- Email: [email protected]
Homeogenetic neural induction in xenopus
DEVELOPMENTALBIOLOGY 147, 73-82 (1991)
Homeogenetic Neural Induction in Xenopus M A R C SERVETNICK AND ROBERT M. G R A I N G E R
Department of Biology, University of Virginia, Charlottesville, Virginia 22901 Accepted May 28, 1991 Neural induction is known to involve an interaction of ectoderm with dorsal mesoderm during gastrulation, but several kinds of studies have argued that competent ectoderm can also be neuralized via an interaction with previously neuralized tissue, a process termed homeogenetic neural induction. Although homeogenetic neural induction has been proposed to play an important role in the normal induction of neural tissue, this process has not been subjected to detailed study using tissue recombinants and molecular markers. We have examined the question of homeogenetic neural induction in Xenopus embryos, both in transplant and recombinant experiments, using the expression of two neural antigens to assay the response. When ectoderm that is competent to be neuralized is transplanted to the region adjacent to the neural plate of early neurula embryos, it forms neural tissue, as assayed by staining with antibodies against the neural cell adhesion molecule, N-CAM. Transplants to the ventral region, far from the neural plate, do not express N-CAM, indicating that neuralization is not occurring as a result of the transplantation procedure itself. Because this response might be occurring as a result of interactions of ectoderm with either adjacent neural plate tissue, or with underlying dorsolateral mesoderm, recombinant experiments were performed to determine the source of the neuralizing signal. Ectoderm cultured in combination with neural plate tissue alone expresses neural markers, while ectoderm cultured in combination with dorsolateral mesoderm does not. We conclude that neural tissue can homeogenetically induce competent ectoderm to form neural tissue and argue that this induction occurs via planar signaling within the ectoderm, a mechanism that, in normal development, may be involved in interactions within presumptive neural ectoderm or in specifying structures that lie near the neural plate. © 1991AcademicPress,Inc.
Another form of signaling may also be important in the induction of the neural plate, namely homeogenetic neural induction, the induction of neural tissue by neural tissue. Homeogenetic neural induction has been proposed to play a central role in the normal formation of the neural plate (Nieuwkoop, 1952, 1985; Albers, 1987) and in specifying the regional nature of induced neural tissue (Nieuwkoop, 1952; Nieuwkoop and Albers, 1990). Although several kinds of experiments have provided evidence for homeogenetic neural induction, these experiments are subject to criticisms that render their interpretation problematic. These criticisms are, first, that the responding tissue in these experiments may have been subject to some influence from mesodermal tissues in addition to those from neural tissue, and, second, that the neural response of the tissue was not verified by the use of unambiguous markers of neural differentiation. Homeogenetic induction was first observed in T r i t u r u s by Mangold and Spemann (1927; s e e also Mangold, 1929, 1933), who placed pieces of neural plate tissue into the blastocoel of a host embryo, which often formed neural tissue in the region of the implant. Such blastocoel implants are difficult to interpret: the precise location, manner of contact, and time of contact are not well controlled in such cases, with the result that the implanted tissue may be acting on ectoderm that has already been acted upon by other tissues, and that may
INTRODUCTION
Although the induction of the central nervous system is one of the primary patterning events in vertebrate development, the process of neural induction is not well understood. Classical evidence (reviewed by Spemann, 1938) indicates that, during gastrulation, dorsal mesoderm transmits signals vertically to the overlying ectoderm, inducing it to form neural tissue; such vertical signaling by the mesoderm has recently been shown to be sufficient to induce the expression of neural-specific genes (Sharpe and Gurdon, 1990; Hemmati-Brivanlou et al., 1990). Recent evidence indicates that a second signal may also be involved in this process (reviewed by Smith, 1989a; New and Smith, 1990). These studies suggest that, prior to invagination of the dorsal mesoderm during gastrulation, a planar signal originating from the region of the dorsal blastopore lip passes laterally through the plane of tissue into the ectoderm. The planar signal has been reported to bias dorsal portions of the ectoderm toward a neural pathway of development (Savage and Phillips, 1989), to specify dorsal ectoderm cells to adopt a neural fate (Dixon and Kintner, 1989), to act synergistically with the vertical signal to elicit maximal expression of neural-specific genes (Dixon and Kintner, 1989), and to play a role in patterning the forming neural plate (Ruiz i Altaba, 1990). 73
0012-1606/91 $3.00 Copyright© 1991by AcademicPress,Inc. All rightsof reproductionin any formreserved.
74
DEVELOPMENTALBIOLOGY
not be completely naive with respect to neural induction. This criticism is especially pertinent given recent results showing that neuralization can be the result of two signals acting synergistically (Dixon and Kintner, 1989). In experiments t h a t circumvent the host effects associated with blastocoel implants, Nieuwkoop (1952), using Ambystoma, Triturus, and Pleurodeles, inserted flaps of folded ectoderm into the presumptive neural plate. He observed neural tissue formed at some distance from the surface of the embryo, arguing that a homeogenetic neuralizing signal can pass through the ectoderm. These experiments have been criticized (Jones and Woodland, 1989) because the ectodermal flaps may have been exposed to signals from the underlying mesoderm, even if only briefly. Recently, Albers (1987; see also Holtfreter, 1933, 1938) performed transplant experiments in Ambystoma in which competent ectoderm was transplanted so as to span the neural-epidermal boundary. The transplanted ectoderm showed neuralization beyond the normal borders of the neural plate, suggesting homeogenetic induction. In these experiments, it has been suggested that the transplanted ectoderm may have come into contact with dorsal mesoderm during neurulation, which could have resulted in neural induction of the transplants (Jones and Woodland, 1989); furthermore, these studies did not test the possibility that the mesoderm outside the neural region, which underlies the transplants, possesses neural-inducing ability. In experiments that exclude any possible influence of mesoderm, Xenopus neural tissue was recombined in vitro with competent ectoderm (Deuchar, 1971; Grunz, 1990). Although ectodermal cells were incorporated into neural structures, no biochemical markers of neural differentiation were used in these studies, raising the possibility that the cells incorporated into neural structures were not themselves neuralized. Such concerns are especially relevant in light of a recent report (Jones and Woodland, 1989) t h a t did not detect homeogenetic neural induction in Xenopus, using both a neural-specific antibody marker and host and donor marking of recombined tissues in transplants as well as recombinants. Here we have reexamined homeogenetic neural induction in Xenopus. Our transplant and recombinant experiments, in conjunction with cell lineage tracing, argue that neural plate tissue can induce neural tissue from competent ectoderm in the absence of mesoderm. We have also shown that the responding cells express two different molecular markers of neural differentiation.
VOLUME147, 1991
MATERIALS AND METHODS
Tissue transplants and recombinants. Xenopus laevis embryos were staged according to Nieuwkoop and Faber (1967). Animal cap tissue was manually removed from FLDx-labeled1 (fluoresceinated dextran; Molecular Probes, Junction City, OR) early gastrula (stage 10) embryos, using sharpened watchmaker's forceps, and transplants were made as described in Servetnick and Grainger (1991). For recombinant experiments, anterior neural tissue was removed either according to the procedure described by Slack (1984), in which ectoderm is dissected in the presence of 0.01% trypsin, followed by washing in 0.02% soy bean trypsin inhibitor, or by dissecting ectoderm in the presence of 1.5 mg/ml collagenase, followed by washing of dissected tissue in 5 mg/ml bovine serum albumin (all proteins from Sigma; all in 100% Steinberg's solution); no differences were observed in results of experiments using the two procedures. Dorsolateral mesoderm was usually dissected without the use of proteolytic enzymes, although it was occasionally isolated from embryos that had been placed in enzyme solutions to isolate other tissues. Dissected neural plate tissue was examined under the dissecting microscope before being apposed to ectoderm in order to verify that no adherent mesoderm cells were present. Recombinants of ectoderm and neural tissue never showed any recognizable mesodermal tissue (notochord or muscle). Although such histological analysis alone cannot rule out the presence of small amounts of head mesoderm (which does not form readily identifiable cell types, and for which no molecular markers are currently available), contamination by even a few mesodermal cells would have been detected microscopically at the time of dissection. Recombinants made with dorsolateral mesoderm were devoid of notochord or muscle (derivatives of dorsal mesoderm) except in a single case, in which mesoderm from a late gastrula (early stage 12) embryo formed notochord; this was probably due to dissection of the wrong region of mesoderm and this case was discarded. Culture of recombinants. Explanted tissues were placed on top of one another on a base of clay and held in tight contact by a curved glass bridge (Henry and Grainger, 1987) for 30-60 min. Recombinants were left in 100% Steinberg's solution overnight, after which the concentration was reduced to ~50% Steinberg's solution. Histology and immunofluorescence. Embryos or re-
1Abbreviationsused:DIC:differentialinterferencecontrast;FLDx: fluoresceinateddextran; N-CAM:neural cell adhesionmolecule;PLR: presumptive lens region.
SERVETNICK AND GRAINGER
combinants were fixed at stages 39-42 (stages of recombinants were monitored using unoperated sibling controls) in either Romeis fixative (for N-CAM staining; Romeis, 1968; Hausen et al., 1985) or in a solution of 2.0 ml 5% trichloroacetic acid, 1.5 ml 37% formaldehyde, 2.5 ml distilled water (for 2G9 staining). Fixation, embedding, sectioning, and staining were as described in Servetnick and Grainger (1991). N-CAM antibodies (provided by U. Rutishauser; see Balak et al., 1987) were used at a dilution of 1:25; monoclonal antibody 2G9 tissue culture supernatant (provided by E. Jones; see Jones and Woodland, 1989) was used undiluted. Cell lineage tracing. To trace the origin of cells in all experiments, one-cell embryos were injected with FLDx (Gimlich and Braun, 1985), and FLDx-labeled ectoderm was used as the test ectoderm in all experiments. We have observed, in some experiments, small amounts of FLDx fluorescence in non-FLDx-injected cells at the junction of labeled and unlabeled tissues, resulting in an apparent gradient of FLDx fluorescence (see, for example, Fig. 3B). The cells that show this effect exhibit weak nuclear fluorescence, rather than the strong, widespread cytoplasmic fluorescence observed in FLDx-injected tissue. This effect is most pronounced in recombinants, although we have also observed it in several transplant experiments in which FLDx-labeled ectoderm transplants are incorporated into the host neural tube. We speculate that it may be due to death of some labeled cells, followed by uptake of released FLDx. Nevertheless, labeled and unlabeled cells are generally easily distinguishable from one another. In the few cases in which there was any ambiguity in identification of cells, these cases were discarded. In scoring experiments, we have been careful to score as positive only those tissues that are positive for antibody staining and t h a t consist of significant blocks of cells exhibiting the bright cytoplasmic fluorescence diagnostic of FLDx-injected cells. RESULTS
Neuralization of Ectoderm Transplants Placed Outside the Neural Plate During the course of experiments on lens induction, we transplanted ectoderm to the presumptive lens region (PLR) of neural plate (stage 14) embryos (Fig. la). Transplanted early gastrula (stage 10) ectoderm, which is competent to form neural tissue but is not yet competent to form lens (Servetnick and Grainger, 1991), often neuralized, despite its location outside the neural plate. The neuralization of this ectoderm was confirmed by staining with an antibody against Xenopus neural cell adhesion molecule, N-CAM, which is expressed specifically in neural tissue at the stages assayed (Balak et al.,
Homeogenetic Neural Induction
75
a
removeectoderm from FLDxlabelled early gastrula
b
transplantto headectodermor ventralectodermof neuralplate stagehost
~
ectodermalone
/ ectoderm ~
y
removeectoderm from FLDx-rabelled early gastrula
anterior neurarplate
~
~
/
ectoderm +
dorsolatera] headmesoderm
FIG. 1. Transplant and recombinant experiments to assay for neural induction. (a) Ectoderm was removed from FLDx-labeled early gastrula (stage 10) embryos and transplanted into a neural plate stage (stage 14) embryo from which the corresponding portion of ectoderm had been removed. Transplants were made to the dorsolateral head region (presumptive lens region), or to the anterior ventral region, as shown. In addition, transplants were made to lateral head ectoderm, corresponding to the region lying between the two areas shown in the figure, and to posterior ventral regions (not shown). (b) Ectoderm was removed from FLDx-labeled early gastrula (stage 10) embryos and cultured either alone, in combination with anterior neural plate tissue, or in combination with mesoderm underlying the dorsolateral head region. The diagram shows ectoderm from a stage 10 embryo and neural plate tissue and head mesoderm from a neural plate stage embryo. However, both late blastula (stage 9) and early gastrula (stage 10) embryos were used as ectoderm donors, and embryos from stages 12 (late gastrula stage) to 14 (neural plate stage) were used as the source of neural plate. In all experiments, embryos or recombinants were cultured to the equivalent of swimming tadpole stages (stages 39-42), fixed, sectioned, and stained by immunofluorescence with antibodies t h a t recognize either N-CAM or the neural antigen 2G9.
1987). An example of such a transplant is shown in Figs. 2A to 2C. The observed neuralization might have occurred as a response of transplanted competent ectoderm to specific inductive signals, or it could have been evoked by the transplantation procedure itself; such "autoneuralization" of ectoderm is known to occur in response to nonspecific stimuli in many species (Holtfreter, 1947; Saxen, 1989), although it is much more difficult to elicit such a response in Xenopus (Grunz, 1985; see also Grunz and Tacke, 1989). To address this point, we compared ectoderm transplants to the PLR (in the dorsolateral
76
DEVELOPMENTAL BIOLOGY
VOLUME147, 1991
SERVETNICKAND GRAINGER
head region, just outside the neural plate) with transplants made to other regions of host embryos. In contrast to transplants to the PLR, transplants placed in the ventral region rarely neuralized (Figs. 2G to 2I). Results of these experiments are summarized in Table 1. In the two cases in which ectoderm transplanted to the ventral region did show a neural response, this was because the ectoderm transplants extended unusually far dorsally (presumably as a result of the extensive stretching of ectoderm during neurulation), reaching the level of the midpharynx. Transplants t h a t were made to lateral head ectoderm (intermediate between the dorsolateral and ventral regions) showed a variable response. In these, the transplants that were located more dorsally formed neural tissue, whereas more ventrally located transplants did not; all of the positive responses observed in this series occurred dorsal to the level of the midpharynx. One of the ventral-most cases of neuralization that we have observed is shown in Figs. 2D to 2F. Presumably, the ectoderm is responding to inductive signals, and these are present in dorsal, but not ventral, regions of the embryo. In transplants to the PLR, a small amount of host ectoderm was often present between the host neural plate and the transplant. In most of these cases, as well as in transplants made to regions further from the neural plate (e.g., Figs. 2D to 2F), the transplants were nevertheless induced to form neural tissue, indicating that direct contact between transplanted ectoderm and host neural plate was not required for neuralization of transplants to occur. These observations demonstrate t h a t competent ectoderm placed some distance from the neural plate can be neuralized, suggesting that a neuralizing signal extends to the region outside the neural plate.
Recombinants of Ectoderm with Neural and Mesodermal Tissue The neuralizing signals in the region outside the neural plate could be originating either from the mesoderm that underlies the transplants, or from the adjacent anterior neural plate. To distinguish between these possibilities, FLDx-labeled early gastrula ectoderm was
Homeogenetic Neural Induction
77
removed and cultured (1) in combination with anterior neural plate tissue from early neurulae, (2) in combination with the mesoderm underlying the presumptive lens region, or (3) as a control, by itself (Fig. 1B). Recombinants were assayed by staining with antibodies to NCAM. To verify that no contaminating dorsal mesoderm adhered to dissected neural plate tissue, the neural tissue was carefully examined under the dissecting microscope; even a few mesodermal cells would have been readily observed (see Materials and Methods). Recombinants of neural plate and ectoderm never contained any recognizable mesoderm cells, although it should be noted t h a t the cell types made by anterior mesoderm are not readily identified histologically. Additionally, some recombinants were made with presumptive neural plate tissue from stage 12 embryos, to which mesoderm does not adhere; the same results were obtained using either stage 12 or later stage 14 neural plate. In the recombinant experiments, summarized in Table 2 (also see Fig. 3), neural inductions were obtained in a high proportion of recombinants with neural plate tissue and occurred in only one case in recombinants containing dorsolateral mesoderm. Ectoderm cultured alone invariably formed compact masses of atypical epidermis (Winklbauer, 1988, and references therein). In the neural-ectodermal recombinants, the tissue derived from neural plate (which we refer to as host neural tissue) usually formed a morphologically discrete ring of neural tissue, resembling a neural tube, which stained strongly for N-CAM (Figs. 3A to 3C). The test ectoderm in all cases formed epidermis t h a t partially or completely surrounded the host neural tube, and, in many cases, formed small amounts of tissue t h a t were clearly N-CAM-positive. The neuralized test tissue in these recombinants was generally in direct contact with the host neural tissue and was in many cases incorporated into the neural tube. The amount of tissue t h a t is neuralized in recombinants is much smaller than t h a t neuralized in transplants to neural plate stage embryos (Figs. 3A to 3C). This suggests either t h a t the tissue contacts made in the recombinants are less than optimal for neural induction or t h a t tissues present in vivo, but absent from recombinants, may potentiate the inducing effect of the neural plate (see Discussion).
FIG. 2. Neural tissue induced from ectoderm transplants. Each row shows a case in which FLDx-labeled early gastrula ectoderm was transplanted to a neural plate stage embryo (as shown in Fig. la). In each row, the left figure shows a section viewed under differential interference contrast (DIC), the middle figure shows FLDx fluorescence, and the right figure shows immunofluorescence staining with anti-NCAM antibody. Note that both the host neural tube and host eye stain strongly for N-CAM. Abbreviations: e - eye, nt = neural tube, p = pharynx. Scale bar = 200 t~m. A-C: Transplant to the dorsolateral head region. Transplanted tissue (labeled tissue in B, marked by arrow) has fused to the host eye and stains with anti-N-CAM (arrow in C). D-F: Transplant to the lateral head area. The transplant (labeled tissue in E, marked by arrow) is located ventral to the level of the eye, but has still made a small neural-tube-like structure that stains positively for N-CAM (arrow in F). G-I: Transplant to anterior ventral ectoderm. The transplant (labeled tissue in H, marked by arrow) is located in the region of the ventral midline, and has formed no neural tissue, as judged by the absence of N-CAM staining of this tissue (I).
78
DEVELOPMENTAL BIOLOGY
TABLE 1 NEURALIZATIONOF TRANSPLANTS OF EARLY GASTRULAECTODERM TO NEURAL PLATE STAGE EMBRYOS
Transplant to:
Number scored
Number N-CAM positive (%)
Dorsal head ectoderm Lateral head ectoderm Ventral ectoderm (total) Anterior Posterior
22 13 21 14 7
20 (91%) 6 (46%) 2 (10%) 2 0
Note. Transplants were performed as described in Fig. la.
Because the induced neural tissue in recombinants did not form clearly recognizable neural structures (for example, neural tube), we wished to verify the neural nature of the induced tissue in recombinant experiments with a second neural marker, which would indicate that at least two neural-specific genes are expressed in induced tissue. To do this, we stained with the monoclonal antibody 2G9, which recognizes a neural-specific cell surface glycoprotein distinct from N-CAM (Jones and Woodland, 1989). Results of 2G9 staining (Table 2 and Figs. 3D to 3F) confirm that neural induction occurs in recombinants with neural plate tissue, but not with dorsolateral mesoderm. The 2G9 antigen is expressed later in normal embryonic development than N-CAM, indicating that later steps in neural differentiation are also triggered by homeogenetic neural induction. DISCUSSION
Our experiments show that competent ectoderm placed adjacent to the neural plate is neuralized and t h a t this neuralization occurs as a result of interactions of ectoderm with the neural plate, not with the mesoderm underlying the transplanted tissue. The neuralization of ectoderm transplanted to the head, but not in direct contact with the neural plate, indicates that neural induction can occur even when the inducing tissue is not in direct contact with the responding ectoderm. This observation bears on the mechanism of homeogenetic induction, as discussed below.
Homeogenetic Neural Induction: Neural Induction in the Absence of Mesoderm The experiments presented here address one of the main concerns t h a t has been raised about previous reports of homeogenetic neural induction, namely that induced ectoderm may have been exposed to mesoderm t h a t can act as a neural inducer (see Introduction). Although transplants to the head region are underlain by mesoderm, this mesoderm is unable to induce a neural
VOLUME147, 1991
response from ectoderm in recombinants, and neuralization of such transplants can be accounted for by signals from the neural plate. The possibility that ectoderm becomes neuralized solely because it comes into contact with dorsal mesoderm during the movements of neurulation (see Jones and Woodland, 1989) seems improbable based on two results. First, ectoderm transplants can be neuralized even when they lie at some distance from the neural tube; such ectoderm is unlikely to contact dorsal mesoderm during neurulation. Second, ectoderm is neuralized in recombinant experiments, in which mesoderm is not present. In any experiment on homeogenetic neural induction, there exists the possibility that, even in the absence of mesodermal contamination, neuralization of ectoderm occurs via mesoderm-derived neural-inducing factors. Such factors might be transmitted from mesoderm to presumptive neural ectoderm and remain adherent to neural ectoderm until it contacts competent ectoderm, which would then be induced by the passively transferred factors. Such passive transfer could lead to the neuralization observed in our recombinants. Passive transfer of peptide factors is unlikely in our experiments, since the neural tissue is dissected in the presence of proteolytic enzymes, although it remains possible that nonprotein factors may be involved. However, if passive transfer is occurring, it is difficult to explain how it could account for the neuralization of ectodermal transplants. Mesoderm-derived factors would have to be transmitted first to the overlying presumptive neural plate then be somehow transferred to the transplanted ectoderm which lies at some distance from the neural plate (Figs. 2D to 2F). Although we cannot rule out the possibility of passive transfer in recombinant experi-
TABLE 2 NEURALIZATION OF EARLY GASTRULA ECTODERM RECOMBINED WITH LATE GASTRULA/EARLY NEURULA TISSUES
N-CAM-stained
Tissues recombined ~ Ectoderm alone Ectoderm plus dorsolateral head mesoderm Ectoderm plus presumptive neural tissue
Number scored
2G9-stained
Number positive
Number scored
Number positive
9
0 (0%)
13
0 (0%)
11
1 (9%)
10
0 (0%)
19
13 (70%)
8
5 (63%)
Ectoderm was removed from late blastula/early gastrula embryos (stages 9-10), presumptive neural tissues were removed from late gastrula/early neurula embryos (stages 12-14), and mesoderm was removed from early neurula (stage 14 embryos), as shown in Fig. lb.
SERVETNICK AND GRAINGER
ments, we believe t h a t the results of transplant experiments render such a possibility implausible. In addition, we believe that the experiments of Nieuwkoop (1952), in which ectodermal flaps were neuralized at some distance from the mesoderm, taken together with recent observations of B. Gallagher and R. Grainger (unpublished data), that mesoderm is not migrating into such ectodermal flaps, argue strongly against passive transfer of inducers and indicate that the neuralization results from signaling within the ectodermal layer. Although passive transfer of neural-inducing factors seems unlikely, it remains possible that mesoderm-derived inducing factors are transmitted to the ectoderm, causing it to neuralize, and that these factors are then transmitted by the neuralized ectoderm to adjacent uninduced ectoderm (see Mangold and Spemann, 1927). However, this in itself would be a form of homeogenetic induction (see Gurdon, 1987). Alternatively, the neural ectoderm may itself actively synthesize inducing agents, causing adjacent competent ectoderm to neuralize. If so, as pointed out by Spemann (1938), homeogenetic neural induction would occur by a process distinct from the induction of neural tissue by mesoderm.
Signaling Mechanisms in Homeogenetic Induction The preponderance of evidence, provided in this paper as well as in previous work (Nieuwkoop, 1952; Albers, 1987), suggests t h a t neural induction can occur homeogenetically via a signal from the neural plate, transmitted through the plane of the ectoderm, t h a t can cause the neuralization of adjacent competent ectoderm. This signal is apparently capable of being transmitted through ectoderm that does not itself become neuralized (see also Albers, 1987). Although our recombinant experiments can be explained either by planar or by vertical signaling, the neuralization of transplants, in the absence of inducing activity of the underlying mesoderm, argues for planar signaling. In light of these conclusions, it should be noted that similar inferences have been made with regard to another form of homeogenetic induction, t h a t of mesoderm. Homeogenetic mesoderm induction occurs via signals within a tissue (Nishijima et aL, 1978; Kurihara and Sasaki, 1981; Suzuki et al., 1984), as does the dorsalization of mesoderm (reviewed by Smith, 1989b), which might also be considered a form of homeogenetic induction (in dorsalization, dorsal mesoderm induces ventral mesoderm to adopt a more dorsal fate). Cooke et al. (1987) have shown that homeogenetic mesoderm induction in recombinants is unlikely to be due to the passive transfer of mesoderm-inducing factors from induced tissue to uninduced tissue, as we have argued above for homeogenetic neural induction.
Homeogenetic Neural Induction
79
Homeogenetic Neural Induction May Occur Only under Restricted Conditions While our results provide evidence for homeogenetic neural induction, recently Jones and Woodland (1989) did not observe such induction in Xenopus, assaying for neural differentiation with the 2G9 antibody. The differences between our results and theirs may be due to one of several factors. First, the position of transplants in the host embryo may affect the ability of neural tissue to differentiate. Jones and Woodland transplanted presumptive neural tissue from midgastrula embryos into the ventral region of host embryos to look for the homeogenetic induction of adjacent host tissue; none was observed. However, the ventral region may well be an inhospitable environment for neural differentiation. There are differences in the ability of different regions of midgastrula ectoderm to neuralize in response to dorsal mesoderm (Machemer, 1932), indicating t h a t either the competence of the ectoderm, or the surrounding environment, may be of considerable importance in its response. Although the neural transplants of Jones and Woodland differentiated well in the ventral region, the ability of adjacent ventral ectoderm to neuralize may be poor, even though this same ectoderm might respond well to neural induction in another context. It is also possible that the neuralizing signal issuing from the small area of transplanted neural tissue in the experiments of Jones and Woodland is weaker than t h a t coming from the intact neural plate, which is the signal we are detecting in our transplant experiments. Jones and Woodland also performed recombinant experiments, but found t h a t homeogenetic neural induction did not occur in these recombinants. It may be t h a t the disposition of neural and ectodermal tissues relative to one another in recombinants affects the transmission of neural-inducing signals and t h a t the arrangement of tissues was different in the two studies. In our experiments, neural and ectodermal tissues were placed in apposition; these tissues presumably healed, at least in some cases, as contiguous sheets, in which at least some planar signals could be transmitted from neural to ectodermal tissue. Jones and Woodland sandwiched neural tissue between two sheets of ectoderm, an arrangement that may be less conducive to formation of contiguous neural-ectodermal sheets and subsequently less conducive to the transmission of planar signals. The amount of tissue t h a t is neuralized in our recombinants is much smaller than t h a t neuralized in transplants to neural plate stage embryos (Figs. 3A to 3F), suggesting that the tissue contacts made in the recombinants may be less than optimal for neural induction. As mentioned above, ectoderm and neural plate probably heal together as a contiguous sheet of tissue in our re-
80
DEVELOPMENTALBIObOGY
VOLUME147, 1991
FIG. 3. Neural tissue induced in recombinants. Each row shows a section of a recombinant (as shown in Fig. lb). In each row, the figure at left shows a section viewed with DIC, the middle figure shows FLDx fluorescence, and the right figure shows immunofluorescence (C stained with anti-N-CAM antibody; F, I stained with monoclonal antibody 2G9.) Abbreviations: e = eye tissue, n = neural tissue, m = mesoderm. Scale bars = 200 ~m. A-C: Recombinant of FLDx-labeled early gastrula (stage 10) ectoderm with stage 14 anterior neural plate stained with anti-N-CAM antibody. Note in (A) that the neural plate tissue has formed a ring of neural tissue that has the appearance of a neural tube (n; also compare A and C) and an eye (e) as judged by the presence of pigmented retina. Test ectoderm (labeled tissue in B) has formed a small projection from the neural tube (large arrow in B) that stains positively for N-CAM (arrow in C). (The small arrow at the bottom of B marks an area of weak FLDx staining within the eye tissue; this is apparently due to small amounts of transfer of FLDx from the test ectoderm as described under Materials
SERVETNICK AND G•AINGER
combinants, but this healing may not occur uniformly at all edges of the recombinant and the time required for such healing may preclude the efficient transmission of planar signals. Another explanation for the small size of induced neural structures is that tissues present in vivo, but absent from recombinants (for example, mesoderm), may potentiate the inducing effect of the neural plate. This has been shown to be the case in lens induction, in which mesoderm, which itself is incapable of inducing a lens, potentiates the ability of presumptive eye tissue to do so (Henry and Grainger, 1990). In sum, homeogenetic neural induction may be affected by both the regional nature of the responding ectoderm and by the arrangement of inducing and responding tissues. The conditions under which homeogenetic neural induction occurs may be sharply restricted; this may be why only small groups of cells are induced in recombinants in our experiments as well as in those of others (Deuchar, 1971; Grunz, 1990).
The Role of Homeogenetic Induction in Embryogenesis Our experiments argue that homeogenetic neural induction can occur in the Xenopus embryo; however, we have no information as to whether it actually does occur during normal development. Nieuwkoop and colleagues (Nieuwkoop, 1952, 1985; Albers, 1987) have argued that homeogenetic neural induction plays a crucial role in normal development, delineating the spatial extent of the neural plate. On the other hand, evidence indicates that the inducing capacity of dorsal mesoderm is sufficient to account for the spatial extent of the neural plate (Holtfreter, 1933; Jones and Woodland, 1989), and therefore homeogenetic induction need not be invoked to account for the neuralization that occurs in normal development. Instead, the signaling mechanisms involved in homeogenetic induction may reflect signaling within a tissue layer, which normally occurs within morphogenetic fields. These interactions are involved in establishing specific cell fates within a previously induced field (see Jacobson and Sater, 1988); however, it could be that as a result of experimental manipulation, these signals can cause competent ectoderm to be incorporated into the morphogenetic field. Such interactions within the mesoderm lead to the dorsalization of mesoderm which is originally specified as ventral (reviewed by Smith, 1989b); these dorsalizing signals may also be ca-
Homeogenetic Neural Induction
81
pable of homeogenetically inducing competent ectoderm to form mesoderm. Analogous interactions, which normally specify cell fates within the neural plate, may lead to homeogenetic neural induction of competent ectoderm. For example, Hemmati-Brivanlou et al. (1990) have proposed that the anterior notochord induces the expression of the engrailed gene in the overlying medial neural plate and that these engrailed-expressing medial neural cells then laterally signal adjacent neural plate cells to express the gene; such a lateral signal might be able to induce competent ectoderm to neuralize. Thus, homeogenetic induction may be a manifestation of the regulative ability of vertebrate embryos. While the signals which lead to homeogenetic neural induction might be involved in either neuralization or in subsequent regulation of neural cell fate, they might also be involved in the specification of placodal tissues, as proposed by Nieuwkoop (1963). According to this model (Nieuwkoop, 1985; Albers, 1987), a single signal is responsible for the induction of both neural and placodal tissues: the signal emanates from the dorsal midline and travels slowly through the ectoderm in a lateral direction. Because the competence of ectoderm changes rapidly during this period (Servetnick and Grainger, 1991), the signal elicits different responses in the ectoderm which it encounters, causing the induction of neural tissue early (and medially) and the induction of placodal tissue later (more laterally). The Nieuwkoop model specifically predicts that if the signal were to contact young ectoderm, it should elicit a neural response, even if surrounding ectoderm were no longer competent to respond by neuralizing. This prediction is borne out by our observations. Although a number of elements of the Nieuwkoop model remain to be tested, we have provided evidence here that a planar signal spreads from the neural plate during early development. We t h a n k Urs Rutihauser and Elizabeth Jones for kindly sharing antibodies; Betty Gallagher for communicating unpublished results; Beverly Smolich and Margaret Saha for comments on the manuscript; and Li-jun Wang, Margie Fox, and Les Cook for technical assistance. This work was supported by a National Research Service Award from the National Eye Institute to M.S. (EY-06173) and grants from the National Eye Institute to R.M.G. (EY-05542 and EY-06675).
Note added in proof Itoh and Kubota (Dev. Growth Differ. 33, 209216) have recently examined homeogenetic neural induction in Xeno-
and Methods.) D-F: Recombinant of FLDx-labeled late blastula (stage 9) ectoderm with stage 14 anterior neural plate stained with antibody 2G9. The neural-plate-derived tissue has formed a small neural mass in the interior of the recombinant (n; cf., panels A and C). Test ectoderm (labeled tissue in E) has been assimilated into the neural tissue (arrow in E); within this composite neural structure, both neural-plate-derived tissue and test ectoderm (arrow in F) stain positively with the 2G9 antibody. G-I: Recombinant of FLDx-labeled late blastula (stage 9) ectoderm and stage 1312dorsolateral head mesoderm (see Fig. lb). In many such ectoderm-mesoderm recombinants, we have observed that, despite the presence of sufficient ectoderm to completely surround the mesoderm, the ectoderm does not do so, but the mesoderm (m) protrudes through the ectodermal ball. Neither the test ectoderm (labeled tissue in H) nor the mesoderm stains with the 2G9 antibody.
82
DEVELOPMENTALBIOLOGY
pus explant recombinants and reported results similar to those presented here. REFERENCES ALBERS, B. (1987). Competence as the main factor determining the size of the neural plate. Dev. Growth Differ. 29, 535-545. BALAK, K., JACOBSON,M., SUNSHINE,J., and RUTISHAUSER,U. (1987). Neural cell adhesion molecule expression in Xenopus embryos. Dev. Biol. 119, 540-550. COOKE,J., SMITH,J. C., SMITH,E. J., and YAQOOB,M. (1987). The organization of mesodermal pattern in Xenopus laevis: Experiments using a Xenopus mesoderm-inducing factor. Development 101, 893-908. DEUCHAR,E. M. (1971). Transfer of the primary induction stimulus by small numbers of amphibian ectoderm cells. Acta Embryol. Exp. 2, 93-101. DIXON, J. E., and KINTNER, C. R. (1989). Cellular contacts required for neural induction in Xenopus embryos: Evidence for two signals. Development 106, 749-757. GIMLICH, R. L., and BRAUN,J. (1985). Improved fluorescent compounds for tracing cell lineage. Dev. BioL 109, 509-514. GRUNZ, H. (1985). Information transfer during embryonic induction in amphibians. J. Embryol Exp. Morphol. 89, (Suppl.), 349-364. GRUNZ, H. (1990). Homoiogenetic neural inducing activity of the presumptive neural plate of Xenopus laevis. Dev. Growth Differ. 32, 583-589. GRUNZ, H., and TACKE, L. (1989). Neural differentiation of Xenopus laevis ectoderm takes place after disaggregation and delayed reaggregation without inducer. Cell Differ Dev. 28, 211-218. GURDON,J. B. (1987). Embryonic induction--Molecular prospects. Development 99, 285-306. HAUSEN, P., WANG, Y. H., DREYER, C., and STICK, R. (1985). Distribution of nuclear proteins during maturation of the Xenopus oocyte. J. Embryol. Exp. Morphol. 89, (suppl.), 17-34. HEMMATI-BRIVANLOU, A., STEWART, R. M., and HARLAND, R. M. (1990). Region-specific neural induction of an engrailed protein by anterior notochord in Xenopus. Science 250, 800-802. HENRY, J. J., and GRAINGER,R. M. (1987). Inductive interactions in the spatial and temporal restriction of lens-forming potential in embryonic ectoderm of Xenopus laevis. Dev. Biol. 124, 200-214. HENRY, J. J., and GRAINGER, R. M. (1990). Early tissue interactions leading to embryonic lens formation in Xenopus laevis. Dev. Biol. 141, 149-163. HOLTFRETER, J. (1933). Nicht typische Gestaltungsbewegungen, sondern Induktionsvorg~inge bedingen medullare Entwicklung von Gastrulaektoderm. Wilhelm Roux Arch. EntwMech. Org. 127, 591618. HOLTFRETER,J. (1938). Ver~inderungen der Reaktionsweise im alternden isolierten Gastrulaektoderm. Wilhelm Roux Arch. EntwMech. Org. 138, 163-196. HOLTFRETER, J. (1947). Neural induction in explants which have passed through a sublethal cytolysis. J. Exp. Zool. 106, 197-222. JACOBSON,A. G., and SAWER,A. K. (1988). Features of embryonic induction. Development 104, 341-359. JONES, E. A., and WOODLAND,H. R. (1989). Spatial aspects of neural induction in Xenopus laevis. Development 167, 785-791. KURIHARA,K., and SASAKI,N. (1981). Transmission of homoiogenetic induction in presumptive ectoderm of newt embryo. Dev. Growth Differ. 23, 361-369.
VOLUME147, 1991
MACHEMER, H. (1932). Experimentelle Untersuchung tiber die Induktionsleistungen der oberen Urmundlippe in ~tlteren Urodelenkeimen. Wilhelm Roux Arch. EntwMech. Org. 126, 391-456. MANGOLD,O. (1929). Experimente zur Analyse der Determination und Induktion der Medullarplatte. Wilhelm Roux Arch. EntwMech. Org. 117, 586-696. MANGOLD, O. (1933). Uber die Induktionsf~higkeit der verschiedenen Bezirke der Neurula yon Urodelen. Naturwissenschaften 43, 761766. MANGOLD,O., and SPEMANN,H. (1927). Uber Induktion yon Medullarplatte durch Medullarplatte im jtingeren Keim, ein Beispiel homSogenetischer oder assimilatorischer Induktion. Wilhelm Roux Arch. EntwMech. Org. l l l , 341-422. NEW, H. V., and SMITH, J. C. (1990). Inductive interactions in early amphibian development. Curr. Opi~ Cell Biol. 2, 969-974. NIEUWKOOP, P. D., et al. (1952). Activation and organization of the central nervous system in amphibians. J. Exp. ZooL 120, 1-108. NIEUWKOOP, P. D. (1963). Pattern formation in artificially activated ectoderm (Rana pipiens and Ambystoma punctatum). Dev. Biol. 7, 255-279. NIEUWKOOP, P. D. (1985). Inductive interactions in early amphibian development and their general nature. J. Embryol. Exp. MorphoL 89, (Suppl.), 333-347. NIEUWKOOP, P. D., and ALBERS, B. (1990). The role of competence in the cranio-caudal segregation of the central nervous system. Dev. Growth Differ. 32, 23-31. NIEUWKOOP, P. D., and FABER, J. (1967). "Normal Table of Xenopus laevis (Daudin)." North-Holland, Amsterdam. NISHIJIMA, K., NODA, S., KURIHARA,K., and SASAKI,N. (1978). Differentiation of partially mesodermalized ectoderm: Homoiogenetic and heterogenetic induction by primarily induced part of the ectoderm. Dev. Growth Differ. 20, 275-281. ROMEIS, B. (1968). "Mikroskopische Technik." p. 81. Oldenburg, Munich/Vienna. RuIz I ALTABA,A. (1990). Neural expression of the Xenopus homeobox gene Xhox3: Evidence for a patterning neural signal that spreads through the ectoderm. Development 108, 595-604. SAVAGE,R., and PHILLIPS, C. R. (1989). Signals from the dorsal blastopore lip region during gastrulation bias the ectoderm toward a nonepidermal pathway of differentiation in Xenopus laevis. Dev. Biol. 133, 157-168. SAXEN, L. (1989). Neural induction. Int. J. Dev. Biol. 33, 21-48. SERVETNICK,M., and GRAINGER,R. M. (1991). Changes in neural and lens competence in Xenopus ectoderm: Evidence for an autonomous developmental timer. Development 112, 177-188. SHARPE, C. R., and GURDON,J. B. (1990). The induction of anterior and posterior neural genes in Xenopus laevis. Development 109, 765-774. SLACK, J. M. W. (1984). Regional biosynthetic markers in the early amphibian embryo. J. Embryol. Exp. Morphol. 80, 289-319. SMITH, J. C. (1989a). Induction and early amphibian development. Cu~. Opin. Cell Biol. 1, 1061-1070. SMITH, J. C. (1989b). Mesoderm induction and mesoderm-inducing factors in early amphibian development. Development 105, 665-677. SPEMANN, H. (1938). "Embryonic Development and Induction." Hafner, New York. SUZUKI,A. S., MIFUNE, Y., and KAN~DA,T. (1984). Germ layer interactions in pattern formation of amphibian mesoderm during primary embryonic induction. Dev. Growth Differ. 26, 81-94. WINKLBAUER, R. (1988). Cell proliferation in ectodermal explants from Xenopus embryos. Roux's Arch. Devl. Biol. 197, 141-147.
Homeogenetic Neural Induction in Xenopus M A R C SERVETNICK AND ROBERT M. G R A I N G E R
Department of Biology, University of Virginia, Charlottesville, Virginia 22901 Accepted May 28, 1991 Neural induction is known to involve an interaction of ectoderm with dorsal mesoderm during gastrulation, but several kinds of studies have argued that competent ectoderm can also be neuralized via an interaction with previously neuralized tissue, a process termed homeogenetic neural induction. Although homeogenetic neural induction has been proposed to play an important role in the normal induction of neural tissue, this process has not been subjected to detailed study using tissue recombinants and molecular markers. We have examined the question of homeogenetic neural induction in Xenopus embryos, both in transplant and recombinant experiments, using the expression of two neural antigens to assay the response. When ectoderm that is competent to be neuralized is transplanted to the region adjacent to the neural plate of early neurula embryos, it forms neural tissue, as assayed by staining with antibodies against the neural cell adhesion molecule, N-CAM. Transplants to the ventral region, far from the neural plate, do not express N-CAM, indicating that neuralization is not occurring as a result of the transplantation procedure itself. Because this response might be occurring as a result of interactions of ectoderm with either adjacent neural plate tissue, or with underlying dorsolateral mesoderm, recombinant experiments were performed to determine the source of the neuralizing signal. Ectoderm cultured in combination with neural plate tissue alone expresses neural markers, while ectoderm cultured in combination with dorsolateral mesoderm does not. We conclude that neural tissue can homeogenetically induce competent ectoderm to form neural tissue and argue that this induction occurs via planar signaling within the ectoderm, a mechanism that, in normal development, may be involved in interactions within presumptive neural ectoderm or in specifying structures that lie near the neural plate. © 1991AcademicPress,Inc.
Another form of signaling may also be important in the induction of the neural plate, namely homeogenetic neural induction, the induction of neural tissue by neural tissue. Homeogenetic neural induction has been proposed to play a central role in the normal formation of the neural plate (Nieuwkoop, 1952, 1985; Albers, 1987) and in specifying the regional nature of induced neural tissue (Nieuwkoop, 1952; Nieuwkoop and Albers, 1990). Although several kinds of experiments have provided evidence for homeogenetic neural induction, these experiments are subject to criticisms that render their interpretation problematic. These criticisms are, first, that the responding tissue in these experiments may have been subject to some influence from mesodermal tissues in addition to those from neural tissue, and, second, that the neural response of the tissue was not verified by the use of unambiguous markers of neural differentiation. Homeogenetic induction was first observed in T r i t u r u s by Mangold and Spemann (1927; s e e also Mangold, 1929, 1933), who placed pieces of neural plate tissue into the blastocoel of a host embryo, which often formed neural tissue in the region of the implant. Such blastocoel implants are difficult to interpret: the precise location, manner of contact, and time of contact are not well controlled in such cases, with the result that the implanted tissue may be acting on ectoderm that has already been acted upon by other tissues, and that may
INTRODUCTION
Although the induction of the central nervous system is one of the primary patterning events in vertebrate development, the process of neural induction is not well understood. Classical evidence (reviewed by Spemann, 1938) indicates that, during gastrulation, dorsal mesoderm transmits signals vertically to the overlying ectoderm, inducing it to form neural tissue; such vertical signaling by the mesoderm has recently been shown to be sufficient to induce the expression of neural-specific genes (Sharpe and Gurdon, 1990; Hemmati-Brivanlou et al., 1990). Recent evidence indicates that a second signal may also be involved in this process (reviewed by Smith, 1989a; New and Smith, 1990). These studies suggest that, prior to invagination of the dorsal mesoderm during gastrulation, a planar signal originating from the region of the dorsal blastopore lip passes laterally through the plane of tissue into the ectoderm. The planar signal has been reported to bias dorsal portions of the ectoderm toward a neural pathway of development (Savage and Phillips, 1989), to specify dorsal ectoderm cells to adopt a neural fate (Dixon and Kintner, 1989), to act synergistically with the vertical signal to elicit maximal expression of neural-specific genes (Dixon and Kintner, 1989), and to play a role in patterning the forming neural plate (Ruiz i Altaba, 1990). 73
0012-1606/91 $3.00 Copyright© 1991by AcademicPress,Inc. All rightsof reproductionin any formreserved.
74
DEVELOPMENTALBIOLOGY
not be completely naive with respect to neural induction. This criticism is especially pertinent given recent results showing that neuralization can be the result of two signals acting synergistically (Dixon and Kintner, 1989). In experiments t h a t circumvent the host effects associated with blastocoel implants, Nieuwkoop (1952), using Ambystoma, Triturus, and Pleurodeles, inserted flaps of folded ectoderm into the presumptive neural plate. He observed neural tissue formed at some distance from the surface of the embryo, arguing that a homeogenetic neuralizing signal can pass through the ectoderm. These experiments have been criticized (Jones and Woodland, 1989) because the ectodermal flaps may have been exposed to signals from the underlying mesoderm, even if only briefly. Recently, Albers (1987; see also Holtfreter, 1933, 1938) performed transplant experiments in Ambystoma in which competent ectoderm was transplanted so as to span the neural-epidermal boundary. The transplanted ectoderm showed neuralization beyond the normal borders of the neural plate, suggesting homeogenetic induction. In these experiments, it has been suggested that the transplanted ectoderm may have come into contact with dorsal mesoderm during neurulation, which could have resulted in neural induction of the transplants (Jones and Woodland, 1989); furthermore, these studies did not test the possibility that the mesoderm outside the neural region, which underlies the transplants, possesses neural-inducing ability. In experiments that exclude any possible influence of mesoderm, Xenopus neural tissue was recombined in vitro with competent ectoderm (Deuchar, 1971; Grunz, 1990). Although ectodermal cells were incorporated into neural structures, no biochemical markers of neural differentiation were used in these studies, raising the possibility that the cells incorporated into neural structures were not themselves neuralized. Such concerns are especially relevant in light of a recent report (Jones and Woodland, 1989) t h a t did not detect homeogenetic neural induction in Xenopus, using both a neural-specific antibody marker and host and donor marking of recombined tissues in transplants as well as recombinants. Here we have reexamined homeogenetic neural induction in Xenopus. Our transplant and recombinant experiments, in conjunction with cell lineage tracing, argue that neural plate tissue can induce neural tissue from competent ectoderm in the absence of mesoderm. We have also shown that the responding cells express two different molecular markers of neural differentiation.
VOLUME147, 1991
MATERIALS AND METHODS
Tissue transplants and recombinants. Xenopus laevis embryos were staged according to Nieuwkoop and Faber (1967). Animal cap tissue was manually removed from FLDx-labeled1 (fluoresceinated dextran; Molecular Probes, Junction City, OR) early gastrula (stage 10) embryos, using sharpened watchmaker's forceps, and transplants were made as described in Servetnick and Grainger (1991). For recombinant experiments, anterior neural tissue was removed either according to the procedure described by Slack (1984), in which ectoderm is dissected in the presence of 0.01% trypsin, followed by washing in 0.02% soy bean trypsin inhibitor, or by dissecting ectoderm in the presence of 1.5 mg/ml collagenase, followed by washing of dissected tissue in 5 mg/ml bovine serum albumin (all proteins from Sigma; all in 100% Steinberg's solution); no differences were observed in results of experiments using the two procedures. Dorsolateral mesoderm was usually dissected without the use of proteolytic enzymes, although it was occasionally isolated from embryos that had been placed in enzyme solutions to isolate other tissues. Dissected neural plate tissue was examined under the dissecting microscope before being apposed to ectoderm in order to verify that no adherent mesoderm cells were present. Recombinants of ectoderm and neural tissue never showed any recognizable mesodermal tissue (notochord or muscle). Although such histological analysis alone cannot rule out the presence of small amounts of head mesoderm (which does not form readily identifiable cell types, and for which no molecular markers are currently available), contamination by even a few mesodermal cells would have been detected microscopically at the time of dissection. Recombinants made with dorsolateral mesoderm were devoid of notochord or muscle (derivatives of dorsal mesoderm) except in a single case, in which mesoderm from a late gastrula (early stage 12) embryo formed notochord; this was probably due to dissection of the wrong region of mesoderm and this case was discarded. Culture of recombinants. Explanted tissues were placed on top of one another on a base of clay and held in tight contact by a curved glass bridge (Henry and Grainger, 1987) for 30-60 min. Recombinants were left in 100% Steinberg's solution overnight, after which the concentration was reduced to ~50% Steinberg's solution. Histology and immunofluorescence. Embryos or re-
1Abbreviationsused:DIC:differentialinterferencecontrast;FLDx: fluoresceinateddextran; N-CAM:neural cell adhesionmolecule;PLR: presumptive lens region.
SERVETNICK AND GRAINGER
combinants were fixed at stages 39-42 (stages of recombinants were monitored using unoperated sibling controls) in either Romeis fixative (for N-CAM staining; Romeis, 1968; Hausen et al., 1985) or in a solution of 2.0 ml 5% trichloroacetic acid, 1.5 ml 37% formaldehyde, 2.5 ml distilled water (for 2G9 staining). Fixation, embedding, sectioning, and staining were as described in Servetnick and Grainger (1991). N-CAM antibodies (provided by U. Rutishauser; see Balak et al., 1987) were used at a dilution of 1:25; monoclonal antibody 2G9 tissue culture supernatant (provided by E. Jones; see Jones and Woodland, 1989) was used undiluted. Cell lineage tracing. To trace the origin of cells in all experiments, one-cell embryos were injected with FLDx (Gimlich and Braun, 1985), and FLDx-labeled ectoderm was used as the test ectoderm in all experiments. We have observed, in some experiments, small amounts of FLDx fluorescence in non-FLDx-injected cells at the junction of labeled and unlabeled tissues, resulting in an apparent gradient of FLDx fluorescence (see, for example, Fig. 3B). The cells that show this effect exhibit weak nuclear fluorescence, rather than the strong, widespread cytoplasmic fluorescence observed in FLDx-injected tissue. This effect is most pronounced in recombinants, although we have also observed it in several transplant experiments in which FLDx-labeled ectoderm transplants are incorporated into the host neural tube. We speculate that it may be due to death of some labeled cells, followed by uptake of released FLDx. Nevertheless, labeled and unlabeled cells are generally easily distinguishable from one another. In the few cases in which there was any ambiguity in identification of cells, these cases were discarded. In scoring experiments, we have been careful to score as positive only those tissues that are positive for antibody staining and t h a t consist of significant blocks of cells exhibiting the bright cytoplasmic fluorescence diagnostic of FLDx-injected cells. RESULTS
Neuralization of Ectoderm Transplants Placed Outside the Neural Plate During the course of experiments on lens induction, we transplanted ectoderm to the presumptive lens region (PLR) of neural plate (stage 14) embryos (Fig. la). Transplanted early gastrula (stage 10) ectoderm, which is competent to form neural tissue but is not yet competent to form lens (Servetnick and Grainger, 1991), often neuralized, despite its location outside the neural plate. The neuralization of this ectoderm was confirmed by staining with an antibody against Xenopus neural cell adhesion molecule, N-CAM, which is expressed specifically in neural tissue at the stages assayed (Balak et al.,
Homeogenetic Neural Induction
75
a
removeectoderm from FLDxlabelled early gastrula
b
transplantto headectodermor ventralectodermof neuralplate stagehost
~
ectodermalone
/ ectoderm ~
y
removeectoderm from FLDx-rabelled early gastrula
anterior neurarplate
~
~
/
ectoderm +
dorsolatera] headmesoderm
FIG. 1. Transplant and recombinant experiments to assay for neural induction. (a) Ectoderm was removed from FLDx-labeled early gastrula (stage 10) embryos and transplanted into a neural plate stage (stage 14) embryo from which the corresponding portion of ectoderm had been removed. Transplants were made to the dorsolateral head region (presumptive lens region), or to the anterior ventral region, as shown. In addition, transplants were made to lateral head ectoderm, corresponding to the region lying between the two areas shown in the figure, and to posterior ventral regions (not shown). (b) Ectoderm was removed from FLDx-labeled early gastrula (stage 10) embryos and cultured either alone, in combination with anterior neural plate tissue, or in combination with mesoderm underlying the dorsolateral head region. The diagram shows ectoderm from a stage 10 embryo and neural plate tissue and head mesoderm from a neural plate stage embryo. However, both late blastula (stage 9) and early gastrula (stage 10) embryos were used as ectoderm donors, and embryos from stages 12 (late gastrula stage) to 14 (neural plate stage) were used as the source of neural plate. In all experiments, embryos or recombinants were cultured to the equivalent of swimming tadpole stages (stages 39-42), fixed, sectioned, and stained by immunofluorescence with antibodies t h a t recognize either N-CAM or the neural antigen 2G9.
1987). An example of such a transplant is shown in Figs. 2A to 2C. The observed neuralization might have occurred as a response of transplanted competent ectoderm to specific inductive signals, or it could have been evoked by the transplantation procedure itself; such "autoneuralization" of ectoderm is known to occur in response to nonspecific stimuli in many species (Holtfreter, 1947; Saxen, 1989), although it is much more difficult to elicit such a response in Xenopus (Grunz, 1985; see also Grunz and Tacke, 1989). To address this point, we compared ectoderm transplants to the PLR (in the dorsolateral
76
DEVELOPMENTAL BIOLOGY
VOLUME147, 1991
SERVETNICKAND GRAINGER
head region, just outside the neural plate) with transplants made to other regions of host embryos. In contrast to transplants to the PLR, transplants placed in the ventral region rarely neuralized (Figs. 2G to 2I). Results of these experiments are summarized in Table 1. In the two cases in which ectoderm transplanted to the ventral region did show a neural response, this was because the ectoderm transplants extended unusually far dorsally (presumably as a result of the extensive stretching of ectoderm during neurulation), reaching the level of the midpharynx. Transplants t h a t were made to lateral head ectoderm (intermediate between the dorsolateral and ventral regions) showed a variable response. In these, the transplants that were located more dorsally formed neural tissue, whereas more ventrally located transplants did not; all of the positive responses observed in this series occurred dorsal to the level of the midpharynx. One of the ventral-most cases of neuralization that we have observed is shown in Figs. 2D to 2F. Presumably, the ectoderm is responding to inductive signals, and these are present in dorsal, but not ventral, regions of the embryo. In transplants to the PLR, a small amount of host ectoderm was often present between the host neural plate and the transplant. In most of these cases, as well as in transplants made to regions further from the neural plate (e.g., Figs. 2D to 2F), the transplants were nevertheless induced to form neural tissue, indicating that direct contact between transplanted ectoderm and host neural plate was not required for neuralization of transplants to occur. These observations demonstrate t h a t competent ectoderm placed some distance from the neural plate can be neuralized, suggesting that a neuralizing signal extends to the region outside the neural plate.
Recombinants of Ectoderm with Neural and Mesodermal Tissue The neuralizing signals in the region outside the neural plate could be originating either from the mesoderm that underlies the transplants, or from the adjacent anterior neural plate. To distinguish between these possibilities, FLDx-labeled early gastrula ectoderm was
Homeogenetic Neural Induction
77
removed and cultured (1) in combination with anterior neural plate tissue from early neurulae, (2) in combination with the mesoderm underlying the presumptive lens region, or (3) as a control, by itself (Fig. 1B). Recombinants were assayed by staining with antibodies to NCAM. To verify that no contaminating dorsal mesoderm adhered to dissected neural plate tissue, the neural tissue was carefully examined under the dissecting microscope; even a few mesodermal cells would have been readily observed (see Materials and Methods). Recombinants of neural plate and ectoderm never contained any recognizable mesoderm cells, although it should be noted t h a t the cell types made by anterior mesoderm are not readily identified histologically. Additionally, some recombinants were made with presumptive neural plate tissue from stage 12 embryos, to which mesoderm does not adhere; the same results were obtained using either stage 12 or later stage 14 neural plate. In the recombinant experiments, summarized in Table 2 (also see Fig. 3), neural inductions were obtained in a high proportion of recombinants with neural plate tissue and occurred in only one case in recombinants containing dorsolateral mesoderm. Ectoderm cultured alone invariably formed compact masses of atypical epidermis (Winklbauer, 1988, and references therein). In the neural-ectodermal recombinants, the tissue derived from neural plate (which we refer to as host neural tissue) usually formed a morphologically discrete ring of neural tissue, resembling a neural tube, which stained strongly for N-CAM (Figs. 3A to 3C). The test ectoderm in all cases formed epidermis t h a t partially or completely surrounded the host neural tube, and, in many cases, formed small amounts of tissue t h a t were clearly N-CAM-positive. The neuralized test tissue in these recombinants was generally in direct contact with the host neural tissue and was in many cases incorporated into the neural tube. The amount of tissue t h a t is neuralized in recombinants is much smaller than t h a t neuralized in transplants to neural plate stage embryos (Figs. 3A to 3C). This suggests either t h a t the tissue contacts made in the recombinants are less than optimal for neural induction or t h a t tissues present in vivo, but absent from recombinants, may potentiate the inducing effect of the neural plate (see Discussion).
FIG. 2. Neural tissue induced from ectoderm transplants. Each row shows a case in which FLDx-labeled early gastrula ectoderm was transplanted to a neural plate stage embryo (as shown in Fig. la). In each row, the left figure shows a section viewed under differential interference contrast (DIC), the middle figure shows FLDx fluorescence, and the right figure shows immunofluorescence staining with anti-NCAM antibody. Note that both the host neural tube and host eye stain strongly for N-CAM. Abbreviations: e - eye, nt = neural tube, p = pharynx. Scale bar = 200 t~m. A-C: Transplant to the dorsolateral head region. Transplanted tissue (labeled tissue in B, marked by arrow) has fused to the host eye and stains with anti-N-CAM (arrow in C). D-F: Transplant to the lateral head area. The transplant (labeled tissue in E, marked by arrow) is located ventral to the level of the eye, but has still made a small neural-tube-like structure that stains positively for N-CAM (arrow in F). G-I: Transplant to anterior ventral ectoderm. The transplant (labeled tissue in H, marked by arrow) is located in the region of the ventral midline, and has formed no neural tissue, as judged by the absence of N-CAM staining of this tissue (I).
78
DEVELOPMENTAL BIOLOGY
TABLE 1 NEURALIZATIONOF TRANSPLANTS OF EARLY GASTRULAECTODERM TO NEURAL PLATE STAGE EMBRYOS
Transplant to:
Number scored
Number N-CAM positive (%)
Dorsal head ectoderm Lateral head ectoderm Ventral ectoderm (total) Anterior Posterior
22 13 21 14 7
20 (91%) 6 (46%) 2 (10%) 2 0
Note. Transplants were performed as described in Fig. la.
Because the induced neural tissue in recombinants did not form clearly recognizable neural structures (for example, neural tube), we wished to verify the neural nature of the induced tissue in recombinant experiments with a second neural marker, which would indicate that at least two neural-specific genes are expressed in induced tissue. To do this, we stained with the monoclonal antibody 2G9, which recognizes a neural-specific cell surface glycoprotein distinct from N-CAM (Jones and Woodland, 1989). Results of 2G9 staining (Table 2 and Figs. 3D to 3F) confirm that neural induction occurs in recombinants with neural plate tissue, but not with dorsolateral mesoderm. The 2G9 antigen is expressed later in normal embryonic development than N-CAM, indicating that later steps in neural differentiation are also triggered by homeogenetic neural induction. DISCUSSION
Our experiments show that competent ectoderm placed adjacent to the neural plate is neuralized and t h a t this neuralization occurs as a result of interactions of ectoderm with the neural plate, not with the mesoderm underlying the transplanted tissue. The neuralization of ectoderm transplanted to the head, but not in direct contact with the neural plate, indicates that neural induction can occur even when the inducing tissue is not in direct contact with the responding ectoderm. This observation bears on the mechanism of homeogenetic induction, as discussed below.
Homeogenetic Neural Induction: Neural Induction in the Absence of Mesoderm The experiments presented here address one of the main concerns t h a t has been raised about previous reports of homeogenetic neural induction, namely that induced ectoderm may have been exposed to mesoderm t h a t can act as a neural inducer (see Introduction). Although transplants to the head region are underlain by mesoderm, this mesoderm is unable to induce a neural
VOLUME147, 1991
response from ectoderm in recombinants, and neuralization of such transplants can be accounted for by signals from the neural plate. The possibility that ectoderm becomes neuralized solely because it comes into contact with dorsal mesoderm during the movements of neurulation (see Jones and Woodland, 1989) seems improbable based on two results. First, ectoderm transplants can be neuralized even when they lie at some distance from the neural tube; such ectoderm is unlikely to contact dorsal mesoderm during neurulation. Second, ectoderm is neuralized in recombinant experiments, in which mesoderm is not present. In any experiment on homeogenetic neural induction, there exists the possibility that, even in the absence of mesodermal contamination, neuralization of ectoderm occurs via mesoderm-derived neural-inducing factors. Such factors might be transmitted from mesoderm to presumptive neural ectoderm and remain adherent to neural ectoderm until it contacts competent ectoderm, which would then be induced by the passively transferred factors. Such passive transfer could lead to the neuralization observed in our recombinants. Passive transfer of peptide factors is unlikely in our experiments, since the neural tissue is dissected in the presence of proteolytic enzymes, although it remains possible that nonprotein factors may be involved. However, if passive transfer is occurring, it is difficult to explain how it could account for the neuralization of ectodermal transplants. Mesoderm-derived factors would have to be transmitted first to the overlying presumptive neural plate then be somehow transferred to the transplanted ectoderm which lies at some distance from the neural plate (Figs. 2D to 2F). Although we cannot rule out the possibility of passive transfer in recombinant experi-
TABLE 2 NEURALIZATION OF EARLY GASTRULA ECTODERM RECOMBINED WITH LATE GASTRULA/EARLY NEURULA TISSUES
N-CAM-stained
Tissues recombined ~ Ectoderm alone Ectoderm plus dorsolateral head mesoderm Ectoderm plus presumptive neural tissue
Number scored
2G9-stained
Number positive
Number scored
Number positive
9
0 (0%)
13
0 (0%)
11
1 (9%)
10
0 (0%)
19
13 (70%)
8
5 (63%)
Ectoderm was removed from late blastula/early gastrula embryos (stages 9-10), presumptive neural tissues were removed from late gastrula/early neurula embryos (stages 12-14), and mesoderm was removed from early neurula (stage 14 embryos), as shown in Fig. lb.
SERVETNICK AND GRAINGER
ments, we believe t h a t the results of transplant experiments render such a possibility implausible. In addition, we believe that the experiments of Nieuwkoop (1952), in which ectodermal flaps were neuralized at some distance from the mesoderm, taken together with recent observations of B. Gallagher and R. Grainger (unpublished data), that mesoderm is not migrating into such ectodermal flaps, argue strongly against passive transfer of inducers and indicate that the neuralization results from signaling within the ectodermal layer. Although passive transfer of neural-inducing factors seems unlikely, it remains possible that mesoderm-derived inducing factors are transmitted to the ectoderm, causing it to neuralize, and that these factors are then transmitted by the neuralized ectoderm to adjacent uninduced ectoderm (see Mangold and Spemann, 1927). However, this in itself would be a form of homeogenetic induction (see Gurdon, 1987). Alternatively, the neural ectoderm may itself actively synthesize inducing agents, causing adjacent competent ectoderm to neuralize. If so, as pointed out by Spemann (1938), homeogenetic neural induction would occur by a process distinct from the induction of neural tissue by mesoderm.
Signaling Mechanisms in Homeogenetic Induction The preponderance of evidence, provided in this paper as well as in previous work (Nieuwkoop, 1952; Albers, 1987), suggests t h a t neural induction can occur homeogenetically via a signal from the neural plate, transmitted through the plane of the ectoderm, t h a t can cause the neuralization of adjacent competent ectoderm. This signal is apparently capable of being transmitted through ectoderm that does not itself become neuralized (see also Albers, 1987). Although our recombinant experiments can be explained either by planar or by vertical signaling, the neuralization of transplants, in the absence of inducing activity of the underlying mesoderm, argues for planar signaling. In light of these conclusions, it should be noted that similar inferences have been made with regard to another form of homeogenetic induction, t h a t of mesoderm. Homeogenetic mesoderm induction occurs via signals within a tissue (Nishijima et aL, 1978; Kurihara and Sasaki, 1981; Suzuki et al., 1984), as does the dorsalization of mesoderm (reviewed by Smith, 1989b), which might also be considered a form of homeogenetic induction (in dorsalization, dorsal mesoderm induces ventral mesoderm to adopt a more dorsal fate). Cooke et al. (1987) have shown that homeogenetic mesoderm induction in recombinants is unlikely to be due to the passive transfer of mesoderm-inducing factors from induced tissue to uninduced tissue, as we have argued above for homeogenetic neural induction.
Homeogenetic Neural Induction
79
Homeogenetic Neural Induction May Occur Only under Restricted Conditions While our results provide evidence for homeogenetic neural induction, recently Jones and Woodland (1989) did not observe such induction in Xenopus, assaying for neural differentiation with the 2G9 antibody. The differences between our results and theirs may be due to one of several factors. First, the position of transplants in the host embryo may affect the ability of neural tissue to differentiate. Jones and Woodland transplanted presumptive neural tissue from midgastrula embryos into the ventral region of host embryos to look for the homeogenetic induction of adjacent host tissue; none was observed. However, the ventral region may well be an inhospitable environment for neural differentiation. There are differences in the ability of different regions of midgastrula ectoderm to neuralize in response to dorsal mesoderm (Machemer, 1932), indicating t h a t either the competence of the ectoderm, or the surrounding environment, may be of considerable importance in its response. Although the neural transplants of Jones and Woodland differentiated well in the ventral region, the ability of adjacent ventral ectoderm to neuralize may be poor, even though this same ectoderm might respond well to neural induction in another context. It is also possible that the neuralizing signal issuing from the small area of transplanted neural tissue in the experiments of Jones and Woodland is weaker than t h a t coming from the intact neural plate, which is the signal we are detecting in our transplant experiments. Jones and Woodland also performed recombinant experiments, but found t h a t homeogenetic neural induction did not occur in these recombinants. It may be t h a t the disposition of neural and ectodermal tissues relative to one another in recombinants affects the transmission of neural-inducing signals and t h a t the arrangement of tissues was different in the two studies. In our experiments, neural and ectodermal tissues were placed in apposition; these tissues presumably healed, at least in some cases, as contiguous sheets, in which at least some planar signals could be transmitted from neural to ectodermal tissue. Jones and Woodland sandwiched neural tissue between two sheets of ectoderm, an arrangement that may be less conducive to formation of contiguous neural-ectodermal sheets and subsequently less conducive to the transmission of planar signals. The amount of tissue t h a t is neuralized in our recombinants is much smaller than t h a t neuralized in transplants to neural plate stage embryos (Figs. 3A to 3F), suggesting that the tissue contacts made in the recombinants may be less than optimal for neural induction. As mentioned above, ectoderm and neural plate probably heal together as a contiguous sheet of tissue in our re-
80
DEVELOPMENTALBIObOGY
VOLUME147, 1991
FIG. 3. Neural tissue induced in recombinants. Each row shows a section of a recombinant (as shown in Fig. lb). In each row, the figure at left shows a section viewed with DIC, the middle figure shows FLDx fluorescence, and the right figure shows immunofluorescence (C stained with anti-N-CAM antibody; F, I stained with monoclonal antibody 2G9.) Abbreviations: e = eye tissue, n = neural tissue, m = mesoderm. Scale bars = 200 ~m. A-C: Recombinant of FLDx-labeled early gastrula (stage 10) ectoderm with stage 14 anterior neural plate stained with anti-N-CAM antibody. Note in (A) that the neural plate tissue has formed a ring of neural tissue that has the appearance of a neural tube (n; also compare A and C) and an eye (e) as judged by the presence of pigmented retina. Test ectoderm (labeled tissue in B) has formed a small projection from the neural tube (large arrow in B) that stains positively for N-CAM (arrow in C). (The small arrow at the bottom of B marks an area of weak FLDx staining within the eye tissue; this is apparently due to small amounts of transfer of FLDx from the test ectoderm as described under Materials
SERVETNICK AND G•AINGER
combinants, but this healing may not occur uniformly at all edges of the recombinant and the time required for such healing may preclude the efficient transmission of planar signals. Another explanation for the small size of induced neural structures is that tissues present in vivo, but absent from recombinants (for example, mesoderm), may potentiate the inducing effect of the neural plate. This has been shown to be the case in lens induction, in which mesoderm, which itself is incapable of inducing a lens, potentiates the ability of presumptive eye tissue to do so (Henry and Grainger, 1990). In sum, homeogenetic neural induction may be affected by both the regional nature of the responding ectoderm and by the arrangement of inducing and responding tissues. The conditions under which homeogenetic neural induction occurs may be sharply restricted; this may be why only small groups of cells are induced in recombinants in our experiments as well as in those of others (Deuchar, 1971; Grunz, 1990).
The Role of Homeogenetic Induction in Embryogenesis Our experiments argue that homeogenetic neural induction can occur in the Xenopus embryo; however, we have no information as to whether it actually does occur during normal development. Nieuwkoop and colleagues (Nieuwkoop, 1952, 1985; Albers, 1987) have argued that homeogenetic neural induction plays a crucial role in normal development, delineating the spatial extent of the neural plate. On the other hand, evidence indicates that the inducing capacity of dorsal mesoderm is sufficient to account for the spatial extent of the neural plate (Holtfreter, 1933; Jones and Woodland, 1989), and therefore homeogenetic induction need not be invoked to account for the neuralization that occurs in normal development. Instead, the signaling mechanisms involved in homeogenetic induction may reflect signaling within a tissue layer, which normally occurs within morphogenetic fields. These interactions are involved in establishing specific cell fates within a previously induced field (see Jacobson and Sater, 1988); however, it could be that as a result of experimental manipulation, these signals can cause competent ectoderm to be incorporated into the morphogenetic field. Such interactions within the mesoderm lead to the dorsalization of mesoderm which is originally specified as ventral (reviewed by Smith, 1989b); these dorsalizing signals may also be ca-
Homeogenetic Neural Induction
81
pable of homeogenetically inducing competent ectoderm to form mesoderm. Analogous interactions, which normally specify cell fates within the neural plate, may lead to homeogenetic neural induction of competent ectoderm. For example, Hemmati-Brivanlou et al. (1990) have proposed that the anterior notochord induces the expression of the engrailed gene in the overlying medial neural plate and that these engrailed-expressing medial neural cells then laterally signal adjacent neural plate cells to express the gene; such a lateral signal might be able to induce competent ectoderm to neuralize. Thus, homeogenetic induction may be a manifestation of the regulative ability of vertebrate embryos. While the signals which lead to homeogenetic neural induction might be involved in either neuralization or in subsequent regulation of neural cell fate, they might also be involved in the specification of placodal tissues, as proposed by Nieuwkoop (1963). According to this model (Nieuwkoop, 1985; Albers, 1987), a single signal is responsible for the induction of both neural and placodal tissues: the signal emanates from the dorsal midline and travels slowly through the ectoderm in a lateral direction. Because the competence of ectoderm changes rapidly during this period (Servetnick and Grainger, 1991), the signal elicits different responses in the ectoderm which it encounters, causing the induction of neural tissue early (and medially) and the induction of placodal tissue later (more laterally). The Nieuwkoop model specifically predicts that if the signal were to contact young ectoderm, it should elicit a neural response, even if surrounding ectoderm were no longer competent to respond by neuralizing. This prediction is borne out by our observations. Although a number of elements of the Nieuwkoop model remain to be tested, we have provided evidence here that a planar signal spreads from the neural plate during early development. We t h a n k Urs Rutihauser and Elizabeth Jones for kindly sharing antibodies; Betty Gallagher for communicating unpublished results; Beverly Smolich and Margaret Saha for comments on the manuscript; and Li-jun Wang, Margie Fox, and Les Cook for technical assistance. This work was supported by a National Research Service Award from the National Eye Institute to M.S. (EY-06173) and grants from the National Eye Institute to R.M.G. (EY-05542 and EY-06675).
Note added in proof Itoh and Kubota (Dev. Growth Differ. 33, 209216) have recently examined homeogenetic neural induction in Xeno-
and Methods.) D-F: Recombinant of FLDx-labeled late blastula (stage 9) ectoderm with stage 14 anterior neural plate stained with antibody 2G9. The neural-plate-derived tissue has formed a small neural mass in the interior of the recombinant (n; cf., panels A and C). Test ectoderm (labeled tissue in E) has been assimilated into the neural tissue (arrow in E); within this composite neural structure, both neural-plate-derived tissue and test ectoderm (arrow in F) stain positively with the 2G9 antibody. G-I: Recombinant of FLDx-labeled late blastula (stage 9) ectoderm and stage 1312dorsolateral head mesoderm (see Fig. lb). In many such ectoderm-mesoderm recombinants, we have observed that, despite the presence of sufficient ectoderm to completely surround the mesoderm, the ectoderm does not do so, but the mesoderm (m) protrudes through the ectodermal ball. Neither the test ectoderm (labeled tissue in H) nor the mesoderm stains with the 2G9 antibody.
82
DEVELOPMENTALBIOLOGY
pus explant recombinants and reported results similar to those presented here. REFERENCES ALBERS, B. (1987). Competence as the main factor determining the size of the neural plate. Dev. Growth Differ. 29, 535-545. BALAK, K., JACOBSON,M., SUNSHINE,J., and RUTISHAUSER,U. (1987). Neural cell adhesion molecule expression in Xenopus embryos. Dev. Biol. 119, 540-550. COOKE,J., SMITH,J. C., SMITH,E. J., and YAQOOB,M. (1987). The organization of mesodermal pattern in Xenopus laevis: Experiments using a Xenopus mesoderm-inducing factor. Development 101, 893-908. DEUCHAR,E. M. (1971). Transfer of the primary induction stimulus by small numbers of amphibian ectoderm cells. Acta Embryol. Exp. 2, 93-101. DIXON, J. E., and KINTNER, C. R. (1989). Cellular contacts required for neural induction in Xenopus embryos: Evidence for two signals. Development 106, 749-757. GIMLICH, R. L., and BRAUN,J. (1985). Improved fluorescent compounds for tracing cell lineage. Dev. BioL 109, 509-514. GRUNZ, H. (1985). Information transfer during embryonic induction in amphibians. J. Embryol Exp. Morphol. 89, (Suppl.), 349-364. GRUNZ, H. (1990). Homoiogenetic neural inducing activity of the presumptive neural plate of Xenopus laevis. Dev. Growth Differ. 32, 583-589. GRUNZ, H., and TACKE, L. (1989). Neural differentiation of Xenopus laevis ectoderm takes place after disaggregation and delayed reaggregation without inducer. Cell Differ Dev. 28, 211-218. GURDON,J. B. (1987). Embryonic induction--Molecular prospects. Development 99, 285-306. HAUSEN, P., WANG, Y. H., DREYER, C., and STICK, R. (1985). Distribution of nuclear proteins during maturation of the Xenopus oocyte. J. Embryol. Exp. Morphol. 89, (suppl.), 17-34. HEMMATI-BRIVANLOU, A., STEWART, R. M., and HARLAND, R. M. (1990). Region-specific neural induction of an engrailed protein by anterior notochord in Xenopus. Science 250, 800-802. HENRY, J. J., and GRAINGER,R. M. (1987). Inductive interactions in the spatial and temporal restriction of lens-forming potential in embryonic ectoderm of Xenopus laevis. Dev. Biol. 124, 200-214. HENRY, J. J., and GRAINGER, R. M. (1990). Early tissue interactions leading to embryonic lens formation in Xenopus laevis. Dev. Biol. 141, 149-163. HOLTFRETER, J. (1933). Nicht typische Gestaltungsbewegungen, sondern Induktionsvorg~inge bedingen medullare Entwicklung von Gastrulaektoderm. Wilhelm Roux Arch. EntwMech. Org. 127, 591618. HOLTFRETER,J. (1938). Ver~inderungen der Reaktionsweise im alternden isolierten Gastrulaektoderm. Wilhelm Roux Arch. EntwMech. Org. 138, 163-196. HOLTFRETER, J. (1947). Neural induction in explants which have passed through a sublethal cytolysis. J. Exp. Zool. 106, 197-222. JACOBSON,A. G., and SAWER,A. K. (1988). Features of embryonic induction. Development 104, 341-359. JONES, E. A., and WOODLAND,H. R. (1989). Spatial aspects of neural induction in Xenopus laevis. Development 167, 785-791. KURIHARA,K., and SASAKI,N. (1981). Transmission of homoiogenetic induction in presumptive ectoderm of newt embryo. Dev. Growth Differ. 23, 361-369.
VOLUME147, 1991
MACHEMER, H. (1932). Experimentelle Untersuchung tiber die Induktionsleistungen der oberen Urmundlippe in ~tlteren Urodelenkeimen. Wilhelm Roux Arch. EntwMech. Org. 126, 391-456. MANGOLD,O. (1929). Experimente zur Analyse der Determination und Induktion der Medullarplatte. Wilhelm Roux Arch. EntwMech. Org. 117, 586-696. MANGOLD, O. (1933). Uber die Induktionsf~higkeit der verschiedenen Bezirke der Neurula yon Urodelen. Naturwissenschaften 43, 761766. MANGOLD,O., and SPEMANN,H. (1927). Uber Induktion yon Medullarplatte durch Medullarplatte im jtingeren Keim, ein Beispiel homSogenetischer oder assimilatorischer Induktion. Wilhelm Roux Arch. EntwMech. Org. l l l , 341-422. NEW, H. V., and SMITH, J. C. (1990). Inductive interactions in early amphibian development. Curr. Opi~ Cell Biol. 2, 969-974. NIEUWKOOP, P. D., et al. (1952). Activation and organization of the central nervous system in amphibians. J. Exp. ZooL 120, 1-108. NIEUWKOOP, P. D. (1963). Pattern formation in artificially activated ectoderm (Rana pipiens and Ambystoma punctatum). Dev. Biol. 7, 255-279. NIEUWKOOP, P. D. (1985). Inductive interactions in early amphibian development and their general nature. J. Embryol. Exp. MorphoL 89, (Suppl.), 333-347. NIEUWKOOP, P. D., and ALBERS, B. (1990). The role of competence in the cranio-caudal segregation of the central nervous system. Dev. Growth Differ. 32, 23-31. NIEUWKOOP, P. D., and FABER, J. (1967). "Normal Table of Xenopus laevis (Daudin)." North-Holland, Amsterdam. NISHIJIMA, K., NODA, S., KURIHARA,K., and SASAKI,N. (1978). Differentiation of partially mesodermalized ectoderm: Homoiogenetic and heterogenetic induction by primarily induced part of the ectoderm. Dev. Growth Differ. 20, 275-281. ROMEIS, B. (1968). "Mikroskopische Technik." p. 81. Oldenburg, Munich/Vienna. RuIz I ALTABA,A. (1990). Neural expression of the Xenopus homeobox gene Xhox3: Evidence for a patterning neural signal that spreads through the ectoderm. Development 108, 595-604. SAVAGE,R., and PHILLIPS, C. R. (1989). Signals from the dorsal blastopore lip region during gastrulation bias the ectoderm toward a nonepidermal pathway of differentiation in Xenopus laevis. Dev. Biol. 133, 157-168. SAXEN, L. (1989). Neural induction. Int. J. Dev. Biol. 33, 21-48. SERVETNICK,M., and GRAINGER,R. M. (1991). Changes in neural and lens competence in Xenopus ectoderm: Evidence for an autonomous developmental timer. Development 112, 177-188. SHARPE, C. R., and GURDON,J. B. (1990). The induction of anterior and posterior neural genes in Xenopus laevis. Development 109, 765-774. SLACK, J. M. W. (1984). Regional biosynthetic markers in the early amphibian embryo. J. Embryol. Exp. Morphol. 80, 289-319. SMITH, J. C. (1989a). Induction and early amphibian development. Cu~. Opin. Cell Biol. 1, 1061-1070. SMITH, J. C. (1989b). Mesoderm induction and mesoderm-inducing factors in early amphibian development. Development 105, 665-677. SPEMANN, H. (1938). "Embryonic Development and Induction." Hafner, New York. SUZUKI,A. S., MIFUNE, Y., and KAN~DA,T. (1984). Germ layer interactions in pattern formation of amphibian mesoderm during primary embryonic induction. Dev. Growth Differ. 26, 81-94. WINKLBAUER, R. (1988). Cell proliferation in ectodermal explants from Xenopus embryos. Roux's Arch. Devl. Biol. 197, 141-147.