive and unified model seems to be needed.
The regulation of the more posterior dorsal endomesoderm including the gastrocoel roof plate
Remarkably, in the medial regions of both somites adjoining the notochord, considerable RD progeny were distributed in all nine of the regulated larvae examined throughout or almost throughout the tissue (Figs 4b–f, 5e–k and Table S1). The left side RD-distribution suggests the ventralization of certain RD progeny. The ventralization had been presumably derived from the unspecified prospective notochord, rather than the prospective right somite, had been recruited into the left somite during development, because the progeny distribution was similar to that of an orthotopically transplanted organizer fragment in which chordin function had been inhibited (Oelgeschläger et al. 2003). Most of these RD progeny in both somites might play roles in regulation as adaxial cells (Grimaldi et al. 2004; Shook et al. 2004). This is thought to be linked with their movement alongside the prospective notochord during gastrulation (Glickman et al. 2003) and their differentiation into the slow muscle cells (Devoto et al. 1996) as in zebrafish development, although this requires further study.
Dorsalization of the RV progeny was observed in the more posterior dorsal endomesoderm, including the GRP derivatives, on the left and at the midline. In the hypochord, and the notochord including its ventral region, the RV progeny were detected at considerably higher percentages than in normal larvae (Figs. 5o–v,o′–v′, 6). This result in the notochord coincides with that reported in other experimental systems (Koga et al. 1986; Stewart & Gerhart 1990; Domingo & Keller 1995). In the somites, including their medial region, and the most dorsal endoderm region, on the left, the RV progeny were observed at higher percentages than in the right counterparts of normal as well as regulated larvae (Figs 5n–u,n′–u′ and Table S1). These results, together with previous findings on the fate of the gastrocoel (the legend for Fig. 7e,f), suggest that the RV descendents, which had been almost reversed, the left-right and dorsoventral orientations, contributed to the left half of the GRP with higher percentages than the RV-percentages in the right half GRP in both normal and regulated larvae (Fig. 7c–f), and, therefore, probably also higher than the left ventral blastomere clone percentages in the left half GRP in normal larvae. This lineage difference in the left half GRP could result in a different utilization of maternal determinants, such as the cortical actin array with a counterclockwise chirality (Danilchik et al. 2006), or dorsal determinants. This might cause impairment of the indispensable function of that half in terms of inducing the unilateral gene expression (Vick et al. 2009), and, especially in the newt RH embryos separated at the two-cell stage, further might result in the reversed heart tube looping (Spemann & Falkenberg 1919; Takano et al. 2007), at a similar percentage as in the neurula stage RH embryo isolation cases (Takano et al. 2007), the RH embryo in which the left half GRP is presumably seriously damaged. This deserves further investigation.
Thus, the prospective dorsal midline in the four-cell stage RH embryos is inferred to be slightly shifted toward the right side compared with normal embryos (Fig. 7a,b, blue arrows).
The right ventral blastomere progeny extensively compensate for the missing left half
The RV progeny in regulated larvae predominately populated most of the ventral tissue areas on both sides (Figs 2e,g, 4g–l, 5l–v) and compensated for the entire left side except for the AE, with the same or higher cell-percentages compared with the right counterparts in both larva types (Figs 4, 5 and Table S1). These RV-percentage differences between the two sides were larger further along the anterior and/or dorsal regions in all tissues (Fig. 5), indicating that the compensatory expense of the right side RV-contribution was higher in the dorsoanterior areas.
The prospective ventral midline in an RH embryo would be drawn on RV, and is closer to RD than that midline in a normal embryo (Fig. 7a,b, red arrows), a shift angle of the midline toward the right side is clearly larger than the counterpart on the dorsal side (Fig. 7a,b, blue arrows). This implies that the determination of the dorsal midline takes precedence over that of the ventral midline, but is not cell-autonomous, and the ventral midline adjusts to the dorsal midline position.
The reliability of the experiments
The present normal embryos provided essentially the same results as in previous studies (Hirose & Jacobson 1979; Masho & Kubota 1986; Dale & Slack 1987; Moody 1987a; Moody & Kline 1990). The present results exclude the possibility that the examined regulated larvae developed mostly from RH embryos including more than half a dose of dorsal determinants, due to the inclination of the first cleavage plane (Black & Vincent 1988; Danilchik & Black 1988). In the absence of a tracer injection, the percentages of the complete larvae development in RH embryos were ordinarily similar to those in both RH and LH embryos reported by Kageura & Yamana (1983), and sometimes reached over 90%, but the incidence of the larvae exhibiting a large head (DAI > 5) or no head (DAI 0, 1) was rare in either case. The same lineage results were obtained for all examined samples for the AE. Additionally, in the present normal embryos, the first cleavage plane demarcates the dorsoventral axis fairly accurately (Fig. 2b,d,f,h), and most of the RD and RV progeny populated on the right side at the tailbud stage (Fig. 5a′–v′), although the ventral region of the fore-and midbrain, as well as the notochord and hypochord showed the bilateral RD-distribution due to cell mixing (Jacobson & Hirose 1978; Hirose & Jacobson 1979; Klein 1987; Cleaver et al. 2000; Shook et al. 2004). All these are similar to the results reported by Klein (1987) that the first cleavage plane demarcates the dorsoventral axis in the embryos obtained by natural mating.
The greater variation of the clonal distributions in regulated than normal larvae are presumably attributable to the increased variation in the arrangement of the early blastomere clones during the healing, as the separated embryos were put onto a flat agar plate for the microinjection in variable orientations and were deformed individually by their own weight.
The incompatible clonal distributions in the ventral regions of the fore-and midbrain between the late gastrula/early neurula and tailbud stages in both embryo types (Figs 2, 5) suggest that bilateral cell mixing occurred after the late neurula stage in both (Jacobson & Hirose 1978).
The observed shift of the blastomere fates remains to be accounted for at the cellular and molecular levels in relation to the formation of the dorsoventral axis, as well as the determination of laterality.
We are grateful to Dr Moody for the technical advice on the experiments. We thank Drs Moody, Gerhart, Grimaldi, and Takano for critical reading of the manuscript. This study was supported in part by funds (No. 086001, 116001-000) from the Central Research Institute of Fukuoka University to M. K. and H. K.