A monoclonal antibody, 2G9, has been identified and characterised as a marker of neural differentiation in Xenopus. The epitope is present throughout the adult central nervous system and in peripheral nerves. Staining is first detected in embryos at stage 21 in the thoracic region. By stage 29 it stains the whole central nervous system, except the tail tip. The epitope is present in a 65K Mr protein, and includes sialic acid. The antibody also reacts with neural tissue in mice and axolotls and newts. 2G9 was used to show that both notochord and somites are capable of neural induction, and the stimulus is present as late as stage 22. Attempts to demonstrate the induction of nervous system by developing nervous system (homoiogenetic induction) were unsuccessful. The view that the lateral extent of the nervous system might be determined by that of the inductive stimulus is discussed. Neural induction was detected as early as stage 10 and occurs in embryos without gastrulation and without cell division from stage .

The formation of the nervous system from the ectoderm of amphibians by neural induction has been one of the most intensively studied areas of experimental embryology, yet most aspects of this process remain a mystery. In Xenopus, the vertebrate species currently most used for studying early development, one does not know what substances bring about the induction, when it starts, precisely which tissues generate the stimulus, for how long it is produced, how long induction takes, or how the anteroposterior organisation of the nervous system arises. However, answers to these questions would provide an essential foundation on which an understanding of neural differentiation could be built.

An initial contribution to answering these questions can be made by providing neural markers for identifying neural tissue by objective criteria. Two nucleic acid probes that have recently been described are derived from the homeobox-containing gene, XIH box6, which identifies posterior nervous system (Sharpe et al. 1987), and from an N-CAM cDNA (Kintner and Melton, 1987; Kintner, 1988). Neither is convenient for routine histological analyses. A different approach is to make antibody markers, a route already taken with an N-CAM antiserum (Jacobsen and Rutishauser, 1986). We describe here a monoclonal antibody marker which can be used to stain the whole central nervous system and peripheral nerves. We use this to find when neural induction begins, which tissues are inductive, for how long this stimulus is produced and to what extent normal gastrulation and cell division are needed for neural induction.

Embryo culture, dissection, sectioning and staining of sectioned embryos were performed as described by Jones and Woodland (1986,1987). In all cases every section of explants, sandwiches or embryos was stained and examined. Whole mounts were stained following the procedure of Dent et al. (1989). Exogastrulae were produced by incubating demembranated blastulae in full-strength modified Barth’s Saline (MBS; see Jones and Woodland, 1986). Cytochalasin-B-treated embryos were incubated in fullstrength MBS containing 10 μgml-1 Cytochalasin B.

Western blotting was carried out by standard techniques. Adult or tadpole Xenopus laevis brains were dissected out and homogenised directly into SDS sample buffer. Samples were spun on a microfuge for 1 min and loaded onto a 12.5 % high bis polyacrylamide gel as indicated in Fig. 2, according to Laemmli (1970). The gel was electroblotted following the method described by Burnette (1981), and the binding visualised using a peroxidase-conjugated rabbit anti-mouse IgG (SIGMA), peroxidase-conjugated goat anti-rabbit IgG (SIGMA) and 4-chloro-l-naphthol.

Production of a neural marker, 2G9

This antibody was prepared by immunizing a Balb C mouse with adult Xenopus laevis brain and spinal cord.

After fusion and plating out of hybridoma cells in 96-well plates by standard methods (Kohler and Milstein, 1975), the supernatants were screened on sections of stage 46 tadpoles. One well gave specific staining to tadpole brain, spinal cord and peripheral nerves (Fig. 1). The cells responsible, designated 2G9, were then cloned by dilution.

Fig. 1.

Whole mount or cryostat-sectioned normal embryos or homoiogenetic sandwich combinations, stained with the anti-nervous system monoclonal 2G9. Monoclonal 2G9 cell supernatant was used as the primary antibody and peroxidase-linked rabbit anti-mouse IgG as the secondary antibody. (A – E) Progressive staining of the nervous system with developmental stage, initiating in the anterior thoracic region and moving both anteriorly and posteriorly as development proceeds. (A) A cryostat section of a stage 21 embryo just as the 2G9 antigen starts to appear. (B – D) Whole-mount immunostained embryos following the procedure described by Dent et al. (1989). (B) A field of embryos from stage 26 – 28, (C) a stage 28 embryo at higher magnification and (D) a stage 26 negative control carried out by omitting the anti-nervous system antibody. (E) A cryostat section from a stage 45 embryo showing the characteristic staining pattern. (F) The result of a homoiogenetic graft made by sandwiching stage 15 X. borealis nervous system into stage 10 – 1012 ectoderm from X. laevis. The nervous system staining using rhodamine-conjugated rabbit antimouse IgG, arrowed, is totally confined to the donor tissue, indicating the absence of homoiogenetic induction in this series of experiments (see Table 3). (G) The same section as in F stained with quinacrine, proving all the cells within the staining nervous tissue are from the donor X. borealis cells, showing their characteristic punctate staining absent from X. laevis. Bars: A,E=50 μm; B=0.9mm; C,D=0.29mm; F,G=50 μm.

Fig. 1.

Whole mount or cryostat-sectioned normal embryos or homoiogenetic sandwich combinations, stained with the anti-nervous system monoclonal 2G9. Monoclonal 2G9 cell supernatant was used as the primary antibody and peroxidase-linked rabbit anti-mouse IgG as the secondary antibody. (A – E) Progressive staining of the nervous system with developmental stage, initiating in the anterior thoracic region and moving both anteriorly and posteriorly as development proceeds. (A) A cryostat section of a stage 21 embryo just as the 2G9 antigen starts to appear. (B – D) Whole-mount immunostained embryos following the procedure described by Dent et al. (1989). (B) A field of embryos from stage 26 – 28, (C) a stage 28 embryo at higher magnification and (D) a stage 26 negative control carried out by omitting the anti-nervous system antibody. (E) A cryostat section from a stage 45 embryo showing the characteristic staining pattern. (F) The result of a homoiogenetic graft made by sandwiching stage 15 X. borealis nervous system into stage 10 – 1012 ectoderm from X. laevis. The nervous system staining using rhodamine-conjugated rabbit antimouse IgG, arrowed, is totally confined to the donor tissue, indicating the absence of homoiogenetic induction in this series of experiments (see Table 3). (G) The same section as in F stained with quinacrine, proving all the cells within the staining nervous tissue are from the donor X. borealis cells, showing their characteristic punctate staining absent from X. laevis. Bars: A,E=50 μm; B=0.9mm; C,D=0.29mm; F,G=50 μm.

2G9 first stains the nervous system at stage 21 in the anterior thoracic region. This rapidly spreads anteriorly and posteriorly to stain all of the nervous system, except the developing tail, by stage 29. Staining of the tail follows a little behind its tip, as it grows.

The antigen in adult brain may be shown, by Western blotting, to be a 65K Mr protein (Fig. 2). A band of the same size may be detected in dissected nervous systems from tadpoles at stage 41 and later. The antigen seems insufficiently abundant to blot from whole tadpoles. Staining of sections is abolished by neuraminidase, suggesting that the antigen is a sialoprotein (data not shown). However, since staining cannot be competed with sialic acid, the epitope must be more complex than sialic acid itself. The pattern of staining suggests that the antigen is associated with the plasma membrane.

Fig. 2.

Western blot of the antigen detected by the monoclonal antibody 2G9. Brains were manually dissected from anaesthetised stage 41 and 50 X. laevis tadpoles or from adult frogs and homogenised directly into SDS sample buffer. Loadings corresponding to up to 12 of a stage 41 or stage 50 brain (track labelled 41 or 50 respectively), or 1/25 of an adult brain (track labelled adult) were run on a 12 % SDS – polyacrylamide gel and blotted onto nitrocellulose filtures. Filters were stained with 2G9 hybridoma supernatant and peroxidase-linked rabbit anti-mouse IgG and peroxidase-linked goat anti-rabbit IgG and developed with 4-chloro-l-naphthol. A control filter loaded with half a stage 50 brain (c) was stained with PBS followed by peroxidase-linked components only. This track was completely blank.

Fig. 2.

Western blot of the antigen detected by the monoclonal antibody 2G9. Brains were manually dissected from anaesthetised stage 41 and 50 X. laevis tadpoles or from adult frogs and homogenised directly into SDS sample buffer. Loadings corresponding to up to 12 of a stage 41 or stage 50 brain (track labelled 41 or 50 respectively), or 1/25 of an adult brain (track labelled adult) were run on a 12 % SDS – polyacrylamide gel and blotted onto nitrocellulose filtures. Filters were stained with 2G9 hybridoma supernatant and peroxidase-linked rabbit anti-mouse IgG and peroxidase-linked goat anti-rabbit IgG and developed with 4-chloro-l-naphthol. A control filter loaded with half a stage 50 brain (c) was stained with PBS followed by peroxidase-linked components only. This track was completely blank.

The epitope recognised by 2G9 seems to be conserved in vertebrate evolution, since the antibody also stains the nervous system of axolotls and mice.

Which regions of mesoderm in the neurula are capable of neural induction

It is well known that the signal of neural induction is generated by dorsal mesoderm, but, in Xenopus, as in most species, the region of mesoderm responsible has not been accurately mapped. We have attempted to map the regions responsible for producing the neural induction signal by dissection from embryos ranging from early gastrula (stage 10) to early tail bud (stage 22). In early gastrula embryos sectors of the marginal zone mesoderm were dissected as described by Dale and Slack (1987b). Three sectors were tested for inducing activity, the dorsal marginal zone (DMZ), the dorsolateral marginal zone (DLMZ) and the ventrolateral marginal zone (VLMZ). In such early gastrula embryos it is not possible to dissect out notochord or somite, but only those regions within which these structures fate-map (Dale and Slack, 1987a). At the end of gastrulation the notochord becomes delimited and it becomes possible to isolate this tissue from the somite regions by dissection. The somite region can only be approximately separated from future lateral plate at this stage, and either fraction might include presumptive kidney. At later neurula stages, myotomes may more accurately be removed. These dissected mesoderm fragments were sandwiched within two pieces of responsive ectoderm from stage 9 or 10 embryos. The mesoderm and ectoderm were taken variously from X. laevis and X. borealis to mark the origin of the tissues formed. Similar results were obtained which ever way round the two species were used. The sandwiches were then cultured to stage 31 – 35, before fixing, sectioning and staining with 2G9 or other tissue-specific antibody markers to score for the induction of neural tissue. At all stages from 10 to 22 the most dorsal mesoderm from the DMZ or isolated notochord induced positively staining areas of nervous sytem (Table 1). This was also true of the DLMZ and somites, though at lower frequency, but more ventral regions of the mesoderm, VLMZ and lateral plate, were uniformly negative. Varying, but substantial amounts of neural tissue were formed from sandwiches including notochord and somites. No attempt was made to quantify this. Thus the extent of the region generating the primary stimulus of neural induction at least approximates to the extent of the neural plate. The accuracy of the notochord dissections was checked for contaminating muscle by staining with the muscle-specific reagent B4 (Jones and Woodland, 1987) and somite dissections for notochord by morphology. In no cases were contaminating cells found. None of the sandwiches containing DLMZ cells contained notochord, consistent with the low percentage of notochord fated to derive from this region (Dale and Slack, 1987a).

Table 1.

A measure of the inductive capacity of different components of mesoderm in gastrula neurula and early tail bud embryos tested by combination in ectodermal sandwiches with competent ectoderm, using X. laevis and X. borealis

A measure of the inductive capacity of different components of mesoderm in gastrula neurula and early tail bud embryos tested by combination in ectodermal sandwiches with competent ectoderm, using X. laevis and X. borealis
A measure of the inductive capacity of different components of mesoderm in gastrula neurula and early tail bud embryos tested by combination in ectodermal sandwiches with competent ectoderm, using X. laevis and X. borealis

Can the nervous system itself induce more nervous system?

Using other species previous authors have presented data that suggest that the neural plate can induce more neural tissue, a process called ‘homoiogenetic’ induction (see Discussion). We have attempted to demonstrate the existence of this phenomenon in two kinds of experiment.

(i) Grafting of neural plate into ventral regions of intact embryos

Superficial ectoderm was removed from presumptive neural plate of stage gastrulae of X. borealis embryos and grafted into ventral positions of X. laevis embryos at stages when the neighbouring ectoderm is still responsive to neural induction (Jones and Woodland, in preparation). As judged by stereomicroscope inspection, the grafts became incorporated within 30 min, resembling similar grafts drawn by Spemann (1938). The grafted embryos were grown on to stage 35, fixed and sectioned. When embryos after stage 11 were used as donors, the small areas of grafted tissue usually formed areas of positive 2G9 staining, often in the form of tubes, sometimes as flat patches. In contrast the adjacent host cells, in this case ventral epidermis, never showed any neural staining; rather they stained in a normal fashion with the epidermal marker 2F7 (Fig. 3, Table 2). In contrast, stage dorsal ectoderm from positions not yet underlain by dorsal mesoderm grafted into a ventral position, formed normal epidermis and integrated fully with surrounding host ventral epidermis. Control grafts from dorsal donors grafted into dorsal positions integrated into normal nervous systems and stained positively with 2G9 as expected. Thus these experiments provide no evidence for homoiogenetic neural induction via lateral transfer of the inducing signal.

Table 2.

Grafting of presumptive neural plate into whole embryos in ventral or dorsal positions to test for homoiogenetic induction

Grafting of presumptive neural plate into whole embryos in ventral or dorsal positions to test for homoiogenetic induction
Grafting of presumptive neural plate into whole embryos in ventral or dorsal positions to test for homoiogenetic induction
Fig. 3.

Homoiogenetic induction; grafting of neural plate into ventral regions of intact embryos. Dorsal ectoderm from stage 1112X. borealis embryos was grafted ventrally into the ectoderm of stage 11 X. laevis hosts. Grafted embryos were cultured until stage 35, fixed and stained as described in Materials and methods. (A) Small ventral nervous system derived from X. borealis donor graft stained with 2G9. (B) Epidermis adjacent to the graft staining normally with the epidermal marker 2F7.C7. (C) Quinacrine staining of the same region showing that no host (X. laevis) cells participate in the grafted nervous system. Bar=50 μm.

Fig. 3.

Homoiogenetic induction; grafting of neural plate into ventral regions of intact embryos. Dorsal ectoderm from stage 1112X. borealis embryos was grafted ventrally into the ectoderm of stage 11 X. laevis hosts. Grafted embryos were cultured until stage 35, fixed and stained as described in Materials and methods. (A) Small ventral nervous system derived from X. borealis donor graft stained with 2G9. (B) Epidermis adjacent to the graft staining normally with the epidermal marker 2F7.C7. (C) Quinacrine staining of the same region showing that no host (X. laevis) cells participate in the grafted nervous system. Bar=50 μm.

(ii) Grafting isolated presumptive neural tissue with isolated ectoderm

The design of experiment described above is limited in the range of graft and host tissues that can be used since it is not applicable to post-neural plate donors. This is not so if neural plate is grafted against a piece of responsive ectoderm, with the two inner surfaces in opposition. Alternatively the neural plate, or even neural tube, can be sandwiched between two pieces of ectoderm. These methods also permit one to use responsive tissue at the blastula stage to allow a considerable period of contact to occur before the ectoderm loses its competence to respond (at about stage , Jones and Woodland, in preparation). The two tissues were variously derived from X. laevis and X. borealis, the results being the same whichever way round the combinations were made. In all instances gastrula and neurula presumptive nervous system developed 2G9 staining, the adjacent or enveloping ectoderm never did so, forming only epidermis (Table 3, Fig. 1). Thus these experiments also show no evidence for homoiogenetic induction of neural tissue by future nervous system.

Table 3.

The lack of occurrence of homoiogenetic induction in sandwich combinations of early gastrula ectoderm and late gastrula and neurula neural tissue

The lack of occurrence of homoiogenetic induction in sandwich combinations of early gastrula ectoderm and late gastrula and neurula neural tissue
The lack of occurrence of homoiogenetic induction in sandwich combinations of early gastrula ectoderm and late gastrula and neurula neural tissue

The stage at which neural induction first occurs

One test for the start of neural induction is to isolate regions of early ectoderm fated to form nervous system and allow them to develop in isolation to see if the 2G9 antigen appears. Blastula ectoderm not underlain by mesoderm invariably forms only epidermis, as judged by 2F7 staining, a result previously described by Jones and Woodland (1986). At stage 10 this is true of all of the ectoderm, except that immediately anterior to the dorsal lip, the area first underlain by gastrulating dorsal mesoderm. In isolation this region forms nervous system (Fig. 4), as well as perfect, circular cement glands. To aid the survival of these small explants, they were sometimes enclosed in stage 9 ectoderm. 12 explants were analysed, all of which contained some 2G9 positive staining at stage 35 equivalent. Control ectoderm from a similar number of stage 9 embryos was always negative for the nervous system marker.

Fig. 4.

Neural induction has already started by stage 10. Ectoderm from the dorsal region of stage 10, already underlain by invaginating dorsal mesoderm, was dissected away from underlying tissue and incubated until control embryos reached stage 35. Explants were fixed, embedded and sectioned as described in Materials and methods before staining with 2G9- and FITC-conjugated rabbit anti-mouse IgG. Bar=50 μm.

Fig. 4.

Neural induction has already started by stage 10. Ectoderm from the dorsal region of stage 10, already underlain by invaginating dorsal mesoderm, was dissected away from underlying tissue and incubated until control embryos reached stage 35. Explants were fixed, embedded and sectioned as described in Materials and methods before staining with 2G9- and FITC-conjugated rabbit anti-mouse IgG. Bar=50 μm.

A problem with this experiment is that it is necessary to be quite certain that the ectodermal explant does not include mesodermal cells. The kinds of dorsal mesoderm known to induce neural plate are notochord and somite (see above). Notochord is easily recognisable morphologically, and did not appear in the developing explants. The possibility of the presence of muscle was tested by staining explants positive for nervous system with an antibody specific for striated muscle, B4 (Jones and Woodland, 1987); no staining was found (data not shown). Thus the kinds of dorsal mesoderm for which we have tests were not present in explants that showed neural development. Furthermore the remainder of the cells in these explants stained with the epidermal marker, 2F7, suggesting that these small explants did not contain mesoderm which had not fully differentiated into identifiable muscle or notochord. We conclude that neural induction is already in progress by stage 10, the start of gastrulation.

Neural induction without gastrulation or cell division

Hypertonic media prevent normal gastrulation, the resulting ‘exogastrula’ having a ball of ectoderm at one end and endoderm at the other, the latter never invaginating in a normal way. The mesoderm is found in a stalk between the two and also extends into the pigmented and yolky regions. In Xenopus exogastrulae this central region also includes superficial ectoderm. This area around the mesoderm stains with 2G9, the neural marker, over large areas consistent with the results of Kintner and Melton (1987). Thus although gastrulation movements are abnormal and highly curtailed, but not absent, neural induction is still extensive.

A more complete arrest of gastrulation movements may be produced by cytochalasin B treatment at blastula stages. Although this agent also produces virtually immediate cessation of cell division, we have previously shown that from stage onwards epidermal development, signalled by 2F7 staining, nevertheless occurs (Jones and Woodland, 1986). We have cultured such embryos in their vitelline membranes until stage 31 in order to test if any neural development occurs. Long before this stage the arrested blastula becomes a bag of rounded, disaggregated cells, whitish in colour, having lost all organised distribution of pigment granules. Each cell is multinucleate. Nevertheless, if the cytochalasin treatment was performed from stage or later some cells of each embryo developed unequivocal staining with 2G9 (Fig. 5). The staining is in a small patch on one side of the embryo. Prior to this stage no staining was observed. Thus neural induction and some steps in neural differentiation can occur even in the absence of cell division from the mid-blastula stage. Presumably mesoderm and ectoderm cells are in sufficient proximity in the disorganised mass of cells for the induction to occur though this has not been formally proven. The failure to see staining when cell division is blocked earlier may be significant, but as with any such negative result, the cause may be artefactual. Attempts to identify differentiated muscle cells in cytochalasin-B-treated embryos have not yielded positive results. In addition notochord morphology was not visible in any cells, although the disruptive effects of cytochalasin B are so great that such morphology would not be expected to be recognisable. Therefore we could not tell if the adjacent 2G9-positive cells were adjacent to dorsal mesodermal cells.

Fig. 5.

Neural induction occurs in cytochalasin-treated embryos. Whole embryos, within their vitelline membranes, were incubated in cytochalasin B (10 μgml-1) to inhibit cytokinesis, until control embryos reached stage 35. Treated embryos were fixed, embedded sectioned and stained with 2G9 as previously described. A small area of staining (arrowed) is evident in the animal pole cells. Bar=25 μm.

Fig. 5.

Neural induction occurs in cytochalasin-treated embryos. Whole embryos, within their vitelline membranes, were incubated in cytochalasin B (10 μgml-1) to inhibit cytokinesis, until control embryos reached stage 35. Treated embryos were fixed, embedded sectioned and stained with 2G9 as previously described. A small area of staining (arrowed) is evident in the animal pole cells. Bar=25 μm.

The results described above, using a new antibody marker staining throughout the nervous system, bear on the mechanism by which the neural plate is formed. One current idea of how the process of neural induction occurs has been expounded by Nieuwkoop and his colleagues (Nieuwkoop et al. 1985; Albers, 1987). It is argued that the neural induction stimulus emanates only from the most dorsal mesoderm, the notochord, and that this neuralises a central longitudinal strip of ectoderm. This stimulus is then propagated in a ventrolateral direction within the ectodermal sheet by homoiogenetic induction, that is, induction of more nervous system by presumptive neural tissue. The stimulus spreads slowly as the ectoderm loses its competence to respond, the dying embers of responsiveness generating the neural crest at the ventral (lateral) margins of the neural plate. The central postulates of this theory are firstly that only the notochord induces nervous system, and secondly that homoiogenetic induction of neural tissue occurs. Our results bear on both of these points.

We show that both the notochord and the somites can induce neural tissue when placed in artificial recombinates. This correlates with previous descriptions of manipulated embryos in which a neural tube forms in an embryo in which the centre of the dorsal lip was removed, producing an embryo with somites, but no notochord (Lehmann, 1928). The same effect was seen when gastrulae were treated with lithium ions (Lehmann, 1935). Recently, in the course of injecting cloned actin genes, we have found Xenopus neurulae that have no notochord, the somites being fused across the central axis of the embryo; nevertheless they form a neural tube. As far as one can tell, the lateral extent of the somites at stage approximates to the extent of the neural plate. There is therefore no reason to exclude the possibility that the lateral extent of the neural plate is determined solely by the lateral extent of the inducing tissue. Confirmation of this idea would partly depend on even more precise mapping of the area of the future somites that can induce, and correlation of this with the extent of the neural plate. At stages when morphological landmarks have not appeared this is not a technically trivial task. Nevertheless, this interpretation of our results suggests that there is no a priori necessity to postulate the occurrence of homoiogenetic neural induction; but the question whether it actually occurs remains.

The clearest way to demonstrate homoiogenetic neural induction is to place neural plate tissue or later neural tubes in contact with responsive ectoderm well away from any dorsal mesoderm and look for neural induction in the test ectoderm. This we have done by moving neural plate tissue to a ventral position or by sandwiching it in responsive ectoderm. In no case did we see neural differentiation in the host ectoderm, judged by neural tube formation or by 2G9 staining, although the grafted presumptive neural tissue developed these characteristics, as expected. These experiments therefore argue against the occurrence of homoiogenetic induction in the Xenopus nervous system. However, the experiments suffer from being negative results, and it is hard to counter the criticism that the conditions were not right to favour this kind of cell interaction, perhaps because cell communication is reestablished too slowly in grafts.

One early demonstration of homoiogenetic induction was that of Mangold and Spemann (1927), who placed pieces of the neural plate into the blastocoel of newt embryos. Subsequently secondary nervous systems were induced. Blastocoel grafts constitute a very complicated situation and this kind of experiment needs further analysis. More recently Albers (1988) has performed experiments in which responsive gastrula ectoderm is placed lateral to the neural plate of later urodele embryos. This graft appeared to form a secondary neural tube. The interpretation was that the neural induction signal emanated from the adjacent neural plates. However, as the neural plate rolled up the graft would have come to lie over the somites, which we show, in accordance with previous work, have inductive capacity in Xenopus at the stage concerned. Thus this experiment provides no unequivocal evidence for homoiogenetic neural induction.

The strongest current evidence for homoiogenetic induction comes from the experiments of Nieuwkoop et al. (1952). Here flaps of ectoderm were placed in the neural plate to grow as a turret. The results were variable, but in many instances nervous system was formed with an anteroposterior organisation. There was reported to be no direct contact with the mesoderm, so only homoiogenetic induction could have provided the effect. Although this seems like a strong experimental design there remains the inevitable problem that neural induction is a rapid process (e.g. some neural tissue is specified even by stage 10), and that the graft is very near to, if not in contact with, dorsal mesoderm. It therefore seems possible that some signal from the latter might diffuse into the graft region. If true homoiogenetic induction occurs in this system we cannot easily see why our tests should be uniformly negative. We feel that at present the case for homoiogenetic induction remains unproven. Moreover there is currently no uncontrovertible evidence that contradicts the simple theory that in Xenopus the lateral extent of the neural plate is solely or primarily determined by the extent of the induction stimulus in underlying tissues, with which it is in close contact. In a limited series of our own graft experiments using such grafted flaps, no induction of the grafted ectoderm occurred. The grafted cells formed epidermis exclusively.

The other information our experiments provide is to show firstly that gastrulation movements are not needed for neural induction and secondly that it occurs without any cell division after stage 7. The fact that epidermal differentiation is also possible in stage embryos blocked in cleavage, but not in those blocked before (Jones and Woodland, 1986), suggests that some critical event in ectoderm development occurs at this stage. It is impossible to say whether this is something trivial relating to resistance to cytochalasin, or an event essential for subsequent ectoderm development of all kinds.

We gratefully acknowledge the technical assistance of Jenni Smith and Surinder Bhamra and the financial support of the MRC and AFRC.

Albers
,
B.
(
1987
).
Competence as the main factor determining the size of the neural plate
.
Development, Growth and Differentiation
29
,
535
545
.
Burnette
,
W. N.
(
1981
).
“Western Blotting”: Electrophoretic transfer of proteins from sodium dodecyl sulphatepolyacrylamide gels to unmodified nitrocellulose and radiographic detection with antibody and radio-iodinated protein A
.
Anal. Biochem
.
112
,
195
203
.
Dale
,
L.
and
Slack
,
J. M. W.
(
1987a
).
Fate map for the 32-cell stage of Xenopus laevis
.
Development
99
,
527
551
.
Dale
,
L.
and
Slack
,
J. M. W.
(
1987b
).
Regional specification within the mesoderm of early embryos by Xenopus laevis
.
Development
100
,
279
295
.
Dent
,
J. A.
,
Polson
,
A. G.
and
Klymkowsky
,
M. W.
(
1989
).
A whole-mount immunocytochemical analysis of the expression of the intermediate filament vimentin in Xenopus
.
Development
105
,
61
74
.
Jacobson
,
M.
and
Rutishauser
,
U.
(
1986
).
Induction of the neural cell adhesion molecule (N-CAM) in Xenopus embryos
.
Devi Biol
.
116
,
524
531
.
Jones
,
E. A.
and
Woodland
,
H. R.
(
1900
).
Neural induction in Xenopus. temporal aspects
.
In preparation
.
Jones
,
E. A.
and
Woodland
,
H. R.
(
1986
).
Development of the ectoderm in Xenopus: tissue specification and the role of cell association and division
.
Cell
44
,
345
355
.
Jones
,
E. A.
and
Woodland
,
H. R.
(
1987
).
The development of animal cap cells in Xenopus: a measure of the start of animal cap competence to form mesoderm
.
Development
101
,
557
563
.
Kintner
,
C. R.
(
1988
).
Effects of altered expression of the neural adhesion molecule, N-CAM on early neural development in Xenopus embryos
.
Neuron
1
,
545
555
.
Kintner
,
C. R.
and
Melton
,
D. M.
(
1987
).
Expression of Xenopus N-CAM RNA is an early response of ectoderm to induction
.
Development
99
,
311
325
.
Kohler
,
G.
and
Milstein
,
C.
(
1975
).
Continuous culture of fused cells secreting antibody of predefined specificity
.
Nature, Lond
.
256
,
495
497
.
Laemmli
,
U. K.
(
1970
).
Cleavage of structural proteins during the assembly of the head of bacteriophage T4
.
Nature, Lond.
121
,
680
685
.
Lehmann
,
F. E.
(
1928
).
Die Bedeutung der Unterlagering fiir die Entwicklung der Medullarplatte von Triton
.
Wilhelm Roux’s Arch. EntwMech. Org
.
113
,
123
171
.
Lehmann
,
F. E.
(
1935
).
Die Entwicklung von Rulckenmark, Spinalganglien und Wirbel anlagen in Chordalosen Korperregionem von Tritonlarven
.
Rev. Suisse de Zool
.
42
,
405
415
.
Mangold
,
O.
and
Spemann
,
H.
(
1927
).
Uber Induktion von Medullarplatte durch Medullarplatte in jiingeren Keim, ein Beispeil homoeogenetischer Oder assimilatorischer Induktion
.
Wilhelm Roux Arch. EntwMech. Org
.
111
,
341
342
.
Nieuwkoop
,
P. D.
,
Boterenbrood
,
E. C.
,
Kremer
,
A.
,
Bloemsma
,
F. F. S. N.
,
Hoessels
,
E. L. M. J.
,
Meyer
,
G.
and
Verheyen
,
F. J.
(
1952
).
Activation and organisation of the central nervous system in amphibians
.
J. Elxp. Zool
.
120
,
1
108
.
Nieuwkoop
,
P. D.
,
Johnen
,
A. G.
and
Albers
,
B.
(
1985
).
The Epigenetic Nature of Early Chordate Development
.
Cambridge University Press
.
Sharpe
,
C. R.
,
Fritz
,
A.
,
De Robertis
,
E. M.
and
Gurdon
,
J. B.
(
1987
).
A homeobox-containing marker of posterior neural differentiation shows the importance of predetermination in neural induction
.
Cell
50
,
749
758
.
Spemann
,
H.
(
1938
).
Embryonic Development and Induction
.
Yale University Press
.