Autonomous differentiation of dorsal axial structures from an animal cap cleavage stage blastomere in Xenopus

The process by which an undifferentiated embryonic cell eventually expresses particular phenotypes probably involves both the acquisition of cytoplasmic factors from the oocyte and a variety of cellular interactions. In Xenopus one of the first steps in differentiation is the establishment of dorsal-ventral polarity, which is specified shortly after fertilization (Vincent and Gerhart, 1987; Elinson and Rowning, 1988; Wakahara, 1989). Several lines of evidence suggest that this polarity may be initiated by the localization of cytoplasmic factors or may itself cause dorsal information to be subsequently localized (see reviews of Elinson and Kao, 1989; Wakahara, 1989). For example, cortico-cytoplasmic rotation during the first cell cycle appears to be necessary for normal axis formation, and it has been postulated that this movement helps to localize or activate cytoplasmic determinants on the prospective dorsal side (Gerhart et al. 1989). Additionally, the removal of dorsal equatorial cytoplasm (Wakahara, 1986) or dorsal equatorial blastomeres (Gimlich, 1986; Takasaki, 1987) results in defective dorsal axes. Finally, transplanting dorsal cells from 8-cell (Kageura and Yamana, 1986) or from 32-cell (Gimlich, 1986; Takasaki and Konishi, 1989; Kageura, 1990) embryos to the ventral midline causes a secondary axis to form, and transplanting dorsal equatorial (32cell stage) or dorsal vegetal (64-cell stage) cells into dorsal-deficient UV-irradiated embryos restores their dorsal axis (Gimlich and Gerhart, 1984; Gimlich, 1986). Hence, a large body of information supports the idea of dorsal determination at an early stage, yet this determination has not been definitively established.

Fate maps using intracellular lineage dyes demonstrate that dorsal axial tissues, most notably notochord, descend from several blastomeres in the dorsal half of 16- and 32-cell embryos (Cooke and Webber, 1985; Dale and Slack, 1987a; Moody, 1987a,b;Takasaki and Konishi, 1989). If dorsal determinants are activated or localized at early cleavage stages, then one would expect that those blastomeres that normally make dorsal axial structures would be able to autonomously differentiate. In order to examine this possibility we used precise lineage-mapping techniques to test whether the dorsal, midline, animal blastomere (D1.1, nomenclature of Hirose and Jacobson, 1979) of the 16cell stage: (1) produces its normal clone after transplantation to a novel site; (2) is necessary for complete dorsal axis formation, and (3) differentiates into its normal progeny when cultured as an explant. The results of these experiments indicate that by the 16-cell stage D1.1 has received instructions regarding its fate either from localized cytoplasmic factors or from an inductive interaction that begins at an earlier stage. Since D1.1 occupies part of the presumptive marginal zone, we also performed these experiments with its tier-1, animal cap daughter (DI.1.1 of Jacobson and Hirose 1981=A1 of Nakamura and Kishiyama, 1971). This blastomere also autonomously differentiates, indicating that the instructions regarding dorsal fate are not confined to the marginal zone or vegetal regions at this stage.

Embryos were obtained from natural matings of adult pairs. Embryos were dejellied and selected at the 2-cell stage as previously described (Moody, 1987a). We only used embryos in which the first cleavage furrow bisected the pale equatorial crescent to reliably identify the dorsal midline of embryos derived from natural matings (Klein, 1987; Masho, 1990). Embryos were selected at the 16- and 32-cell stages only if their cleavage furrows occurred in a stereotypic radial fashion (described in detail in Moody, 1987a,b;Fig. 1). Specifically, animal hemisphere blastomeres whose third cleavage furrow extended toward the geometric equator were not used because these cells might acquire autonomy by the receipt of dorsal vegetal cytoplasm (Gimlich, 1986; London et al. 1988; Masho, 1988).

Blastomeres to be transplanted were injected with 1 nl of 10 % horseradish peroxidase (Boerhinger-Mannheim). Some embryos were incubated in Ca²⁺/Mg²⁺-free modified Stern’s solution (Nakatsuji and Johnson, 1982) with 20mM sodium citrate for 5 –10min, then transferred to an agar-coated dish containing Ca²⁺/Mg²⁺-free Stern’s solution and an unlabeled host. In other cases the operation was performed entirely in 50% Steinberg’s solution. Blastomere V2.1 was completely removed from the host with sharpened forceps, and the labeled blastomere (either D1.1 [n=14], VI.2 [n =12] or V2.1 [n=9]; Fig. 1) was teased from the donor and placed in the gap created by removing V2.1. In some experiments the transplantation was performed at the 32-cell stage. In these cases either both tier-1 (DI.1.1, n = 18) or both tier-2 (DI.1.2, n=8) daughters of D1.1 were transplanted into the position of the daughters of V2.1 (Fig. 2). Since a bilateral pair of cells was transplanted at the 32-cell stage, either bilateral pairs of tier-3 (n = 11), bilateral pairs of tier-4 (n=6), or unilateral tier 3+4 (n=9) ventral midline cells were removed to make enough room for the donor cells. Ca²⁺ and Mg²⁺ were replaced in the medium to promote healing. Embryos were fixed at stages 32 –34 (Nieuwkoop and Faber, 1967) and histologically processed to reveal the labeled clone of the transplanted cell in serial sections, as previously described (Moody, 1987a).

The descendants were quantified, as previously described (Klein and Moody, 1989) to compare the transplanted clone to that of a normal embryo. Quantitative estimates of cell fate were made by assigning a numerical value to each organ as follows: no labeled cells=0; five or fewer labeled cells=0.5; many labeled cells=1.5; an almost completely labeled organ=3.0. The contribution of a particular blastomere to each organ was determined by averaging the values from all specimens in which that blastomere had been labeled. The difference between the fate of the transplanted blastomere and the normal fate of its host position was determined by subtracting the average value of intact V2.1 from the average value of the transplanted blastomere (Table 1). Control blastomeres also were analyzed in this manner. Differences of ±1 were considered to be significant because they distinguish between ‘few’ and ‘many’ labeled progeny (Klein and Moody, 1989).

D1.1 or either of its daughters was removed either bilaterally or unilaterally. As a control, VI.1 or either of its daughters was removed in other embryos. The embryos were placed in 50% Steinberg’s solution. Late in the cleavage cycle the desired blastomere(s) was pulled from the embryo with sharpened forceps. Any embryos that failed to heal within 1 h were discarded. The remaining embryos were analyzed for gross morphological defects at stages 32 –38.

In order to test whether D1.1 or its animal cap daughter (DI.1.1) can differentiate autonomously in culture, blastomeres were isolated as described for the transplants, and grown in groups of 4 –6 on an agar bed in normal amphibian Ringer’s solution (NAM, Messenger and Warner, 1979) or in 67% L-15 medium (Rosa et al. 1988). The components of these media are defined and do not contain serum, embryo extracts or growth factors; thus, they do not contain any of the molecules known to promote dorsal differentiation (Rosa et al. 1988). After 30 –36 h (when controls reached stages 22 –25) the explants were scored for elongation. The elongated morphology indicates differentiation of dorsal axial structures (Symes and Smith, 1987; Rosa et al. 1988; Godsave et al. 1988), because in vivo elongation results from the convergence-extension movements of presumptive dorsal mesoderm (Keller and Danilchik, 1988) and the differentiation of notochord and somitic muscle (Adams et al. 1990). We categorized elongation in two groups: “+ ‘indicated obvious lengthening (length/width>1.7), and ‘±’ indicated slight lengthening (1.7>length/width>1.2). The latter group was tabulated separately in order to be confident that the elongation was caused by dorsal differentiation and was not an artifact of end-to-end healing of the group of blastomeres. Some explants were fixed in 4% paraformaldehyde, serially sectioned with a cryostat, and processed for the immunofluor-escent detection of notochord, using a keratan sulfate antibody (MZ-15) that recognizes only notochord between stages 14 –27 in Xenopus embryos (Zanetti et al. 1985; Smith and Watt, 1985). Explants also were examined histologically for the presence of differentiated cement glands.

We used the classical approach of transplanting animal hemisphere blastomeres to a novel location to test their state of determination (Slack, 1983). Two different blastomeres (D1.1 and VI.2), which normally give rise to substantial amounts of head structures and central nervous system (CNS) (Hirose and Jacobson, 1979; Moody, 1987a), were separately transplanted to the position of a ventral, vegetal midline blastomere (V2.1; Figs 1, 2), which normally produces trunk endoderm and mesoderm but very little ectoderm (Fig. 3B). These differences between donor and host fates were used to stringently test in vivo the state of determination of the transplanted cells.

Ectodermal descendants of V2.1 (the host position) normally include small numbers of cells in the trunk epidermis, trunk neural crest, and an occasional cell in the dorsal spinal cord (Fig. 3B; Moody, 1987a). Mesodermal descendants are located mostly in the dorsal and ventral thirds of the trunk somites, the nephric tubules and the lateral plate. Endodermal descendants are located throughout the hindgut and proctodeum. When V2.1 was transplanted to its original position in another embryo, as a sham operation control, its clone distributed to these same tissues (Fig. 3E). Analysis of the clone size in each organ demonstrates that a detectable difference between the clone which descended from transplanted V2.1 and that which descended from an intact V2.1 exists in only a few structures (Table 1). Therefore, the manipulation alone does not significantly alter cell fate.

Blastomere VI.2 normally contributes progeny to many head and anterior trunk ectodermal, mesodermal and endodermal structures (Fig. 3C; Moody, 1987a). When VI.2 was transplanted to the position of V2.1, its clone resembled that of its new position (compare Fig. 3F to Fig. 3B and 3C). Most of the head structures normally populated by VI.2 contained virtually no labeled cells (cement gland, olfactory placode, head epidermis, retina, lens, otocyst, forebrain, midbrain, heart, pharynx and foregut). This lack of labeled cells in the head is typical of the normal V2.1 clone. When the clone size of transplanted VI.2 in each tissue was compared to that of normal V2.1, very few differences were detected (Table 1). It was notable that transplanted VI.2 contributed fewer progeny than did normal (or transplanted) V2.1 to ventral trunk mesoderm or hindgut (Table 1). However, the overall distribution of transplanted V1.2’s progeny closely resembled that of the V2.1 clone (Fig. 3F; Table 1). Thus, the local ventral vegetal environment can influence the fate manifested by a transplanted animal blastomere.

Blastomere D1.1 normally contributes ectodermal progeny primarily to head structures and large areas of the brain. It also contributes to the entire length of the ventral spinal cord (Fig. 3A; Moody, 1987a). When it was transplanted to the V2.1 position, many of its ectodermal progeny maintained their original fate (Fig. 3D). Unlike the normal V2.1 clone, transplanted D1.1 contributed substantially to cranial ganglia, head neural crest, and all levels of the brain and spinal cord. However, the most rostral head structures (cement gland, olfactory placode, lens), which normally descend from this cell, did not receive progeny from the transplanted D1.1. The mesodermal progeny of the transplanted D1.1 substantially populated dorsal head mesenchyme, branchial arches, heart, notochord, head somites and trunk central somite (Fig. 3D, 4A; Table 1). These all are regions appropriate for the donor but not for the host blastomere (Table 1). D1.1 also contributed endodermal progeny according to its original fate, e.g. pharynx, foregut and archenteron roof. In general, the number of transplanted D1.1 progeny found in the different tissues was quite different from that of normal V2.1, and was very similar to that of normal D1.1 (Table 1). These results demonstrate that D1.1 manifests its original fate even when transplanted to the ventral vegetal midline (compare Fig. 3D to Fig. 3A and 3B).

In the majority of cases the clone from the transplanted D1.1 incorporated appropriately into the host organs, even though the blastomere was placed at the opposite pole of the embryo (Fig. 1,2). However, in 43 % of the cases partial secondary axes also developed lateral to the primary axis. The secondary axes contained both donor and host cells, and the donor cells contributed to tissues appropriate for their original D1.1 fate (e.g. notochord, branchial arch, somite, ventral neural tube, heart; Fig. 4; Table 2).

Previous studies have suggested that putative dorsal determinants extend towards the animal pole as far as tier-2 (Elinson and Kao, 1989). Therefore, D1.1, which is the mother cell to tier-1 and tier-2 dorsal midline blastomeres, may manifest its normal fate when transplanted simply because of the inclusion of its tier-2 daughter. However, fate maps show that the tier-1 daughter normally contributes to dorsal axial structures. One map shows that this cell (called Al) gives rise to 1 % of the total volume of notochord (Dale and Slack, 1987a), and another map shows that this cell produces notochord in 91 % of embryos (Moody, 1987b). In order to demonstrate which daughter is responsible for the autonomous formation of dorsal axis from a novel site we performed the transplantation with a bilateral pair of either the tier-1 daughters (D1.1.1) or the tier-2 daughters (D1.1.2; Fig. 1, 2). The transplanted DI. 1.1 clone was found in the host CNS and axial muscle in 69% of embryos in which label was detected, and in the notochord in 23 % of these embryos (Fig. 5; Table 2). The transplanted D1.1.2 clone was found in host CNS and axial muscle in 100 %, and in notochord in 40 % of embryos in which label was detected. In both experiments a secondary axis developed in about two-thirds of the embryos (Fig. 2C; Table 2). The supernumerary axes always contained neural tube and muscle (Fig. 5C), and many (42% for D1.1.1; 60% for D1. 1.2) also contained a histologically recognizable notochord (Fig. 5D; Table 2). The transplanted D1. 1.1 clone populated the secondary CNS and axial muscle in 88%, and the secondary notochord in 20% of the cases in which label was detected (Fig. 5; Table 2). The transplanted D1.1.2 clone populated the secondary CNS and axial muscle in 100 % of the cases in which label was detected; because of labeling problems in those cases containing a secondary notochord (n=3), we could not tell if the transplanted clones populated this structure (Table 2). These data demonstrate that both daughters of D1.1 are able to form dorsal axial structures autonomously after transplantation to an ectopic position.

If D1.1 is required for the formation of dorsal structures, then its removal should produce embryos with deficits in dorsal axial structures (Gimlich, 1986; Takasaki, 1987). To test this possibility, we deleted D1.1 or one of its 32-cell daughters, either unilaterally or bilaterally. Sixty-two percent of the embryos with bilateral D1.1 ablations had dorsal defects, which were defined as missing segments of spinal cord, brain, notochord and somitic muscle (Table 3). Bilateral removal of either daughter resulted in dorsal defects in similar numbers of embryos (by analysis, P>0.25). In each experiment the frequency of dorsal defects was smaller after unilateral ablations. The frequency of dorsal defects after dorsal blastomere deletion was significantly greater (P<0.05) than after ventral animal blastomere deletion. Therefore, the presence of D1.1 seems to be important for the normal formation of the entire length of the dorsal axis, and both daughters are equally important in this process.

If D1.1 already has received instructions as to its dorsal fate, then we would expect it to develop at least some dorsal axial structures autonomously in explant culture. We monitored the differentiation of explanted blastomeres by noting the extent of their elongation and the differentiation of notochord and cement gland. A ventral animal blastomere (V1.1; Fig. 1) did not elongate in defined media (Table 4). Twenty-four percent of the V1. 1 explants were scored ‘±’ but they never contained immunofluorescent notochord or differentiated cement gland (Fig. 6). A dorsal, vegetal blastomere (D2.1; Fig. 1), which has been identified as a site of putative dorsal determinants (Gimlich and Gerhart, 1984; Elinson and Kao, 1989) elongated substantially (+) in a few of the explants, and elongated slightly (±) in 19% of the explants cultured in L-15 (Table 4). This frequency was not statistically different from the response of the VI. 1 explants (χ² analysis, P>0.1), but D2.1 samples taken from both ‘+ ‘and ‘±’ categories frequently (50%) contained immunofluorescent notochord. Thus, at least some of the ‘±’ explants demonstrated dorsal differentiation, in contrast to VI.1 explants. Cement gland was not observed in any of the D2.1 explants. Dl.l elongated in both media; the frequency of elongation was significantly greater than that of either VI.1 or D2.1 (P<0.001). Although the D1.1 explants usually contained 4–6 blastomeres, two DLLs were sufficient for elongation to occur. Immunofluorescent notochord was observed frequently (43 %) in samples from both the “+ ‘and ‘±’ categories (Fig. 6), showing that at least some of the ‘±’ explants of this blastomere expressed dorsal differentiation. None of the explants sampled from the ‘category contained immunopositive notochord. Differentiated cement gland was identified in 18% of those D1.1 explants cultured in L-15 medium and in 50% of those cultured in NAM. These results demonstrate that both dorsal midline 16-cell blastomeres can autonomously differentiate, the animal cell (D1.1) significantly more so than the vegetal cell (D2.1). The ventral animal blastomere (VI. 1) showed no detectable dorsal differentiation in culture.

In order to determine whether the elongation of the dorsal animal explants was due to dorsal mesoderm being induced in the equatorial (tier-2, Dl.l.2) daughter of Dl.l, we cultured the animal cap daughter (tier-1, D1.1.1) as an explant. The explants elongated in both media, the frequency being significantly different from that of V1.1 when cultured in NAM (P<0.03) but only slightly different when cultured in L-15 (P<0.2). Notochord immunofluorescence was present in 17 % of the elongated explants sampled; this percentage is very similar to that of transplanted DI.1.1 clones that contribute to notochord (Table 2). Cement gland was detected in 40% (L-15) to 50% (NAM) of the explants. These results show that the tier-1 daughter, although presumably not a recipient of dorsal mesoderm inductive signals (Gurdon et al. 1985), can autonomously differentiate in a significant number of the explants.

These studies show that the progeny of the dorsal, midline, animal blastomere (D1.1): (1) follow their normal lineage program to populate dorsal axial structures after D1.1 is transplanted to the opposite pole of the embryo; (2) form a secondary axis in nearly half of these embryos; (3) are important for the normal formation of the entire length of the dorsal axis, and (4) autonomously differentiate in explant culture in the absence of exogenous growth factor signals. These data indicate that by the 16-cell stage this blastomere has received instructions regarding its fate and is intrinsically capable of carrying out some degree of its developmental program. The molecular nature of this early step in cellular determination of D1.1 is not known, but there are two likely candidates. D1.1 may contain dorsal determinants inherited from the egg cytoplasm, and/or it may receive instructions regarding its fate from an inductive interaction prior to the 16-cell stage.

A role for localized cytoplasmic molecules in the determination of cell fate has been suggested in numerous developing systems (reviewed in Davidson, 1986) . In Xenopus there are a few examples of localized material: (1) ‘germ plasm’ (reviewed in Ressom and Dixon, 1988); (2) maternal mRNA’s (King and Barklis, 1985; Rebagliati et al. 1985); (3) a yolk-free, RNA-rich cytoplasm in the dorsal, animal region of the fertilized egg (Herkovits and Ubbels, 1979; Imoh, 1984), which becomes concentrated, but not exclusively localized, in the dorsal animal blastomeres, and (4) a few proteins, as determined by 2-dimensional PAGE (Miyata et al. 1987) and by protein synthetic patterns (Klein and King, 1988), which are contained in different dorsal and ventral blastomeres.

One postulated function of dorsally localized cytoplasmic molecules in Xenopus is the determination of the dorsal axis (Elinson and Kao, 1989; Wakahara, 1989). Several experiments have suggested that a developmental bias to produce dorsal structures exists prior to gastrulation (Kageura and Yamana, 1983; Gurdon et al. 1985; Sharpe et al. 1987; London et al. 1988; Thomsen et al. 1990). Furthermore, the existence of unique cytoplasmic moieties has been surmised based on the observations that: (1) centrifugation of the fertilized egg results in two dorsal axes, presumably because centrifugation physically displaces a localized cytoplasmic factor (Gerhart et al. 1989); (2) prevention of cortical rotation in the egg prevents dorsal axis formation (Gerhart et al. 1989); (3) removal of dorsal equatorial cytoplasm results in dorsal axial defects (Wakahara, 1986); (4) removal of dorsal midline blastomeres results in dorsal axial defects (Kageura and Yamana, 1984; Gimlich, 1986; Takasaki, 1987; this study); (5) transplanting dorsal midline blastomeres to the ventral midline results in secondary axis formation (Kageura and Yamana, 1986; Gimlich, 1986; Takasaki and Konishi, 1989; Kageura, 1990; this study), and (6) embryos made deficient in dorsal axial structures by UV-irradiation can be rescued by implanting dorsal midline blastomeres (Gimlich and Gerhart, 1984; Gimlich, 1986). These studies have led to the proposal that ‘dorsal determinants’ may be contained in dorsal midline vegetal and equatorial cells of the 32-cell embryo (Elinson and Kao, 1989). Although dorsal tier-2 equatorial cells cannot rescue the dorsal axis of UV-irradiated hosts (Gimlich, 1986), our results and those of Kageura (1990) demonstrate that they manifest their dorsal axial fate from an ectopic ventral position.

An important result of our studies is that the tier-1 blastomere also may contain these moieties. DI. 1.1 induced secondary axes and contributed to dorsal axial structures as frequently as its tier-2 sister. Furthermore, it often elongated when placed in explant culture and produced differentiated notochord and cement gland. The frequency at which it differentiated in explant culture was lower than that of its mother cell (D1.1), but was significantly greater than that of a ventral, animal blastomere when cultured in NAM. In the intact embryo DI. 1.1 normally contributes extensively to dorsal axial structures, including notochord (Dale and Slack, 1987a; Moody, 1987b). Similarly, Kageura (1990) reported that when DI. 1.1 was transplanted to the ventral animal hemisphere it often induced a secondary axis. These studies are the first to suggest that putative ‘dorsal determinants’ also may be localized to the tier-1 dorsal, animal blastomeres of the 32-cell embryo. These results have important implications for studies of mesoderm induction, which often assume that the animal cap is fated to produce only ectoderm.

Deletion of dorsal midline blastomeres at various stages result frequently in dorsal axial defects (Kageura and Yamana, 1984; Cooke and Webber, 1985; Gimlich, 1986; Takasaki, 1987; this study). Although these studies do not prove the existence of ‘dorsal determinants’, they are consistent with that hypothesis. However, since dorsal defects are not exhibited by all of the experimental embryos, nor is the entire axis missing, either there is sufficient ‘dorsal’ information elsewhere in the embryo to compensate for the ablated cells (Koga et al. 1986; Yuge and Yamana, 1989), or the surgical procedure may physically prevent a sufficient number of dorsal mesodermal cells from invaginating. The latter possibility has been tested by implanting a ventral cell into the dorsal gap. In some cases normal embryos result (Kageura and Yamana, 1986; Koga et al. 1986), and in some cases embryos still have dorsal defects (Takasaki and Konishi, 1989). When we replaced the excised DI.1.2 with a ventral cell (VI.1.2), the transplanted clone compensated by producing CNS, notochord and axial muscle, but the embryos still had posterior axial defects (data not shown). Although these results argue that deletions are not simply causing non-specific damage, they can not distinguish whether dorsal defects are caused by the loss of ‘dorsal determinants’ or the loss of a sufficient number of cells to produce adequate dorsal axial material in response to mesoderm inductive signals.

Blastomere D1.1 may acquire the ability to differentiate autonomously from dorsal-specific mesoderm inductive signals (Smith and Slack, 1983; Gimlich and Gerhart, 1984; Dale and Slack, 1987b). However, although the 16-cell vegetal pole cells are clearly inductive, animal cap cells are not thought to be competent to respond to inductive signals until after the 64-cell stage (Jones and Woodland, 1987). Many studies of animal cap explants from cleavage stages concur with this conclusion, although their results vary according to the stage of removal and whether the culture medium contained serum (Vintemberger, 1934; Grunz, 1977; Nakamura et al. 1970; Kageura and Yamana, 1983; Shiokawa et al. 1984; Jones and Woodland, 1987; London et al. 1988; Pierce and Brothers, 1988). In contrast, our data indicate that Dl.l and its daughters may respond to mesoderm inducing signals as early as the 16-cell stage. One difference in our experiments that may account for consistent elongation and differentiation of D1.1 explants is that we cultured the equivalent of an animal cap, but it was composed of only one cell type (i.e. all D1.1 or all DI.1.1). This could affect the frequency of elongation in a few ways. Perhaps only a small proportion of D1.1 cells normally is sufficiently determined at early cleavage to elongate. In this case only a small number of animal cap explants will elongate, but D1.1 explants will elongate in proportion to the number of determined D1.1’s contained in the explant. Alternatively, D1.1’s differentiation into dorsal mesodermal tissue may depend upon a community of similarly differentiating neighbors (Gurdon, 1988). Finally, D1.1 explants might be activated nonspecifically by a factor in the medium or by the manipulation itself (e.g. Dale and Slack, 19876; Godsave and Slack, 1989). If true, however, it is significant that D1.1 is much more sensitive to this activation than its vegetal neighbor or a ventral animal cell.

We assayed the ‘dorsal’ state of transplanted blastomeres by directly observing whether members of the labeled clone integrated into central nervous system, notochord and axial muscle. Explants are more difficult to analyze because characteristic tissue organization may not be evident and cell-type specific protein markers may only be detectable in mature cells. Many studies have monitored the expression of the musclespecific cardiac actin mRNA, but this molecule is expressed in the muscle progeny of both dorsal and ventral cells. Many studies also have used the elongation of the explant as an indicator of dorsal axial structures (e.g. Symes and Smith, 1987; Rosa et al. 1988; Pierce and Brothers, 1988; Godsave et al. 1988) because in vivo elongation of embryos results from the convergence-extension movements of the gastrulating presumptive dorsal mesoderm (Keller and Danilchik, 1988) and from the elongation of the differentiating notochord (Adams et al. 1990). In addition, elongated animal caps cultured for long durations contain neural tube, notochord and somites (Pierce and Brothers, 1988). We found that elongation distinguished between dorsal and ventral explants very well (Table 4). However, because we combined several blastomeres in one explant, it was possible that they would heal end-to-end, and thus develop into an elongated structure without the aid of convergence-extension or notochord differentiation. Therefore, we tested samples from each group with an antibody (Zanètti et al. 1985) that is specific for Xenopus notochord until stage 27 (Smith and Watt, 1985). Only explants from dorsal blastomeres were immunoreactive with MZ-15;’ventral explants that elongated slightly never were immunoreactive.

Another indicator of dorsal fate is the presence of a cement gland. Recent molecular studies demonstrate that cement gland differentiation indicates the presence of dorsal mesoderm, which has induced this gland (Jamrich and Sato, 1989; Sives et al. 1989; see also Pierce and Brothers, 1988). In our experiments nearly half of D1.1 and D1.1.1 explants contained morphologically obvious cement glands that were fully functional, as indicated by secreted mucus, but none of the D2.1 or V1.1 explants contained similar glandular tissue. Since ectoderm from any region of the embryo is competent to form cement gland (see citations in Jamrich and Sato, 1989), the presence of this gland only in the dorsal explants indicates that only in these cases did dorsal mesoderm differentiate.

Many studies in which dorsal tissue or cells were transplanted into ventral regions report that dorsally derived transplants induce secondary axes (Spemann, 1967; Gimlich, 1986; Kageura and Yamana, 1986; Koga et al. 1986; Cardellini, 1988; Takasaki and Konishi, 1989; Kageura, 1990). However, there is variation between studies in the percentage of cases having secondary axes. Differences in the placement of the transplant may be one cause of the variation. In our experiments the frequency of secondary axes was the same whether the transplanted blastomere was placed only in tier-3 (82%) or only in tier-4 (83%). Differences in whether the donor cells were able to migrate into the host axis may be another cause of the variation. These cells would go unnoticed in studies that did not use an autonomous marker. In our experiments a more equatorial position allowed the transplanted clone slightly better access to the host’s dorsal axis (86% of those placed in tier-3 versus 67% of those placed in either tier-4 or tiers-3 and 4). It will be most informative to test, by following the gastrulation movements of transplanted clones, whether secondary axes form because the transplant cannot fully integrate into the host axis.

We are grateful for the helpful discussions with Drs Steven Klein and Robert Grainger, and the gift of MZ-15 antibody from Drs Fiona Watt and Jim Smith. We thank Dr Anne Warner for convincing us to use NAM for the explant cultures. This research has been supported by NIH grants NS23158 and NS20604 (S. A.M.), a National Research Service Award (NS07868) to B.C.G., and the NBD training grant at the University of Virginia (HD07323).