The cleavage stage origin of Spemann’s Organizer: analysis of the movements of blastomere clones before and during gastrulation in Xenopus

Gastrulation movements morphologically transform the amphibian embryo from a ball of cells into an elongated tadpole with distinct dorsal-ventral and anterior-posterior axes. In addition, during this time, many tissue and phenotype specifications occur, and many region-specific regulatory genes are first expressed, e.g., Xlim-1, fork head, and goosecoid (Taira et al., 1992; Dirksen and Jamrich, 1992; Blumberg et al., 1991). Identification of the developmental history of the cells occupying the various regions of the gastrula is fundamental for understanding the role of these genes and the upstream events that lead to their region-specific expression. Currently, we know the developmental fate of the early cleavage stage blastomeres (Hirose and Jacobson, 1979; Jacobson and Hirose, 1981; Jacobson, 1983; Dale and Slack, 1987; Moody, 1987a,b; Moody and Kline, 1990), and there is a detailed fate map of the early gastrula (Keller, 1975, 1976). However, there are few data that relate these two maps; we do not know where the blastomere clones are located in the gastrula. In this study, the locations of the blastomere clones prior to and during gastrulation were mapped so that the existing early and late maps could be integrated and the blastomere progenitors of regions of the gastrula that express unique gene products could be identified. Since many experiments use the cleavage stage embryo for targeting foreign gene products, and rely on changes in fate for the interpretation of gene function, the comprehensive map of blastomere clones at blastula and gastrula stages will allow investigators to target correct progenitors for gene misexpression, ‘knockout,’ and dominant negative experiments. This study demonstrates that some clones move prior to gastrulation, such that important developmental regions of the embryo change in composition over time. For example, the animal cap is used to assay the inductive capacity of exogenous molecules or ectopically expressed gene products (e.g., Slack et al., 1987; Kimelman and Kirschner, 1987; Smith, 1987; Rosa et al., 1988; Sokol and Melton, 1992; Christian et al., 1992). Investigators have used animal caps of various sizes and stages, which can lead to different results (see Dawid, 1991). Our study demonstrates that animal caps of different sizes at different stages in fact contain different clones. Another important developmental region is Spemann’s Organizer, the inducer of the nervous system. This region has been identified in the stage 10 embryo (Keller, 1976), but investigators disagree on the identity of the cleavage stage progenitors (Gimlich, 1986; Takasaki, 1987; Masho, 1988). It often has been assumed that Spemann’s Organizer region arises from the vegetal hemisphere of the cleavage stage embryo (see review Elinson and Kao, 1989) but, in fact, our initial studies (Hainski and Moody, 1992) and those of others (Takasaki, 1987; Masho, 1988) show that most of the dorsal blastopore lip actually arises from an animal hemisphere blastomere. In the present study, we detail the blastomeres (from 16and 32-cell embryos) that contribute to the Organizer region, as well as the lateral and ventral lips, and provide a developmental history of the pregastrulation movements of these important clones.

Embryos were obtained from natural matings of adult frogs that had been induced to mate with chorionic gonadotropin (Sigma). Fertilized eggs were dejellied and selected for lineage dye injections as detailed in previous reports (Moody, 1987a,b). Only embryos with stereotyped cleavage furrows (Fig. 1) were used in order to label consistently the same progenitor in all embryos. Embryos were held in Steinberg’s solution until they reached the 16or 32-cell stage.To study clones derived from the 16-cell embryo, each of the eight different blastomeres (Fig. 1A) was injected with 1 nl of 5% horseradish peroxidase (HRP, Boeringer-Mannheim). To study clones derived from the 32-cell embryo, two neighboring blastomeres were injected, one with 1 nl of 0.5% Texas Red-dextran-amine (TRDA, Molecular Probes) and the other with 2 nl of 0.5% fluorescein-dextran-amine (FDA, Molecular Probes). Only the midline blastomeres of the three animalmost tiers were examined (Fig. 1B,C). For easier reading, the nomenclature of Nakamura (Fig. 1C) is used in the text when referring to 32-cell blastomeres. However, Jacobson’s nomenclature is illustrated in Fig. 1B so that reference can be made to mother cells (Fig. 1A) and to the 32-cell fate map of Moody (1987b).

Injected embryos were raised in Steinberg’s solution, the fluorescent ones in the dark, at room temperature. Embryos were fixed, at intervals from stage 7 to stage 13 (Nieuwkoop and Faber, 1967), in 4% paraformaldehyde in 0.1 M P0₄ buffer (pH 7.4). Most of the HRPlabeled embryos were processed as whole mounts. They were washed, reacted in 3,3′-diaminobenzidine (Sigma), dehydrated, cut in half and embedded in clear plastic (Eukitt, Calibrated Instruments, Inc.). Most of the fluorescently labeled embryos were sectioned at 17 μm with a cryostat, washed and coverslipped with Tris/glycerol. For stages 710, the animal pole was identified by determining the center of the blastocoel, from its dorsal/ventral and left/right walls, and projecting a line to the surface.

To quantify the extent of movement of the B1 clone between stages 8 and 9, video images of sagittal sections were measured with a Hamamatsu Argus-10 image processor. Sections were chosen from each embryo (four at stage 8 and five at stage 9) that contained the largest area of the labeled B1 clone. The circumference of the tissue section was divided into 360° of arc. The animal-most and vegetalmost boundaries of the B1 clone were measured relative to a linear projection of the floor of the blastocoel to the surface. These boundaries were expressed in degrees of arc.

Although the appearance of the dorsal lip of the blastopore at stage 10 is commonly used as the indicator of the onset of gastrulation, Xenopus blastula cells become motile at stage 8 (Newport and Kirschner, 1982), clones have intermixed by three cell diameters by stage 9 (Wetts and Fraser, 1989) and cellular events indicative of gastrulation movements begin nearly an hour before stage 10 (Keller, 1978). We investigated both the movements of each blastomere clone and the mixing between clones before stage 10 to determine whether these movements reorganize the clones prior to the invagination at the dorsal blastopore lip.

The stage 7 clones were in the original position and wedge shape of the injected blastomere (Figs 2,3). Clones of animal blastomeres interdigitated along their edges with neighboring unlabeled cells, especially at the marginal zone border of the clone (Fig. 3). The D1.1 clone was the most intermixed (Fig. 3B), especially with the contralateral D1.1 clone (Fig. 4). Clones of vegetal blastomeres had little mixing at the borders (Figs 2,3). At stage 8 some clones began to shift positions. The A4 clone extended a few cell diameters across the geometric animal pole into the dorsal area, while the animal-most descendants of A1 receded one or two cell diameters away from the animal pole (Figs 5,6B). In addition, all of the clones mixed along their borders with neighboring clones at a depth of one or two cell diameters (Figs 5,6B).

At stage 9, several of the clones had moved from their original positions. The ventral animal clones (A4, B4) pushed toward the dorsal side, over the animal pole (Figs 6C, 7). In the blastocoel roof, these clones overlapped one another in different layers, superficial or deep. In some cases, B4-derived cells occupied the superficial layer over the A4-derived cells lining the blastocoel; in other cases, the positions of the clones were reversed. The dorsal animal clones moved vegetally toward the marginal zone (Figs 6C, 8). The animal-most descendants of A1 were nearly 45° distant from that pole and the vegetal-most descendants were within two to three cell diameters of the blastocoel floor (Fig. 6C). The B1 clone occupied the dorsal marginal zone (Fig. 8); its rostral extent was about five cell diameters animal to the floor of the blastocoel. The C1 clone was compressed toward the vegetal pole, and most of its constituents had left the dorsal marginal zone (Fig. 8). In addition, as described by Wetts and Fraser (1989) for dorsal animal clones, mixing at the boundaries between clones occurred at a depth of three to four cell diameters (Figs 6C, 7, 8).

During gastrulation, the main bodies of the clones continued to change positions as involution proceeded. At stage 10, the spread of ventral animal clones (A4, B4) over the blastocoel roof was extensive (Figs 6D, 9) and, at the midline, they covered about three-quarters of the roof. At stage 11, the A4 clone stretched across nearly the entire blastocoel roof, the B4 clone extended from the middle of the blastocoel roof into the preinvolution mesoderm close to the ventral lip, and the C4 clone was involuting (Fig. 6E). On the dorsal side, the animal clones (A1, B1) at stage 10 had condensed to fill the dorsal marginal zone and the blastopore lip (Figs 6D, 11-13). At stage 11, members of the A1 and B1 clones had involuted (Fig. 10), and members of the C1 and D1 clones that had formed the dorsal floor of the blastocoel were pulled toward the animal pole by the leading edge of the involuting mesoderm (Fig. 6E). At stage 12, these descendants of C1 and D1 formed the floor of the archenteron and the prechordal mesoderm of the head (Fig. 6F).

During gastrulation, the blastomere clones were still recognizable as discrete masses. Surface cells were mostly contiguous with members of their own clone, but deep cells began to mix extensively with neighboring clones (Fig. 6D-F). The deep cells of the dorsal clones mixed as they crowded toward the dorsal lip of the blastopore and involuted. In the stage 11 blastopore lip, for example, the descendants of A1 and B1 were well circumscribed in the preinvolution mesoderm but thoroughly mixed in the postinvolution mesoderm (Fig. 10). By the end of gastrulation, each descendant of A1 and B1 was within three cell diameters of a descendant of both A1 and B1 (Fig. 14). Deep cells of the ventral clones mixed as they spread over the embryo, and involuted at the ventral lip of the blastopore (Figs 6E-F, 9, 15).

The movements of epiboly and perhaps an asymmetric expansion of the blastocoel reorganize the relative positions of the blastomere clones, such that the constituents of the animal cap gradually change. Since the animal cap is used extensively in mesoderm induction assays, we detailed the clonal composition of these caps at different stages. At stages 7 and 8, each animal blastomere occupied nearly its original position with respect to the blastocoel (Figs 2, 3, 6B). However, at stage 8 some dorsal animal clones (D1.2 and A1) had receded slightly from the geometric pole, and some ventral animal clones (A4 and V1.2) had moved dorsolaterally a few cell diameters over the animal pole (Figs 5, 6B). Between stage 8 and 9, the B1 clone moved to the vegetal part of the dorsal marginal zone, and had very few descendants in the animal cap (Figs 6C, 8). At stage 8, the mean rostral border of the B1 clone was 24° animal to the blastocoel floor (n=4) and the mean caudal border was 4° vegetal to that border. At stage 9 the mean rostral border of the B1 clone was 9° animal to the blastocoel floor (n=5) and the mean caudal border was 26° vegetal to that border. At stage 9, the A1 clone had moved to the original position of B1, and the A4 clone stretched across the animal pole into the dorsal animal cap (Figs 6C, 7). At stage 10, the B1 clone no longer contributed to the animal cap (Figs 6D, 11). The animal portion of the dorsal marginal zone was occupied by descendants of A1 (Fig. 6D). The A4 and B4 clones constituted nearly threequarters of the animal cap at the midline (Fig. 9).

At early stage 10, the D1.1 clone constituted the majority of the dorsal blastopore lip (Fig. 16). The lateral edges of the lip consisted of descendants of D2.2. Descendants of D2.1 usually occupied the yolk plug ventral to the lip, but occasionally some of its descendants were located on the dorsal side of the lip. Labeling B1 and C1 in the same 32-cell embryo revealed that descendants of B1 populated the dorsal surface of the lip, as well as the deep cells at the site of invagination (Figs 6D, 11), while the descendants of C1 populated the yolk plug side of the lip. Most bottle cells derived from C1. Although the majority of the stage 10 dorsal lip consisted of descendants of B1, the proportion of contributing C1 descendants varied from one embryo to the next (cf. Figs 11, 12, 13). However, in all cases, the site of invagination was less than three cell diameters away from the interface between the B1 and C1 clones.

At stage 11, the D1.1 clone populated the midline dorsal lip, and descendants of D1.2 and D2.2 populated the more lateral regions (Fig. 16); the deeper layers also contained scattered cells from D2.1. It is notable that the D1.2 clone invaded the vegetal hemisphere by this stage, separating the D1.1 and D2.2 clones. Both daughters of D1.1 contributed to the stage 11 dorsal lip. The A1 clone did not reach the dorsal lip on the surface, but deep descendants extended into the preinvolution and postinvolution mesoderm (Fig. 10). While the B1 clone formed most of the postinvolution mesoderm and archenteron roof (Figs 6E, 10), we estimated that between 3 and 12% of the descendants of C1 also contributed to these structures (Fig. 6E).

At stage 12, the clones derived from the three dorsal blastomeres (D1.1, D1.2, D2.2) were compressed into the dorsal third of the blastopore lip (Fig. 16). D1.1 and D1.2 clones formed stripes in the developing neural plate, parallel to the anterior-posterior axis, and extended through the lip to contribute to mesoderm and archenteron roof. The superficial descendants of A1 just reached the blastopore lip with a few descendants intermixed with the B1 clone in the region of involution (Fig. 14). The B1 clone extended from the region of involution across the archenteron roof and throughout the dorsal axial mesoderm (Fig. 14). With the reduction in the size of the blastopore, only a few descendants of D2.1 remained in the yolk plug. Because the region of involution lies along the dorsal side of the D2.1 clone, its descendants lined the archenteron floor (Fig. 6F).

At early stage 10, the lateral edges of the lip consisted of descendants of D2.2 (Fig. 16). At stage 11, the lateral lip was composed mostly of a mixture from the lateral vegetal clones (D2.2 and V2.2) with a few cells from V1.2. In about half of the specimens, the surface cells of the V1.2 clone were at the lip and, in the rest, they were a few cell diameters away. In all cases, the deep members of the V1.2 clone were in the preinvolution mesoderm at the lip, but in only a few specimens had labeled cells involuted. At stage 12, the surface descendants of D2.2 occupied a relatively small patch in the lateral blastopore lip, while the surface descendants of V1.2 had spread to occupy nearly a quarter of the lateral surface, as viewed from the blastopore (Fig. 16).

At stage 11, the ventral lip consisted of a mixture of cells derived from V2.2 and V2.1 (Fig. 16). Only the tier-3 daughter of V2.1 (C4) contributed to the involuting zone; the tier-4 daughter (D4) formed only the yolk plug (Fig. 6E). At stage 12, the ventral lip consisted mostly of descendants of V2.1 and V2.2, the clones of which extended in both superficial and deep layers from the marginal zone through the lip. In addition, the surface descendants of V1.1 almost reached the site of involution, intermixing with surface cells from V2.1. Only the deep descendants of the vegetal daughter of V1.1 (B4) contributed cells to the involuting mesoderm (Fig. 15).

Xenopus embryos and animal caps isolated from them are commonly used to study the induction of fate changes by growth factors or genetic manipulations. Understanding the potential of a tissue in situ or in vitro requires full knowledge of its origin and the fates of its progenitors, especially since some fates are determined prior to these manipulations (e.g., Takasaki, 1987; Kageura, 1990; Gallagher et al., 1991). Our study was designed to investigate the movements of the blastomere clones between the morula and gastrulation, to evaluate the consequences of these movements for the ultimate fates of the blastomere clones, and to determine the identities of the blastomere progenitors of the animal cap and blastopore lip. We found that clones move prior to gastrulation, such that the constituents of the animal cap and dorsal marginal zone change between stages 8 and 10. These data are important for interpreting animal cap assays and fate changes resulting from genetic and other cellular manipulations.

Each blastomere of the cleavage stage embryo contributes to widely diverse tissues in the tailbud stage (Dale and Slack, 1987; Moody, 1987a,b). Because the tissues of the tailbud embryo are represented by discrete territories in stage 10 embryos (Keller, 1975, 1976), we mapped the positions of the blastomere clones onto these territories (Table 1). For the midline clones, the 32-cell data are presented and, for the lateral clones, the 16-cell data are presented. Table 1 shows where the clones are located at stage 10, in general and using Keller’s specific regional descriptions, what these regions are fated to become (Keller, 1975, 1976), and what these blastomeres are fated to become (Moody, 1987a,b). This comparison establishes the concordance of the fate maps derived from cleavage stage embryos with those derived from the early gastrula. It provides much more phenotypic detail to some regions of Keller’s map (e.g., ectodermal specializations of the head, neural crest derivatives and branchial arches), it demonstrates that Keller’s maps cannot be superimposed over the original blastomeres (Fig. 6A), and it clarifies the process by which blastomere clones move through gastrulation to achieve their eventual fates.

Study of the pregastrula using cinemicrographic techniques indicates that the apical surfaces of superficial cells in the animal and marginal regions expand between stages 7 and 9 (Keller, 1978). After stage 9, dorsal marginal cells expand even more rapidly, extending the marginal zone vegetally toward the forming blastopore invagination. As a consequence of these pregastrulation movements, blastomere clones occupy very different space in the embryo at stage 10 than they did at cleavage stages. With the formation of the blastocoel and the movements of epiboly, each blastomere clone changes shape and position to occupy a characteristic region of the gastrula that is quite different from its original position (Fig. 6; Table 1). These changes in the positions of blastomere clones were not predicted by the cinemicrography. For example, it appeared that the expansion of the animal cap was uniform (Fig. 5A; Keller, 1978), but our data show shifts of specific groups of cells (clones A4 and B4) from ventral to dorsal between stages 8 and 10, and then the spread of these groups from animal to ventral between stages 10 and 12. The expansion of the apices of superficial cells also did not predict that groups of cells (both deep and superficial members of the clone) would move en masse from animal to dorsal marginal zone regions.

A feature of these clonal movements is the mixing of cells between clones. In zebrafish, whose blastomere clones do not have predictable fates and can contribute to all regions of the embryo (Kimmel and Law, 1985), the clones mix freely during gastrulation and descendants of distant clones come in contact with one another (Warga and Kimmel, 1990). However, in Xenopus, whose blastomere clones are predictable and regionally discrete (Dale and Slack, 1987; Moody, 1987a,b), the clones mix only along their borders until just before involution. Wetts and Fraser (1989) showed that, at stage 9, dorsal animal clones interdigitated by about three cell diameters. We show that this intermixing begins as early as stage 7, prior to the onset of cellular motility (Newport and Kirschner, 1982). Mixing occurs more in animal clones than vegetal clones and more in the marginal zone. The earliest and most extensive mixing occurs across the dorsal animal midline, which predicts the bilateral origin of forebrain structures (Jacobson and Hirose, 1978; Huang and Moody, 1992, 1993). Mixing that occurs prior to stage 8 probably results from the interdigitation of cells during mitoses and permissiveness to mixing at clonal boundaries.

During epiboly the mixing of clones increases. Superficial layers remain nearly coherent, with a few interposing cells from other clones. This is consistent with the cinemicrographic observation that during epiboly cells from the superficial and deep layers do not mix (Keller, 1978). In contrast, cells from different clones in the deep layers mix extensively. In the deep layers of the blastocoel roof, this mixing is accomplished by clones sliding past one another and cells intercalating (Keller, 1980). In the deep layers of the marginal zone, the mixing is accomplished by the radial intercalation of cells in the preinvolution zone and the mediolateral intercalation during involution (Wilson and Keller, 1991) These movements cause thorough mixing of the dorsal midline clones along the anteroposterior axis. For example, the clone of A1, which involutes after the clone of B1, extends across the entire anteriorposterior extent of the dorsal axial tissues and is coextensive with descendants of B1 through all but the most anterior sensory epithelium (Fig. 14). Even after gastrulation a slow progressive mixing of clones continues, at least in the nervous system (Wetts and Fraser, 1989).

Isolated Xenopus animal caps form only atypical epidermis in culture, but can be induced to form mesoderm (Sudarwati and Nieuwkoop, 1971). Thus, they are used in a standard assay to test the effectiveness of factors to induce mesoderm. However, investigators have used large or small animal caps from various stages, sometimes with different results (Dawid, 1991). For example, it has been suggested that different laboratories observed different abilities of Xwnt-8 to induce mesoderm because of differences in explant size (Christian et al., 1992) or stage of isolation (Sokol, 1993). While it has been proposed that these differences are due to the diffusion from the vegetal hemisphere to the animal pole of an endogenous signal (Sokol, 1993), an alternative hypothesis is that the constituents of the animal cap change during pregastrula stages. Although the stage 10 fate maps show that the animal cap produces only ectodermal derivatives (Keller, 1975), the normal fates of stage 8 or 9 animal caps in the intact embryo have not been mapped. Since the 16- and 32-cell fate maps show that the animal cap blastomeres (tiers-1 and -2) contribute to all three germ layers, and transplantation studies show that dorsal animal blastomeres are determined to produce dorsal mesodermal tissues (Takasaki, 1987; Kageura, 1990; Gallagher et al., 1991), it is possible that nonectodermal cells are within the animal cap of the blastula and leave it by stage 10. These proposed changes in the constituents of the animal caps could result in changes in their competence to respond to manipulations.

In fact, our data demonstrate that the clones comprising the animal cap change after the midblastula transition. The stage 8 animal cap, whether large (including the entire blastocoel roof) or small (including only cells within 45° of the animal pole), contains the entire clone of B1. Previous experiments indicate that this blastomere can autonomously differentiate to form dorsal mesoderm (Takasaki, 1987; Kageura, 1990; Gallagher et al., 1991) and that it contains dorsal information (Elinson and Kao, 1989; Hainski and Moody, 1992). Furthermore, B1 is the primary progenitor of Spemann’s Organizer and a major progenitor of the notochord (Dale and Slack, 1987; Moody, 1987b). The presence of the B1 clone, whose dorsal mesoderm fate seems determined before stage 8, probably is a significant factor in the state of competence of the stage 8 animal cap. For example, the presence of some B1 descendants in the stage 8 cap may account for the ability of activin to induce dorsal mesoderm only from dorsal halves of the cap (Sokol and Melton, 1991).

At stage 9, the B1 clone has moved vegetally into the marginal zone. Consequently, the small stage 9 animal cap contains primarily the descendants of A4, B4 and A1. The large stage 9 animal cap, however, still includes some descendants of B1. This may result in differences in inducibility between large and small stage 9 animal caps (e.g., Christian et al., 1992; Sokol, 1993). In addition, the movement of B1 out of the animal cap correlates with the observation that mesoderm does not form in stage 8 or large stage 9 animal caps isolated from embryos injected with exogenous Xwnt-8 mRNA (Sokol, 1993). Perhaps the presence of the B1 clone inhibits the Xwnt-8 signal.

By stage 10 the B1 clone occupies the dorsal lip of the blastopore, so that a large animal cap includes only neuroectodermal (A1) and epidermal (A4, B4) progenitors, while a small cap includes only epidermal descendants of A4 and B4. Interestingly, the competence of the animal cap to form mesoderm ceases at this stage (Jones and Woodland, 1987), when all the blastomere clones known to produce mesoderm have exited from the animal cap. Thus, the different constituents of the animal cap at different stages may strongly influence the competence of the cap to be induced, and its responsiveness to the presence of exogenous gene products.

The dorsal lip of the blastopore is the Organizer of the dorsoventral axis of the embryo (Spemann and Mangold, 1924) and recently novel gene transcripts have been localized to this region (Blumberg et al., 1991; Dirksen and Jamrich, 1992; Taira et al., 1992). If the progenitor of the dorsal lip is to be targeted for genetic manipulation, one needs to specifically identify that progenitor. Generalizing the 32-cell fate map to resemble the stage 10 fate map has led to the common assumption that the dorsal lip arises from C1, the tier-3 dorsal midline cell. Although one lineage tracing study reported that the site of invagination is in the cleavage furrow between the C1 and D1 blastomeres (Gimlich, 1986), others have placed it close to the cleavage furrow between C1 and B1 (Masho, 1988; Takasaki, 1987), with the B1 clone forming most of the preinvolution mesoderm and the C1 clone forming most of the bottle cells (Takasaki, 1987). By observing many embryos in which both B1 and C1 were labeled, we concur that the clone of B1 is the primary contributor to the dorsal lip. Although C1 does have descendants in this region, especially the bottle cells, they were never the major constituents. The proportion of C1 descendants in the dorsal lip varied from embryo to embryo (Figs 11-13), probably due to variability in the location of the third cleavage furrow (see also Masho, 1988). Thus, the majority of evidence indicates that the Organizer develops mostly from a blastomere in the animal hemisphere and that, if one wants to target Spemann’s Organizer for genetic manipulations, B1 is the best candidate blastomere. Furthermore, our study demonstrates that descendants of B1 orient and drive the initial mediolateral intercalation of cells at the dorsal lip. Recent work by Shih and Keller (1992a,b) shows that the suprablastoporal endoderm can organize mesodermal behavior and induce dorsal axis formation, while immediately adjacent involuting marginal zone drives mediolateral intercalation. B1 is the primary contributor to both of these regions (Table 1).

Two important techniques for testing the function of presumed regulatory genes are to express the gene in an inappropriate location or to block the function of the endogenous product of the gene in its native location. Several studies have shown that the site of injection of mRNAs at morula stages can determine the effectiveness of the genetic manipulation. By detailing the movements of blastomere clones prior to gastrulation and relating the 32-cell and stage 10 fate maps, the effectiveness of targeted genes may be improved. For example, Steinbesser et al. (1993) recently expressed goosecoid mRNA in a tier-3 blastomere of a ventralized embryo without achieving complete rescue of the dorsal axis. The data herein demonstrate that B1, a tier-2 blastomere, normally produces most of Spemann’s Organizer, which is the normal site of goosecoid expression (Cho et al., 1991). Perhaps if goosecoid mRNA is expressed in tier-2 cells the dorsal axis would be rescued completely.

Understanding the movements of clones through gastrulation also may help explain the results of manipulations on a mechanical versus genetic basis. For example, Xwnt-8 mRNA (Smith and Harland, 1991), noggin mRNA (Smith and Harland, 1992) and a native RNA extracted from dorsal blastomeres (Hainski and Moody, 1992) are more effective at rescuing the dorsal axis of ventralized embryos or inducing a secondary axis when injected into ventral vegetal, rather than animal cells. It seems unlikely that injections of exogenous mRNA into ventral animal cells fail to effect changes in axial fate due to a lack in a signalling pathway or a lack of competence, because ventral animal cells transplanted dorsally can make axial structures (Huang and Moody, 1993), and in culture can be induced to produce dorsal mesoderm (Hainski, 1992). What is notable from this study of gastrulation movements is that the ventral animal clone (V1.1) does not involute. If a blastomere is competent, but its clone physically cannot ingress, there is little chance that an axis can be organized by exogenous gene products. This idea is supported by the observations that dorsal animal blastomeres can organize secondary dorsal axes when transplanted to ventral tier-3 or tier-4 positions (Gallagher et al., 1991), but rarely do so when transplanted to ventral tier-1 and tier-2 positions (Kageura, 1990; Huang and Moody, 1993).

We wish to thank Mrs. Lianhua Yang for her excellent technical assistance. This work was supported by NIH grant NS23158.