ABSTRACT
In amphibian gastrulae, the cell population of the organizer region of the marginal zone (MZ) establishes morphogenesis and patterning within itself and within surrounding regions of the MZ, presumptive neurecto-derm, and archenteron roof. We have tested the effects on pattern of reducing the amount of organizer region by recombining halves of Xenopus laevis late blastulae cut at different angles from the bilateral plane. When regions within 30° of the dorsal midline are excluded from recombinants, ventralized embryos develop lacking the entire anterior-posterior sequence of dorsal structures, suggesting that the organizer is only 60° wide (centered on the dorsal midline) at the late blastula stage. As more and more dorsal MZ (organizer) is included in the recombinant, progressively more anterior dorsal structures are formed. In all cases, when any dorsal structures are missing they are deleted serially from the anterior end. Thus, we suggest that the amount (lateral width) of the organizer in the MZ determines the anterior extent of dorsal development.
INTRODUCTION
Gastrulation transforms the organization of the cellularized amphibian egg (the late blastula) into the organization of the embryo, with a body axis characterized by an anterior-posterior succession of dorsal structures. This transformation entails widespread migration and repacking of cells, and many inductive interactions influencing all germ layers. In the early gastrula, the cells most responsible for this reorganization occupy the marginal zone (MZ), a torus-shaped region at the equatorial margin of the animal hemisphere; the vegetal edge of the zone is marked by the bottle cells of the blastopore, and the animal edge by the limit of involution (see Keller, 1976; and Gerhart and Keller, 1986, for the case of Xenopus). This zone arises at the margin of the hemispheres because vegetal cells of the blastula induce nearby animal hemisphere cells to gain MZ properties (Nieuwkoop, 1973). During and after gastrulation, MZ cells differentiate into mesodermal tissues, including major dorsal elements of the axis (notochord and somites) and lateroventral tissues (lateral plate and blood islands). They also give rise to archenteron roof endoderm, which eventually differentiates parts of the definitive gut. Thus, they have endo-mesodermal fates.
The MZ is well known to contain a specialized subregion, the organizer, occupying the sector that is the first to gastrulate (the dorsal lip of the blastopore) and which straddles the embryo’s prospective dorsal midline (see Spemann, 1938). The organizer can initiate the development of a complete secondary body axis when transplanted at the early gastrula stage to the prospective ventral midline of the MZ of another embryo (Spemann and Mangold, 1924). The grafted organizer induces the host to form most or all of the neural tube and somites of the secondary axis, while the grafted tissue differentiates largely to notochord. The resident organizer is presumed to play a similar role in the normal development of the primary axis.
In this paper, we utilize a novel transplantation technique (see Results) to determine the types of embryonic defects generated by amounts of organizer less than sufficient for normal development. Most studies on the organizer have analyzed its ability to induce neural structures or to dorsalize the ventral marginal zone, without discerning how varying the quantity of organizer affects anterior-posterior pattern (Slack, 1983; Hamburger, 1988). One exception is Schmidt (1936) who used anuran embryos (not including Xenopus) and noted that, when less than sufficient amounts of organizer are used, the embryos tend to be deficient in structures at the anterior end. We are particularly interested in the anterior defectiveness of such embryos because treatments of the Xenopus egg in the first cell cycle that block the first cell cycle cortico-cytoplasmic rotation (eg. UV-irradiation, cold shock, pressure, nocodazole; see Gerhart and Keller, 1986), or treatments of the early blastula that prevent the induction of dorsal mesoderm (eg. removal of dorsal vegetal cells; see Gimlich, 1986) result in identical phenotypes. The more first cell cycle rotation or dorsal mesoderm induction is inhibited, the more extensively are anterior dorsal structures missing from the final axis. In this paper, we show that the quantity of organizer in the Xenopus late blastula embryo also correlates with the anterior extent of dorsal development; as more and more organizer is surgically removed from the embryo, more anterior structures are missing from the body axis. This correlation allows us to unify our earlier observations and suggest that inhibition of first cell cycle rotation or dorsal mesoderm induction causes a reduction in the size of the organizer which then leads to the observed anterior defects.
MATERIALS AND METHODS
Eggs and embryos
Procedures for rearing adult Xenopus laevis and for the ovulation, fertilization and removal of the jelly layer of eggs have been described previously (Vincent et al. 1986). Our Ringer’s solution (abbreviated ‘R’) contains: 100 mM NaCl, 2mM KC1, 1mM MgCl2, 2mM CaCl2, 5mM sodium Hepes, brought to pH 7.1 with NaOH. Experiments were carried out at 19±2°C. Embryos were staged according to Nieuwkoop and Faber (1967). Unoperated embryos almost always form normal tadpoles (<3% are DAI grade 6 or greater; 2% are DAI grade 4 or less), see Kao and Elinson (1988) and the Results section for a description of the DAI scale). UV-irradiation of eggs was performed with a Mineralight UV light (Ultraviolet Products, San Gabriel, CA) before normalized time 0.4 of the first cell cycle (where 0.0 is fertilization and 1.0 is first cleavage) as described previously (Scharf and Gerhart, 1980; Gimlich and Gerhart, 1984).
Determination of the position of the dorsal midline of the late blastula
We chose to operate on embryos at the late blastula stage (stage 9, 1–2 h prior to blastopore lip appearance) in order to allow extensive healing before the morphogenetic movements of gastrulation begin. Since there is no reliable marker of dorsoventral orientation at the time of operation, we prepared embryos in which the presumptive dorsal midline would be at a predictable location, as follows. Eggs were placed in a plastic tissue culture dish (Falcon, 60× 15 mm) in a solution of 7% Ficoll in equal parts of R/3 (one-third strength Ringers) and 0.067 M sodium phosphate pH 7.8, and tipped 90° off the normal gravitational axis before time 0.4 of the first cell cycle. A small crystal of alkalinized Nile blue was placed on the uppermost meridian to stain a small spot (Kirschner and Hara, 1980). After first cleavage, embryos were transferred to R/4. In 92% of the unoperated normal embryos (280 eggs from 11 different frogs), the Nile blue spot was located within 15° of the center of the dorsal lip of the early gastrula. In 7% of the unoperated normal embryos, the spot was 15°-4O° from the dorsal lip, and in 1% of the cases was 40°-60° away. The centerline of the dorsal lip of the early gastrula correlates extremely well with the centerline of the neural plate, the average angle separating the two lines being only 8.6° (Danilchik and Black, 1988). Thus, our Nile blue spots accurately predict the position of the dorsal midline of the embryo, as is necessary for our experiments.
Surgery
Late blastula embryos were rested on a bed of 2% agarose (high gel temperature; Sigma) in a low calcium-magnesium modified Ringer’s (LCMR: 66mM NaCl, 1.33mM KC1, 0.33mM CaC12, 0.17mM MgC12, with 50μgml−1 gentamycin and 5mM Hepes; brought to pH7.1 with NaOH), and cut in half with eyebrow knives, along the animal-vegetal axis. UV-irradiated embryos of the same age were bisected in a similar fashion. Five minutes after bisection, cell debris was gently blown from the cut surface using a Pasteur pipet. Halves were transferred to a plastic tissue culture dish (Falcon) containing LCMR. To prevent cells from adhering to the plastic, the bottom of the dish had been previously coated with poly-HEMA (polymer type NCC, Hydron Laboratories, New Brunswick, NJ). We used hair loops and watchmaker’s forceps to place two halves together in matching animal-vegetal orientations. The recombinant was then held between the edges of two glass microscope slides also coated on the edges with poly-HEMA. After 5–15 min, recombinants had partially healed and were transferred to round-bottom agarose wells in R/6 plus 0.25 mg ml−1 BSA and 50 μg ml−’gentamycin. External healing was usually complete within 30min-l h, approximately an hour before the start of gastrulation.
Scoring of the dorsal axial development of recombinant embryos
Recombinant embryos were cultured until control siblings reached stage 41 and then were scored using the dorsoanterior index (DAI) of Kao and Elinson (1988). Details of the scoring criteria are summarized in Results.
Lineage tracing and histology
Embryos were injected with fluoresceinated lysine dextran (FDA; 10 nl of 20 mg ml−1 stock) at the one-cell stage as described previously (Gimlich and Braun, 1985). Embryos were fixed in Bouin’s fixative for 24 h, washed thoroughly in 50% ethanol-50 mM NH4OH, dehydrated through an ethanol series, and embedded in Paraplast using standard procedures. Blocks were sectioned at 10 μm thickness. Mounted sections were examined with epifluorescence optics optimized for fluorescein.
Sources of variation in the development of recombinants
For a description of the terminology used for recombinants see Fig. 1 and Results. Individual 0/UV recombinants gave DAI scores ranging from 0-5 with 28 of 61 (46%) clustering at DAI grade 3. One potential source of variation was variability in the amount of damage done from operation to operation. Indeed, recombinants of normal dorsal halves and UV-irradiated halves (90+/UV) usually failed to develop forebrain structures (DAI=4.0, n=11, see Fig. 2) indicating that the operation itself had some effect. While we were unable to address variation due to operation damage, we could discern two other sources of variation for these recombinants as well as for other classes of recombinants (i.e. 30+/UV and 15∼/UV) by analyzing 20 reciprocal pairs of 0/UV recombinants (Table 1). These potential sources of variation were as follows.
Development of reciprocal pairs of recombinant embryos derived from lateral halves of normal embryos combined each with half of a UV-irradiated embryo (0/UV recombinants). DAI1 and DAI2 indicate the dorsoanterior index grades of the two members of each reciprocal pair of recombinants

The surgical operations to produce lateral half/UV (0/UV) and 30−/UV recombinants. Late blastula embryos are cut in half through the animal-vegetal axis and recombined in the correct animal-vegetal orientation. The organizer region of the marginal zone (MZ) is cross-hatched. On the far left, the normal embryo is cut along the dorsal midline. Third from the left, the normal embryo is cut 30° away from the dorsal midline, but still through the animal-vegetal axis (position of dorsal midline indicated by dotted line). Recombinant embryos are incubated to stage 41 and scored for the anterior extent of development of the body axis. See Results for a description of the DAI scale.
The surgical operations to produce lateral half/UV (0/UV) and 30−/UV recombinants. Late blastula embryos are cut in half through the animal-vegetal axis and recombined in the correct animal-vegetal orientation. The organizer region of the marginal zone (MZ) is cross-hatched. On the far left, the normal embryo is cut along the dorsal midline. Third from the left, the normal embryo is cut 30° away from the dorsal midline, but still through the animal-vegetal axis (position of dorsal midline indicated by dotted line). Recombinant embryos are incubated to stage 41 and scored for the anterior extent of development of the body axis. See Results for a description of the DAI scale.
The amount of organizer correlates with the dorsoanterior completeness of the body axis. Normal embryos are bisected through the animal-vegetal axis at various angles from the dorsal midline (DML). Halves from normal and UV-irradiated embryos are combined as described in Fig. 1. Each square represents one operated embryo. The horizontal dotted line represents the average DAI for unoperated UV-irradiated siblings. The solid line connects average DAI scores for operated embryos. The extent of axis formation is scored on the DAI scale as depicted on the abscissa. To the right of the thin vertical line are two control experiments. 90+ or 30+ halves are combined with UV-irradiated halves to assess the extent of axis deficiency caused by the operation itself. 90+ halves should contain a full-sized organizer with the cut being far away from the organizer. The average DAI of 90+/UV recombinants is 4.0. Thus, the operation itself causes some axis deficiency. In 30+/UV recombinants, the cut is close to the side of the organizer. In these embryos the average DAI is 3.5. Thus more axis deficiency is seen with cuts and healing close to (but not within) the organizer region. Some categories of operation display a wide variation in phenotypes (eg. 0/UV recombinants give DAI scores ranging from 0-5). This variation could be due to variability in the placement of the cut, variation in the amount of damage due to the operation itself, or to natural variability in organizer size (see Materials and methods).
The amount of organizer correlates with the dorsoanterior completeness of the body axis. Normal embryos are bisected through the animal-vegetal axis at various angles from the dorsal midline (DML). Halves from normal and UV-irradiated embryos are combined as described in Fig. 1. Each square represents one operated embryo. The horizontal dotted line represents the average DAI for unoperated UV-irradiated siblings. The solid line connects average DAI scores for operated embryos. The extent of axis formation is scored on the DAI scale as depicted on the abscissa. To the right of the thin vertical line are two control experiments. 90+ or 30+ halves are combined with UV-irradiated halves to assess the extent of axis deficiency caused by the operation itself. 90+ halves should contain a full-sized organizer with the cut being far away from the organizer. The average DAI of 90+/UV recombinants is 4.0. Thus, the operation itself causes some axis deficiency. In 30+/UV recombinants, the cut is close to the side of the organizer. In these embryos the average DAI is 3.5. Thus more axis deficiency is seen with cuts and healing close to (but not within) the organizer region. Some categories of operation display a wide variation in phenotypes (eg. 0/UV recombinants give DAI scores ranging from 0-5). This variation could be due to variability in the placement of the cut, variation in the amount of damage due to the operation itself, or to natural variability in organizer size (see Materials and methods).
Variation in the placement of the cut. A cut inadvertently made 5–10° off the midline (due to an imperfect prediction of the position of the midline) would have resulted in one recombinant embryo receiving less than half an organizer and the other recombinant embryo receiving more. In such cases, members of the reciprocal pair would differ greatly in their DAI scores, one high and one low. Table 1 indicated that 25% (5/20) of the reciprocal pairs met this expectation, with a grade 4 or 5 embryo paired with one of grade 2 or less (pairs 8,14,15, 16,19; see the ‘pair difference’ column). Thus, some of the variation in the data of Fig. 2 (for the 15”/UV, 0/UV and 30+/UV entries) was due to cutting error.
Normal embryos may vary in their amount of organizer. To evaluate this, we examined the spread in the summed performance of 0/UV reciprocal pairs (see the ‘pair sum’ column in Table 1); there was a considerable spread in this sum (ranging from 4 to 9). 40% of the pairs showed a summed DAI of 7 or greater, while 20% of reciprocal pairs showed a summed DAI of 5 or less. We suggest that these latter pairs derive from embryos possessing a narrower than normal organizer, although such an organizer is probably sufficient for normal development if left intact.
RESULTS
The amount of late blastula organizer correlates with the anterior extent of dorsal axial development in the tadpole
In order to determine how a reduction in the amount of organizer affects embryonic pattern, we bisected late blastula embryos along the animal-vegetal axis at various angles to the bilateral plane, and then made full-sized recombinants from two of the various halves so that recombinant full-sized embryos contained a full sized MZ with less than normal amounts of organizer. When we needed a half that contained little or no autonomous ability for dorsoanterior development, we bisected a ‘ventralized’ late blastula, that is, one derived from an egg UV-irradiated in the first cell cycle to prevent cortico-cytoplasmic rotation (Grant and Wacaster, 1972; Malacinski et al. 1974; Chung and Malcinski, 1975; Scharf and Gerhart, 1980). Ventralized embryos are normal in terms of the number of cells in the MZ, but the entire MZ resembles the most ventral sector of the MZ of a normal embryo in that the cells engage in a ventral type of gastrulation and differentiate only ventral cell types such as blood islands and coelomic mesoderm (Malacinski et al. 1977; Scharf and Gerhart, 1980,1983; Cooke and Smith, 1987). Based on their ventral type behavior during gastrulation and their subsequent ventral differentiation, UV-irradiated embryos are presumed to contain no organizer tissue. Our method of recombining halves was advantageous to other methods of tissue transplantation or explantation because it produced fewer cut edges at which damage could occur, and it reduced exposure of tissue to the culture medium. Most importantly, the spatial context of the embryo was preserved, which was important for a coherent morphogenesis allowing well-patterned differentiation. Thus, this method allowed us to score for anterior-posterior completeness of the body axis.
A schematic diagram of two different operations is shown in Fig. 1. On the left, normal embryos were bisected through the animal and vegetal poles and along the dorsal midline, producing lateral halves also called ‘0’ (zero) halves since they were produced by cuts at 0° from the dorsal midline. On the right of the figure, normal embryos were bisected throught the animal and vegetal poles but along a plane 30° away from the dorsal midline. These cuts generated two halves: one contained the dorsal midline and is referred to as the 30+ piece, the other did not contain the dorsal midline and is referred to as the 30” piece. These pieces were combined with halves from UV-irradiated embryos to produce recombinant embryos (referred to as 30+/UV and 30∼/UV recombinants) with MZ composition different than that of a normal embryo.
Figs 2 and 3 show the results from a series of experiments designed to determine the types of embryonic patterns generated by recombinants containing less than sufficient organizer amounts. After generating and analyzing a number of recombinants, we found that the recombinants could be scored according to the dorsoanterior index (DAI) since they were indistinguishable from embryos developing from eggs blocked in the cortico-cytoplasmic rotation of the first cell cycle, the latter embryos being the type for which the index was originally devised (Malacinski et al. 1974; Scharf and Gerhart, 1980, 1983; Vincent et al. 1986; Kao and Elinson, 1988). The dorsoanterior extent of development was scored as follows: 5 - normal tadpole, 4 - embryo with two eyes close together and a reduced forehead, 3 - embryo with just one eye, 2 - embryo with no eyes but at least one otolith, 1 - embryo with only trunk and tail or just tail structures, 0 - embryo with no dorsal axial structures.
Recombinant embryos showing the correlation between amount of organizer and the anterior extent of dorsal axial development. Operations were performed at stage 9, allowed to develop until controls had reached stage 41, and photographed. (A) Typical 30−/UV recombinant contains no organizer and does not form any dorsal axial structures (DAI 0). (B) Typical 15−/UV recombinant contains approximately one-fourth an organizer and forms tail, trunk, and hindbrain (DAI 2). (C)Typical 0/UV recombinant contains half an organizer and forms tail, trunk, hindbrain and midbrain with one eye in this case (DAI 3). (D) Typical 90+/UV recombinant contains a full organizer and forms tail, trunk, hindbrain, midbrain, two eyes, and some forebrain (DAI 4). (E) A normal tadpole (DAI 5).
Recombinant embryos showing the correlation between amount of organizer and the anterior extent of dorsal axial development. Operations were performed at stage 9, allowed to develop until controls had reached stage 41, and photographed. (A) Typical 30−/UV recombinant contains no organizer and does not form any dorsal axial structures (DAI 0). (B) Typical 15−/UV recombinant contains approximately one-fourth an organizer and forms tail, trunk, and hindbrain (DAI 2). (C)Typical 0/UV recombinant contains half an organizer and forms tail, trunk, hindbrain and midbrain with one eye in this case (DAI 3). (D) Typical 90+/UV recombinant contains a full organizer and forms tail, trunk, hindbrain, midbrain, two eyes, and some forebrain (DAI 4). (E) A normal tadpole (DAI 5).
90−/UV recombinants developed as embryos entirely lacking dorsal structures, or at best containing the most posterior dorsal structures such as the tail (see leftmost column of Fig. 2). 90∼/UV recombinants (average DAI of 0.2) showed no more dorsoanterior development than sibling UV-irradiated control embryos (average DAI of 0.4) and hence we concluded that the 90” half provided no organizer material. Similar results were obtained with 60−/UV and 45−/ UV recombinants. Even 30”/UV recombinants with an average DAI of 0.4, showed no more dorsoanterior development than sibling UV-irradiated controls (see Fig. 3A). Either the organizer does not extend beyond 30°, or the amount beyond 30° is too little to exert an effect on embryonic pattern in this assay. We therefore suggest that the organizer is restricted to a domain 60° wide, centered on the prospective dorsal midline in the stage 9 blastula.
15−/UV recombinants (see Figs 2,3B) were the first to show significant dorsoanterior development; these were bilateral in their organization and possessed body axes with tails and trunks complete on average to the hindbrain level (DAI=2.2), well above the average DAI of 30−/UV recombinants or UV-irradiated controls. Thus, even with approximately one-fourth of an organizer (15°), there was extensive dorsoanterior development. The variation in the dorsoanterior development of 15−/UV recombinants was large (individual DAI scores range from 0-4). Possible sources of variation are discussed in Materials and methods.
The 0/UV recombinants (containing a half organizer from a normal embryo) developed quite well giving an average DAI of 3.1 for the population, with 46 of 61 (75%) possessing body axes complete at least to the eye level (see Figs 2,3C). The 30+/UV recombinants, containing a full-sized organizer, developed only slightly better giving an average DAI of 3.5. Finally, as a control for the operation, we made 90+/UV recombinants and these gave an average DAI of 4.0 (see Figs 2,3D). This last case suggested that the operation itself reduced axial development by approximately one DAI grade even when the cut was made at a 60° distance from the organizer boundary. Thus, the operation set an upper limit on the dorsoanterior development of recombinants. On the whole, the results of Figs 2 and 3 clearly demonstrate that, as the amount of organizer is increased in the late blastula recombinant, the extent of anterior development of dorsal structures also increases. This is the first demonstration of this correlation.
The organizer does not exhibit regional medio-lateral pattern at stage 9
It is possible that the organizer may possess a detailed medio-lateral pattern at stage 9 (late blastula), and that the range of phenotypes seen in the recombinant embryos described in this paper may be due, not to the amount of late blastula organizer, but to the inclusion or exclusion of specific types of organizer tissue in the different recombinants. For instance, posterior organizing cells might reside in the lateral flanks of the organizer (15–30° from the dorsal midline) while anterior organizing cells might reside close to the dorsal midline (0–15°). Therefore, the different recombinants (i.e. 15”/UV, 0/UV) might show progressively more dorsoanterior development due to the quality of organizer retained in the normal half. In order to test this assertion, we compared recombinants containing equal quantities of only medial or lateral parts of the organizer. First, we made 15−/15− recombinants. These contained two juxtaposed lateral flanks of organizer each contributing approximately one-fourth of an organizer for a total of one-half an organizer (from the lateral regions exclusively). Second, we made recombinants where a 30° wedge centered on the dorsal midline of a normal embryo was removed and placed into a UV-irradiated embryo in the correct animal-vegetal orientation. These recombinants contained approximately one-half an organizer containing only the medial region of the organizer. If the medial region of the organizer is qualitatively different than the lateral regions, the ‘medial 30°/UV’ recombinants should develop anterior but not posterior structures while the 15−/15− recombinants should develop posterior but not anterior structures. However, we find that the 15−/15− and ‘medial 30°/UV’ recombinants gave the same range of embryonic axes, were all scorable on the DAI scale, with equivalent average DAI’s (2.6 and 2.7 respectively, see Table 2). These data strongly favor the interpretation that the quantity (the lateral width), and not quality, of the stage 9 organizer correlates with the anterior extent of dorsal development.
UV-irradiated tissue in 0/UV recombinants contributed to half the neural tube, half the somitic mesoderm and, in most cases, to a portion of the notochord
0/UV recombinants contained only half an organizer yet they often formed full-sized normal tadpoles with a complete set of dorsoanterior structures. Often both members of the reciprocal pair of recombinants prepared from one bisected normal embryo gave significant axial development (Fig. 4A), and these invariably exhibited bilateral symmetry (see Fig. 4B). The UV-irradiated half of the recombinant contributed an entire side to the bilateral embryo as shown in Fig. 4C, a trunk-level transverse section taken from a recombinant in which the UV-irradiated half had been labelled with a cell autonomous fluorescent lineage tracer (FDA; Gimlich and Braun, 1985). Note that the UV-irradiated half gave rise to a number of differentiated dorsal elements, namely, to somites on one side and half the neural tube. In many cases (11 of 13), the notochord derived from both the non-UV and the UV-irradiated tissue, with the UV-irradiated tissue contributing on average 11% of the notochord (as estimated by counting vacuoles surrounded by labelled or unlabelled cytoplasm in transverse sections). Thus, even at the late blastula stage, UV-irradiated tissue, which lacked any autonomous ability for axial development on its own, was induced to form somites, neural tube and even some notochord, if surgically combined with an organizer. In addition, dorsal axial structures from the UV-irradiated half, such as eyes, otoliths, somites,, neural tube and notochord, were arranged in the proper orientation to one another. Therefore, the half organizer from the normal half has the ability to establish both anterior-posterior as well as dorsoventral pattern within the UV-irradiated tissue.
The development of recombinants containing a lateral half of a normal late blastula and half a UV-irradiated embryo of the same age (0/UV recombinants). (A) The two tadpoles at the top are a reciprocal pair of recombinants. Both embryos form extensive dorsoanterior axial structures (DAI grade 4), almost to the level of the unoperated control shown below. (B) A dorsal view of a recombinant embryo reveals bilateral symmetry. All 0/UV recombinants are bilaterally symmetric. (C) A trunk level section of a recombinant where the UV-irradiated tissue was labelled with FDA. Note that all the somite material on one side, as well as half the neural tube and a portion of the notochord (two of five notochord vacuoles in this section) are derived from the UV-irradiated tissue.
The development of recombinants containing a lateral half of a normal late blastula and half a UV-irradiated embryo of the same age (0/UV recombinants). (A) The two tadpoles at the top are a reciprocal pair of recombinants. Both embryos form extensive dorsoanterior axial structures (DAI grade 4), almost to the level of the unoperated control shown below. (B) A dorsal view of a recombinant embryo reveals bilateral symmetry. All 0/UV recombinants are bilaterally symmetric. (C) A trunk level section of a recombinant where the UV-irradiated tissue was labelled with FDA. Note that all the somite material on one side, as well as half the neural tube and a portion of the notochord (two of five notochord vacuoles in this section) are derived from the UV-irradiated tissue.
The organizer expands between stage 9 and stage 10
All operations mentioned thus far were performed at stage 9 (late blastula). In order to determine if the organizer changes in size between stage 9 and stage 10 (early gastrula) we performed 30−/UV and 45−/UV recombinants at stage 10 and found that 30−/UV recombinants showed some dorsoanterior development (average DAI of 1.6, n=12), while 45−/UV recombinants showed no more dorsoanterior development than UV-irradiated controls (average DAI of 0.3, n=16). Thus the organizer expands between stage 9 (when it is ⩽60° wide) and stage 10 (when it is ⩽90° wide). This is in agreement with other estimates of organizer width for the early gastrula (Dale and Slack, 1987).
DISCUSSION
The anterior extent of dorsal development of the embryonic axis depends on the amount of organizer We have surgically constructed embryos to contain a full-sized marginal zone (MZ) with reduced amounts of the organizer region and have assessed their subsequent anterior dorsal axial development. If we exclude MZ material within 30° of either side of the prospective dorsal midline, the recombinant embryos develop no dorsal axial structures. Thus, we conclude that the organizer has a width of 60° or less in the normal stage 9 embryo (late blastula). When a quarter of an organizer (15°) is included, the subsequent axis is complete to the hindbrain-ear vesicle level. When a half organizer (30°) is included, the axis is complete to the midbrain-eye level. Importantly, whether the half organizer is composed of cells solely from either the medial region of the organizer or from the lateral region, the recombinant embryos develop equivalent axes, posteriorly complete to the midbrain-eye level. Finally, a full organizer (60°) gives even more anterior development often including forebrain. Therefore, we conclude that the anterior extent of dorsal development of the body axis depends on the quantity (lateral width) of the organizer in the late blastula. (Our results argue against qualitative differences in the organizer in the medio-lateral direction. They do not concern animal-vegetal differences within the organizer, a subject we are currently studying).
Two early events probably linked to the establishment of the organizer size in the late blastula are cortico-cytoplasmic rotation during the first cell cycle, and dorsal mesoderm induction during the blastula stages (Ancel and Vintemberger, 1948; Elinson, 1980; Vincent et al. 1986; Vincent and Gerhart, 1987; Nieuw-koop, 1973; Gimlich and Gerhart, 1984). If either of these events is inhibited, embryos are deficient in dorsoanterior structures (Scharf and Gerhart, 1980; 1983; Gimlich, 1986). In fact, the amount of rotation correlates with the anterior extent of development: a small amount of rotation leads to tail, while more and more rotation leads to trunk and head development (Vincent and Gerhart, 1987). In this paper, we show that our recombinant embryos not only display the same dorsoanterior deficient phenotypes, but also that the anterior extent of dorsal development is dependent on the amount of organizer present in the late blastula. We propose that reduction in the amount of first cell cycle rotation results in proportional reduction in the strength of dorsal mesoderm induction in the blastula which then results in a proportional reduction in organizer size.
How does organizer size limit the anterior extent of development?
How does a reduction in the quantity of organizer tissue leads to the qualitative effect of serial deletion of dorsal axial structures from the anterior end? Models for anterior-posterior patterning of neural structures in amphibians usually invoke two signals, one which ‘neuralizes’ ectodermal tissue, and one which ‘cauda-lizes’ this tissue from the posterior end (see Hamburger, 1988 for review). If neuralized tissue is far from the caudalizing source, it will develop as anterior structures such as forebrain and eye, while tissue that is closer to the caudalizing source will develop as more posterior structures. The organizer cell population lengthens in the anterior-posterior direction during gastrulation and neurulation in a process referred to as convergent extension (Keller et al. 1985; Keller and Danilchik, 1988; Wilson et al. 1989). We speculate that the size of the organizer determines the amount of convergent extension, which in turn determines how far tissue can escape from the caudalizing signal at the posterior end. In support of this speculation, when trypan blue (an inhibitor of convergent extension) is injected into the blastocoel early in gastrulation, the resulting embryos possess only posterior structures such as trunk and tail. As trypan blue is injected later and later in gastrulation, increasingly more anterior structures can be formed (Gerhart et al. 1984; Danilchik et al. 1990; Doniach et al. 1990). Thus, in normal untreated embryos, the formation of more and more anterior structures may depend on the cumulative amount of convergent extension, which in turn may depend on the amount of organizer present.
Analysis of 0/ UV recombinants
The 0/UV recombinants described in this paper demonstrate capabilities of the organizer not previously tested in any amphibian embryo. In our 0/UV recombinants, the half organizer must establish anterior-posterior and dorsal-ventral pattern in twice as much non-organizer tissue as was present in classical experiments performed with newt embryos where a half organizer was shown to be able to pattern a half-sized or smaller embryonic axis (Spemann and Falkenberg, 1919; Mayer, 1935). Our experiments extend classical results by showing that a Xenopus half organizer is often sufficient for the development of a full-sized tadpole with the complete anterior-posterior set of dorsal structures, thus establishing a lower limit for the amount of organizer necessary for the development of a full-sized tadpole. As a test of the upper limit, Cooke (1972) transplanted dorsal lips and showed that fullsized embryos containing two adjacent organizers often develop normally. When all experiments are considered, it appears that the organizer can vary between one-half and twice its normal size and still lead to normal development of a full-sized tadpole.
In addition to having the complete anterior-posterior set of dorsal structures, 0/UV recombinants containing either a right half or a left half of an organizer develop into bilaterally symmetric tadpoles. Similarly, bilateral secondary axes often form when, in urodeles, a left or right half of an early gastrula dorsal lip is transplanted to the ventral side of a host of the same age (Mayer, 1935). These results show that the late blastula or early gastrula organizer does not serve as a source of bilateral axial symmetry. This symmetry seems to originate from interactions occurring after the time of surgery. During gastrulation, the organizer cell population converges and extends, and induces neighboring non-organizer cells, which are arranged bilaterally to the organizer in the MZ, to alter their morphogenetic movements and to develop as dorsal mesoderm. Thus, bilateral development depends at least in part on the bilateral disposition of non-organizer cells responsive to the organizer (see Gerhart, 1980). Spemann and Falkenberg (1919) used a hair loop to constrict and subsequently divide urodele gastrulae along the bilateral plane producing two lateral or ‘0’ halves, each of which was allowed to heal along its cut edge. The imperfect bilateral symmetry of embryos arising from these halves may be due to: (1) the limited reserve of non-organizer MZ cells on the side of the cut that are brought into contact with the organizer by healing, and/or (2) the lack of time for cell interactions and movements to occur before cell fates are established. Roux (1888) killed one cell in the 2-cell embryo producing lateral halves. These embryos later do not exhibit bilateral symmetry but rather produce either a left or right half neurula, probably because the half organizer of the hemi-gastrula is bordered by inducible non-organizer MZ material on only one side (the other side being bordered by the dead cell). The 0/UV recombinants described in this paper exhibit bilateral symmetry probably because the organizer has inducible non-organizer MZ on both sides and because there is sufficient time for inductions to occur.
The bilateral symmetry of our stage 9 0/UV recombinants is due largely to the induction of dorsal axial structures (i.e. somites, notochord and neural tube) in the UV-irradiated half since lineage-labelling experiments show that somites, notochord and neural tube are derived from both the normal and UV-irradiated halves, even though unoperated UV-irradiated embryos never form notochord and rarely form somites and neural tube. These structures are induced in the UV-irradiated half by the lateral half. Many researchers have reported the induction of neural tube and somites in ventral tissue at stages after the late blastula (Spemann, 1938; Yamada, 1950; Slack and Forman, 1980; Gimlich and Cooke, 1983; Smith and Slack, 1983). The induction of notochord in ventral (or UV-irradiated) tissue after the late blastula stage has rarely been observed (Spemann and Mangold, 1924; Kaneda, 1981; Smith and Slack, 1983; Gimlich and Gerhart, 1984). Lineage labelling experiments in this paper provide a clear example of induction of notochord after the late blastula stage. Lineage labelling and fate mapping experiments further indicate that induction of notochord after the late blastula stage plays a role in normal development (Stewart and Gerhart, 1990).
ACKNOWLEDGEMENTS
We thank K. Anderson, B. Chasan, E. Ferguson, B. Kimmel, C. Phillips, D. Schneider, K. Symes, and the members of our lab for comments on the manuscript, and R. Gimlich (Oregon State Univ.) for a gift of FDA. The research was supported by USPHS grant GM19363 to J.C.G. and a predoctoral NSF fellowship (1-782000-21602-3) to R.M.S.