The Regional Properties of Amphibian Organizer Tissue After Disaggregation of Its Cells in Alkali

The idea that regional differentiation in the embryo is governed by ‘fields’ or ‘gradients’ in the metabolism of the organizer has for some time been popular among experimental embryologists. It was a development of Child’s famous ‘theory of axial gradients’ (Child, 1916a, b; also review, 1946), according to which anteroposterior differentiation in animals depended upon differences in the metabolic rate of the tissues. Although there are many objections to this theory, it has not so far been disproved. Substantial evidence has accumulated that certain biochemical gradients do exist in vertebrate embryos, and, although the connexion between these and morphological differentiation is unknown, it has remained convenient to use the terms ‘field’ and ‘gradient’ when describing the properties of the organizer.

Waddington & Schmidt (1933) introduced the term ‘individuation field’ in connexion with the pattern of antero-posterior determination in the chick, and the influence of the host on grafted tissues. They emphasized the action of the organizer as a unifying system. Any competent tissue added to the system tends to be incorporated into it so that a single normal embryo results ; the region over which such a unifying activity occurs is the individuation field. Further, when an organizer is grafted into the neighbourhood of competent tissue, it may not only induce in that tissue structures which become assimilated into a unified system together with the derivatives of the organizer, but it may induce organs which remain independent of the graft. Such inductions, in so far as they are morphologically organized, are themselves ‘individuated’. In many cases, particular parts of the organizer are known to induce specific types of organs, and in these the organizer must have transmitted some influence which has controlled the individuation of the induction. Waddington & Schmidt argued that this transmissible influence was connected with the individuation field by which the organizer’s own morphological unity is ensured. Since any specific organ which is induced will be characteristic of some particular region of the embryo, the capacity of an organizer to transmit specific individuating influences may be referred to as its ‘regionality’. It may be noted, however, that it is quite possible that within an induction individuating forces might arise which were not specifically determined by any influence transmitted by the inductor; this would be ‘self-individuation’ (cf. Waddington, 1950; Lehmann, 1945, ‘Selbstorganisation’).

Using a more analytical approach, Dalcq (1940) attempted to explain differentiation in the amphibian embryo in terms of ‘morphogenetic potentials’. The future dorsal lip region, and other tissues in relation to it, were supposed to be determined by potentials computed from their positions along two axial metabolic gradients: one dorsoventral, and the other antero-posterior. The existence of these two gradients is fairly well established, but it is questionable how far they control regional differentiation.

Since no qualitative explanation has ever been given which could explain all the known facts about regional differentiation, many leading embryologists (e.g. Dalcq, 1940; Lehmann, 1945; Nieuwkoop, 1946; Waddington, 1937; Yamada, 1950) still favour the point of view that there are quantitative biochemical gradients within the organizer, at least in the early stages of development, although these probably later give rise to qualitative differences. Waddington & Yao (1950) have paid particular attention to the space-time relationship between archenteron roof and dorsal ectoderm during gastrulation, and have given contour maps of the suggested distribution of the evocator in levels of concentration extending outwards from a maximum in the anterior mesoderm. In their work the integrative quality of organizer action is once more emphasized. It is suggested that this quality normally depends upon both the temporal and the spatial sequence of invagination.

Although little is known about the influence of time on the regional properties of the organizer, a few observations have been made on the influence of its spatial configuration. Spemann (19316) noted that the success of secondary inductions depended partly on the direction and extent to which the implanted organizer had invaginated. Altekriiger (1932) claimed that the nature of inductions was determined by the external shape of the implanted tissue.

In the light of the ideas which have just been discussed, it was remarkable when Holtfreter (1944b) showed that the organizer can apparently induce normally after its cells have been first disaggregated in alkali and then allowed to reaggregate in random order. He found that the now chaotically arranged cells were still able to induce a normal symmetrical nervous system when implanted into an early gastrula. Such a treatment must have profoundly disturbed any individuation field which depends on the arrangement of the cells with respect to one another. Since the structures induced by such organizers are reasonably well individuated and therefore regionally specific, we have to consider how far the individuation of the disaggregated cells becomes restored, and whether the initial regional specificity of the parts of the organizer is also regained or whether the individuation of the induced structures arises autonomously.

There are, however, two points to be borne in mind in considering Holtfreter’s results. One is, that in the young organizer (early gastrula dorsal lip) which he used, regional differences were not yet established. There is abundant evidence (e.g. Spemann, 1931b; Holtfreter, 1933a; Hall, 1937; Waddington & Yao, 1950) that the regional properties of the organizer are modifiable at this stage. Secondly, it could be suggested that the symmetry of the nervous tissue induced by reaggregated organizer was imposed by the host’s individuation field.

The following experiments were designed to test these two possibilities, by repeating Holtfreter’s procedure, but this time using organizer tissue from the late gastrula (Harrison’s stage 12–$1212$⁠) and explanting it into early gastrula ectoderm in order to avoid any possible influence from a host embryo. In addition, a few expiants and implants of early gastrula dorsal lip were carried out to confirm Holtfreter’s results.

Yamada (1937) and Waddington & Yao (1950) have presented evidence that regional differentiation in the archenteron roof of the late gastrula is still labile. On the other hand, Mangold (1929 a) and ter Horst (1948) were able to detect regional differences of a quantitative nature in the archenteron roof of the early neurula. In the present work it was found that the completed gastrula (H. 12–$1212$⁠), was the latest stage at which it was possible successfully to separate the archenteron roof from the dorsal ectoderm. Okada & Hama (1945) appear to be the only workers who have studied the regional properties of the organizer at this stage, using expiants. But the differences which they noticed were in the types of sense organs induced : they did not describe any tissue recognizable as brain or neural tube. Since sense organs are secondary inductions, they do not seem a sound criterion for recognition of regional differences in the primary inductor. In the following work, therefore, an extensive series of control expiants was carried out (series 4 (a) and (b)) in order to establish the existence of regionality in the normal primary organizer, before studying the effects of disaggregation of its cells.

The species used for these experiments was Triturus alpestris, obtained from sources in Belgium and Switzerland, and kept in laboratory aquaria. Their normal breeding season was prolonged by injecting females with Pregnyl (Organon Laboratories Ltd.).

The embryos were decapsulated, washed and operated on in Holtfreter’s saline (Holtfreter, 1931) containing a phosphate buffer instead of the usual sodium bicarbonate buffer. To every 50 ml. of Holtfreter saline, 1 ml. of the following mixture of phosphates was added: 1 part 2·3 % KH₂PO₄, to 3 parts 1·5 % K₂HPO₄ solution. This gave a constant pH of 7·0, as measured by an electrolytic pH meter. It proved more satisfactory than sodium bicarbonate which cannot be sterilized in solution and gives a pH nearer 8 0, which is not so suitable for expiants.

Expiants were cultured in full-strength Holtfreter saline, and embryos in the saline diluted to one-tenth, in dishes floored with a layer of 2 % agar.

In spite of precautions to maintain asepsis (see Hamburger, 1942), over 60% mortality occurred in earlier experiments. So in the latter half of the 1950 season sodium sulphadiazine (May & Baker) was used as a bactericidal agent. Detwiler, Copenhaver & Robinson (1947) recommend an 0·5% solution of this drug in Holtfreter saline, but it was found in the present work that any attempt to buffer the solution at this concentration resulted in a precipitate of sodium hydrogen phosphate and a rise in pH to 8·2. Finally it was found satisfactory to use the sodium sulphadiazine at 0·1 % in the buffered Holtfreter saline, adding in addition to the mixed phosphate buffer, 1 ml. of M/15 KH₂PO₄ to every 15 ml. of saline. This gave a pH of 6·85. Crystals of sodium hydrogen phosphate formed slowly, but this had no adverse effect on the cultured tissues. Over 90% of expiants survived in this solution, for up to 14 days.

Operations were carried out with fine, two-edged mounted steel needles, the embryo resting in a depression on the floor of a waxed solid watch-glass. Presumptive ectoderm was taken from the early gastrula at Harrison’s stage 10, except in series 6 where ventral ectoderm from a completed gastrula (Harrison’s stage 12–$1212$⁠) was used. As organizer, the complete dorsal lip from the early gastrula (series 2, 5 and 6), or the archenteron roof mesoderm from the late gastrula (series 3 and 4) was used.

To disaggregate the organizer, it was placed in Holtfreter saline containing 0·7 g./l. of potassium hydroxide. The pH of this solution had been adjusted to 10·0 by addition of 1 ml. of borax-soda buffer solution (Reilly & Rae, 1943), to every 50 ml. of saline. After 2-5 min., varying with the size of the piece of tissue, the cells had rounded off and fallen apart. Further exposure to the alkali resulted in their complete disintegration. The disaggregation process was watched under the binocular, and as soon as it was complete, the loose cells were transferred in a very fine pipette, with as little of the alkali as possible, to a solid watch glass containing normal Holtfreter saline and floored with 2 % agar. The cells were collected together into a heap with the help of needles, and left undisturbed for 2 hr. or more. In this time they reaggregated into a sheet or ball of tissue which could either be explanted into ectoderm or implanted into another gastrula. The expiants or implants were left undisturbed in the watch-glasses overnight, then transferred to culture dishes.

Expiants were cultured for 9-14 days, and implants for 6-9 days. They were observed daily, disturbing them as little as possible. They were finally fixed in Bouin’s fixative, dehydrated through 50, 70 and 95 % alcohols, cleared in methyl benzoate, and taken via benzene to paraffin wax of 52° C. melting-point. Sections were cut at 10 μ, and stained with Ehrlich’s or with Delafield’s haematoxylin.

Barth (1941) and Holtfreter (1944a, 1945, 1947) found that the early gastrula ectoderm of Amblystoma punctatum and Triturus torosus was capable of differentiating into neural tissue even in the absence of an organizer, if the pH of the culture solution was suitably adjusted. Since the saline used in the present experiments had been modified both in pH and salt content by the addition of phosphate buffers and of sodium sulphadiazine, a control series of explants of early gastrula ectoderm alone was carried out to verify that neuralation could not occur under these conditions.

Twenty-one explants of early gastrula ectoderm were cultured for 12 days in the same saline as was used throughout the experiments. After this time four explants had disintegrated, and the remaining seventeen were irregularly shaped, with folds on their surface. In histological sections, no neural tissue was seen.

It was clear that the ectoderm had no power to self-differentiate into neural tissue, and that neither the modified saline nor the sodium sulphadiazine acted as evocators. All neural structures obtained in the following experiments could therefore be attributed to the inducing action of the mesoderm.

In a series of explants of which three survived, the dorsal lip was excised from the early gastrula (Harrison’s stage 10), placed in alkaline saline until its cells had disaggregated, then returned to normal saline to reaggregate. It was wrapped in early gastrula ectoderm and cultured for 10-12 days.

Eight control explants were made in addition, using normal, intact dorsal lip.

The different types of inductions obtained in experiments and controls are summarized in Table 1. When actual numbers, not percentages, were analysed statistically, it was found that there was no significant difference between experiments and controls.

Certain failures of bilateral symmetry were noticed in the nervous tissue of both experimental and control explants. The neural tube, for instance, frequently had a very small, circular lumen, and whichever wall lay next to partly or completely differentiated somite tissue was thickened. Pl. 2, fig. A shows an interesting case where the neural tube lies on its side with respect to the epidermis, having formed a thickened ‘lateral’ wall adjacent to the somite tissue lying between it and the notochord. This is in accordance with Holtfreter’s observation (1933a) that the symmetry of the neural tube is determined by the position of somites and notochord.

The brains formed were frequently asymmetrical or radially symmetrical, and eye inductions were nearly always single. There was only one explant containing paired eyes, and these were contiguous in the mid-line owing to the absence of an infundibulum.

The absence of any significant difference between the results for experiments and controls in Table 1 supports Holtfreter’s assertion (1944b) that disaggregation of the cells of the organizer does not affect its ability to induce a regionally differentiated nervous system. Although too much stress should not be laid on statistical differences where such small numbers are concerned, it is at least clear that recognizable brain, sense organs and neural tube can be induced by the reaggregated organizer. The degree of differentiation was not as complete as Holtfreter describes in his implants, nor had the tissues been organized into anything like a complete neural axis, but such differences are usual when explants and implants are compared (cf. Gallera, 1948; Pasteels, 1949). The essential result here is the similarity between inductions by disaggregated and by normal organizer.

For these experiments, embryos of Harrison’s stages 12 and $1212$ were used (small yolk-plug or slit blastopore stages). After deflecting the dorsal ectoderm, the whole archenteron roof mesoderm was removed in one piece for explantation.

Owing to shortage of material, ventral ectoderm from the same late gastrula was used to enwrap its explanted archenteron roof : this made it possible to use only one embryo per explant instead of two.

In some 100 experiments, of which seventy-three survived to give clear results, the archenteron roof tissue was disaggregated and reaggregated as described previously, then explanted into the ventral ectoderm and cultured for 14 days.

Fifteen (i.e. 40-55 %) of the controls elongated to resemble an embryonic axis. In two of these a normal-looking head developed, including a well-differentiated eye with lens and iris. There was also a tail outgrowth with a fin. The other explants were either spherical or irregular in shape.

In the experimental series only eleven (i.e. 30·55 %) formed anything resembling an embryonic axis, and none of these was as complete as the two controls mentioned above. Some had distorted tail outgrowths, and in others gills or balancer were recognizable.

The morphological features of the nervous tissues induced were not noticeably different as between experiments and controls. In both series they tended to be incomplete and asymmetrical. The brain was usually radially symmetrical or asymmetrical with one thickened wall next to a sense organ (eye or otic capsule). Pl. 2, fig. B, shows a very small piece of brain tissue thickened on the. side, next the single eye which it has induced.. Pl. 2, fig. C, shows a piece of hindbrain also thickened on one, side, next, to the single otic; capsule Both of these cases were explants of disaggregated organizer, but there were similar cases among the controls. The sense organs were nearly always single : in only one experimental explant, and one control, were there paired eyes.

The neural tubes in these explants were also atypical. There was not a single case where an elongated, bilaterally symmetrical tube had been formed. Instead, short thin-walled tubules or solid rods of neuroid cells were the rule.

Since regional differentiation implies a distinction between head and trunk nervous system within the same unit of tissue, a comparison was made between the numbers of experiments and controls which had both brain and neural tube in the same explant (col. 9 of Table 2). These figures were found not to differ significantly.

The numbers of brain, eye and neural tube inductions obtained are given in Table 2. When analysed statistically they show no significant difference between experiments and controls. But the total number of inductions of all kinds was significantly lower in the experiments. This may have been due in some cases to excessive treatment with alkali, since in several explants where there had been no induction the mesoderm appeared dead. It is interesting that dead mesoderm never appeared to induce : on the contrary, induction was invariably associated with the survival and differentiation of the mesoderm into notochord or mesenchyme.

The above results show quite clearly that the qualitative inducing powers of the archenteron roof of the late gastrula are not impaired by disaggregation of its cells. All the recognizably different types of neural tissue arose both in the experiments and in the controls. Brain, sense organs, and neural tube could be distinguished in both series, although they tended to be incompletely differentiated and irregular in shape. The possible reasons for this irregularity, typical of all induction in explants, are discussed later.

In both experiments and controls, the frequencies of brain and eye inductions were surprisingly low. This was probably due to the advanced age of the ectoderm, since Holtfreter (1938) found that the ventral ectoderm of the late gastrula has lost some of its competence for neural differentiation, and Waddington (1936) showed that at this stage even the lateral ectoderm is not as reactive to the organizer as is the dorsal ectoderm. So in 1950, when material was more abundant, a few additional explants were made using ectoderm from an early gastrula to enclose the archenteron roof.

Ten explants were made of untreated archenteron roof into early gastrula ectoderm. Only four survived, but all of these had formed brain, and three had eyes also. In all three cases the eyes were paired, but contiguous in the mid-ventral line owing to the absence of the infundibulum.

Two explants showed complete, symmetrical embryonic axes: in fact, they differed from normal embryos only in the absence of endoderm.

The markedly higher percentage of brain and eye inductions in this series than in the 1949 series indicates that early gastrula ectoderm has a much greater competence for neural differentiation than late gastrula ventral ectoderm. The cause of this lowered competence in the late gastrula ectoderm has never been explained: but it could possibly be due to its lower potentiality for growth than young gastrula ectoderm. The piece excised from the late gastrula was what would normally form only the ventral epidermis, whereas that used from an early gastrula represented the whole of the presumptive epidermis, as well as presumptive neural plate. These areas would be expected to show more growth under the influence of an organizer than would ventral epidermis alone.

The ability of the series (b) explants to form symmetrical embryonic axes is also worthy of comment. It indicates that individuation depends in part on factors residing in the dorsal ectoderm.

In the following series, control explants were carried out to test for regional differences in the normal archenteron roof, before investigating the effect of disaggregation on the organizer at this stage.

The archenteron roof of the late gastrula (stage H. 12–$1212$⁠) was divided into equal sized anterior, middle and posterior parts. Each part was explanted into one half of the ectodermal area from the early gastrula.

After 10-12 days little differentiation was visible externally, and only very few explants showed any elongation or tail outgrowths (see Table 3, col. 8). The degree of external differentiation did not depend on the region of the archenteron roof explanted. In section, brain tissue was identified in 47·9 % of the inductions, eye tissue in 35·4% and neural tube in 31·6%, averaging the figures for the three types of explant (Table 3). When the frequencies of these types of induction were compared as between different archenteron roof thirds, no significant difference was shown. The posterior thirds appeared just as capable of inducing midbrain structures as the anterior two-thirds, and conversely, the anterior third could induce neural tube. Pl. 2, fig. D, shows a symmetrical brain, with an eye cup, induced by the posterior third of the archenteron roof.

In all these explants there was the same asymmetry and incomplete differentiation as had been seen in series 2 and 3. Pl. 2, fig. D, shows the only case where the brain was bilaterally symmetrical. Further, the total number of inductions was very low. It was thought that this might be due to some factor of size, limiting the inducing powers of small organizer pieces. So a second series of control explants was made, in each of which three anterior, three middle, or three posterior thirds were combined, so that the total mass of mesoderm was equal to a complete archenteron roof.

In this series, the whole ectodermal area from an early gastrula was used to wrap each group of three archenteron roof thirds.

The total percentage of inductions (60% ; see Table 4) was much higher than in series (a). Moreover, some regional differences were now evident between the different organizer thirds, since anterior thirds gave a significantly higher percentage of brain and eye inductions than middle or posterior thirds. This falls into line with ter Horst’s (1948) and Mangold’s (1929a) results obtained with the archenteron roof of the early neurula.

Having found conditions in which regional specificity could be demonstrated in different parts of the normal organizer, it was possible to test the effects of disaggregating its cells.

Anterior, middle and posterior thirds of the archenteron roof were disaggregated separately in alkaline saline. The cells from three anterior, three middle, or three posterior thirds were then combined and explanted into the complete ectodermal area from an early gastrula.

Fewer of these explants showed elongation or tail outgrowths after 10-12 days’ culture than had either of the control groups. On histological examination, however, it was found that the total percentage of inductions (73·2% ; see Table 5) compared favourably with group (b). The frequencies of brain, eye and neural tube inductions by corresponding thirds in groups (b) and (c) did not differ significantly (cf. Table 4 with Table 5).

Observations were also made on the histological differentiation of the mesoderm, and it was noticed as in series 3 (a), that whenever a neural tube had been induced some notochord tissue had formed. Muscle tissue also formed occasionally, but was never organized into a regular series of somites. However, the ability of trunk mesoderm to form notochord showed for certain that it had not been killed by the alkali treatment, and also indicated that it had undergone considerable reorganization after reaggregation.

An analysis of the data of Table 5 showed that there was no significant difference between anterior, middle and posterior thirds as regards the numbers of each type of induction given (cols. 4-7). So judged on a quantitative basis, it appears that the regional specificity of the different parts of the organizer had been destroyed by disaggregation. It must, however, be emphasized that there is no sharp distinction between the results of group (b) and group (c) : untreated posterior thirds induced anterior nervous tissue just as often as disaggregated posterior thirds, but because middle and anterior thirds gave more brain inductions in series (b) than after disaggregation in (c), differences between the three regions reached the level of significance in (b). The alkali treatment reduced the incidence of cephalic inductions by anterior parts of the organizer, and this was responsible for the apparent loss of regional specificity in the separate thirds in (c) when the types of inductions which they gave were compared.

Since all types of neural tissue in these explants tended to be irregular in shape, they were identified by their histological character rather than their form. For instance, a neural vesicle was classed as brain not merely because its diameter was larger than a neural tube, but because of having extremely thickened walls consisting of several layers of cells with small, closely packed nuclei orientated towards the lumen of the vesicle. This is in contrast to a normal neural tube, which has few layers of cells, the innermost layer containing large, elongated nuclei.

The nerve tubes in these explants were atypical : the lumen was circular instead of the normal elongated dumbbell shape, and the cells were orientated in radial instead of bilateral symmetry. The whole tube seldom extended for more than ten sections (100 μ). There was one case where a solid neural rod had formed in the centre of the explant, evidently through differentiation of the mesoderm, not the ectoderm.

An attempt was made to compare the degree of differentiation of the neural tissue induced by different organizer thirds. This was not easy, since some explants had been cultured for longer than others which had been fixed early owing to signs of ill-health. However, inductions by the middle third of the organizer were, on the whole, the most complete and easily recognizable: in contrast, inductions by the anterior third appeared the least successful, and were sometimes rather doubtfully neural. It is interesting that Okada & Hama (1945) found that the anterior one-third of the late gastrula archenteron roof did not induce at all when implanted into the neurula. They did not, however, try explanting it.

The results in group (b) demonstrate the existence of a regional specificity in different thirds of the archenteron roof at the late gastrula stage. This specificity showed itself by quantitative differences in the frequency of brain, eye and neural tube induced by the separate thirds. Although no qualitative regional differences could be demonstrated—the posterior third was able to induce the same range of structures as more anterior parts—the quantitative differences were large enough for one legitimately to distinguish three regions in the organizer at this stage. The conclusions of Okada & Hama (1945), referred to on p. 20, are, therefore, confirmed.

Comparing the quantitative results with those of ter Horst (1948), it appears that in the present experiments a more posterior region of the archenteron roof was able to induce an eye. In her results, no eye was induced by the two posterior fifths of the archenteron roof. However, at stage H. 12, these parts would not yet have invaginated (cf. Pasteels, 1943). Mesoderm which was in the posterior third of the archenteron roof in the late gastrula would have reached the middle fifth of it by the early neurula stage. This region, ter Horst found, induced eye in 33 % cases : exactly the same figure was obtained in group (b) here for the middle thirds of the archenteron roof.

Comparison of the results in group (6) and (c) shows that after disaggregation the power of the archenteron roof to induce brain or eye tissue is reduced in the anterior two-thirds, and that therefore, regional differences, detected by the present criteria, are no longer apparent. But it is not justifiable to conclude that regional specificity has been entirely destroyed : some more accurate means of measuring it is desirable.

It was also observed that inductions were more successful in control group (b) than in group (a), and that regional differences became apparent in (b) although they could not be detected in (a). The possible reasons for this result are discussed later.

The following experiments were carried out to compare the degree of differentiation and bilateral symmetry attained by induced neural tissues in implants with that attained in explants where, as has been seen, the formations were far from normal.

The dorsal lip was excised from an early gastrula (stage H. 10) and disaggregated by the usual alkali treatment. The reaggregated cells were implanted under the prospective ventral ectoderm of an early gastrula (Mangold’s ‘Einsteckung’ technique). It was found best to insert the tissue through an incision made in the dorsal ectoderm near the anterior end of the embryo. The implant then usually remained in its anterior position, partly through becoming fused to the ectoderm as the wound healed, and partly because the embryo had flattened dorso-ventrally after the removal of its vitelline membrane.

Holtfreter (1944 b) used the material from several blastopore lips for his implants, but in the present experiments the material from one lip was found to be sufficient, since very few cells were lost during the alkali treatment. More than one reaggregated dorsal lip formed, in fact, too large a piece of tissue to implant conveniently.

Five control implants were also made, using untreated blastopore lip tissue.

In fourteen out of fifteen experiments, and in all the controls, inductions were obtained. One control formed a complete secondary larva ventrally, but there was no comparable case among the experiments, though many of them formed fairly definite secondary axes. In six of the experimental series there was secondary hindbrain, but none had formed forebrain or midbrain. Even in the controls, the one complete larva was the only secondary induction which included anterior parts of the brain. This is in striking contrast to Holtfreter’s claims to have obtained complete heads induced by disaggregated dorsal lip.

Tables 6 and 7 give in detail the tissues present in the secondary axes of both experiments and controls. It is not possible to make statistical comparisons on such few data, but one gains the general impression that, apart from the one complete larva, the controls did not feature any more complete or successful inductions than the experiments. Hindbrain was induced in two further controls, and in six out of the fourteen experiments. Other neural tissue consisted only of neural tube, while the mesoderm itself formed notochord and somites.

The nervous tissue in all these secondary inductions was elongated antero-posteriorly and was bilaterally symmetrical. Histologically it was much more regular than any inductions obtained in explants. Although both neural tube and brain consisted usually of fewer cells than had been observed in explants—because the embryos had been cultured for a shorter time (6-9 days) and cell-multiplication had not, therefore, proceeded so far—these cells were arranged compactly, in a definite orientation with respect to the lumen.

The position of the secondary axes varied in both experiments and controls. Though the implant had always been placed ventrally, it had evidently been carried dorsalwards in some cases. This phenomenon has been noticed in implants, by Yamada (1937) and by Holtfreter (1944b), who postulated that there was some attractive force from the host’s axis which drew the graft towards it. Recently, Dalcq & Minganti (1949) have observed the migration of grafted cells towards the host’s axial mesoderm. However, it is more probable that in the present experiments the graft was shifted by the invagination movements of the host.

In the experimental series it was noticeable that dorsally placed secondary axes were better differentiated than ventral ones. For instance, the secondary axis of Text-fig. 1 shows the outline of a brain, and also a dorsal fin. The ventral axis in Text-fig. 2 is by comparison far less definite. The data of Table 7 show further that brain was more frequently induced in lateral than in ventral secondary axes.

The mesoderm of the lateral axes also had a more symmetrical organization than in ventral axes. There were paired somites in all of them, whereas if somites were present in the ventral axes they formed a single block in the mid-line. A further observation was that a secondary archenteron was associated with all but one of the lateral inductions, though none ever formed with the ventral inductions. This suggested that the differentiation of the host endoderm might influence the layout and symmetry of the grafted mesoderm.

In order to investigate further the possible action of the endoderm on the graft, some implants were made into ventral halves of early gastrulae. These are described in the next section.

The dorsal lip and complete mesodermal area were removed from an early gastrula, and disaggregated in alkaline saline. The remaining part of the embryo, to be used as host, was left in the operating dish resting in such a position that its own weight pressed the cut surfaces together. After 2-3 hr., the wound had partially healed, leaving a small area through which the reaggregated dorsal lip tissue could be inserted into the blastocoel. However, subsequent healing was seldom successful : only two operations succeeded out of eleven performed.

Seven ventral halves of early gastrulae were cultured without grafts, as controls.

The two implants which survived gave rise to elongated embryonic axes on the dorsal side of the host. Pl. 2, fig. E, shows the appearance of one of them after 7 days. On histological examination, the neural axis of this embryo was found to consist only of nerve tube with some neuroid tissue at the anterior end. In the second embryo, however, there was midbrain, hindbrain and otic tissue. The otic region is shown in Pl. 2, fig. F. Both embryos possessed symmetrically arranged paired somites on either side of a notochord. The endoderm had differentiated into a normal gut.

The control half-embryos were all spherical in shape after 7 days’ culture. The dorsal ectoderm had expanded by growth, but no neural tissue had formed.

The method of operation used in these experiments was unsatisfactory because so few embryos healed successfully. Filatov (1937) experienced the same difficulty when implanting eye cups into ventral halves of early gastrulae. However, it is desirable to carry out more experiments of this type, since the ventral half of a gastrula is an ideal host, offering normal mechanical conditions for the invagination of grafted mesoderm but having no axis of its own to exert any inducing action.

The two positive results were interesting because they showed that paired somites will form in the presence of endoderm. This is in contrast to the irregular single mass of somite tissue which was the most that was obtained in explants even of normal organizer when no endoderm was present.

The conclusion already suggested by the observations in series 5 is, therefore, confirmed : namely, that the endoderm exerts some influence on the individuation of the somite mesoderm in amphibian embryos.

The most clear-cut result of the present work has been the confirmation of Holt-freter’s observations (1944b). It has been shown that in both explants and implants the dorsal lip can both differentiate and induce, exhibiting regional properties, even after its cells have been disaggregated in alkali, and reaggregated in random order. Evidently its antero-posterior specificity can be altered according to the new arrangement of the cells. Rather more surprising, however, was the observation in series 3 that even at as late a stage as the end of gastrulation, rearrangement of the archenteron roof cells in this manner did not destroy the regional properties of the tissue as a whole. At this stage Okada & Hama (1945) have been able to demonstrate differences in inducing power in different parts of the archenteron roof explanted separately, while our own data in series 4 show differences in the percentages of brain and eye inductions by explants of mesoderm derived from three different regions of the archenteron roof. To explain the results of series 3, then, it is necessary to assume that these regional differences are reversible, and can be re-established after they have been experimentally disturbed.

However, the evaluation of qualitative differences by means of quantitative data must always be a somewhat doubtful procedure, especially with operative work on amphibian embryos, where it is seldom possible in one season to collect enough data for a convincing statistical analysis. Some of the morphological features of the present results have also suggested that the quantitative data should be accepted with reserve.

To begin with, it should again be remarked that all the nervous tissue induced in explants was atypical, so that the absence of differences between experiments and controls simply meant that inductions after disaggregation were no less typical than before disaggregation : it did not imply that they were bilaterally symmetrical in either case. It was, therefore, a negative observation: a failure to detect regional alterations, and not a proof that none had taken place. Secondly, it should be noted that some inductions which were not definitely recognizable as either brain or neural tube had to be classed as ‘neuroid’, and were not included in the evaluation of the regional properties of the organizer. This will have tended to produce uniformity in results which might otherwise have shown differences between experiments and controls.

The abnormality of inductions obtained in explants has been remarked on by Gallera (1948) and Pasteels (1949). Chuang (1939), though he claims to have obtained one or two symmetrical brain inductions by non-specific inductors, depicts cases which show quite clearly the lack of symmetry in explants as compared with implants. None of these authors has attempted to explain this phenomenon.

Reasons might, however, be suggested for some of the abnormalities which arose in the present work. In the asymmetrical neural tubes, for example, the thickening of the wall on the side adjacent to somite mesoderm, in accordance with Holt-freter’s observations (1933 a), could be explained by his view that the somites have some influence on the differentiation of the neural tube. The occasional radial symmetry of brain tissue might be due to the fact that a complete sphere of ectoderm had come under the influence of the organizer, whereas in a whole embryo only the dorsal ectoderm can be affected, since the ventral ectoderm is separated from the organizer by endoderm. In addition, it could be suggested that the development of either radial or bilateral symmetry is controlled by mechanical factors. In a whole embryo, where the presumptive neural plate is stretched as a roof over the yolky ventral tissues, the orientation of the neural folds must to some extent be governed by the direction in which overall elongation is occurring in the embryo. Clearly it would be a mechanical feat of some difficulty for transversely orientated neural folds to grow up and then close across the ever-increasing longitudinal axis of the embryo to meet in the dorsal mid-line. The normal mode of neurulation is much easier to accomplish. In an explant, on the other hand, there are no mechanical restrictions due to ventral tissues, so that elongation could occur in any direction : the fact that it may occur equally in all directions and result in radial symmetry is not surprising.

In contrast to the brain and neural tube, no asymmetries were observed in individual sense organs. This is probably because their intrinsic symmetry is normally radial, not bilateral, and so is not upset by the conditions of explantation. It should also be noted that the differentiation of sense organs depends chiefly on histological changes: foldings of tissue and other morphogenetic movements are on a very small scale.

One abnormal feature of the sense organs, however, was that they were seldom paired in explants. It was noticed particularly that the eyes were nearly always single. Adelmann (1936), in a review of the causes of cyclopia, has pointed out that the eye cups invariably fuse in the mid-ventral line if for any reason the infundibulum fails to develop. Mangold (1936) and Nieuwkoop (1946) reported instances where absence of foregut endoderm was associated with fusion of the eye cups and failure of the infundibulum to develop. The absence of paired eyes in the present explants might, therefore, be explained by the fact that they contained no endoderm.

It was noticed that the mesoderm was itself able to reorganize to a considerable extent after reaggregation. In the explants of series 3 and 4, as has already been remarked (pp. 24 and 27), some notochord tissue was invariably present when neural tube had been induced, whereas brain inductions were accompanied only by mesenchyme or by undifferentiated mesoderm. It appears, then, that even at the end of gastrulation the differentiating properties of the mesoderm cells can be modified, so that they can form an axial notochord even after having been disarranged by alkali treatment. It would seem also that notochord-formation is essential to the induction of neural tube. This gives some support to Yamada’s hypothesis (1950) that the induction of trunk nervous system requires special mechanical conditions, one of which is the elongation of the mesoderm which occurs when it forms a notochord.

Another observation by Yamada (1937) was that somites formed preferentially in explants which included some notochordal tissue, and differentiated more regularly under the influence of the notochord. In the present explants, on the other hand, muscle tissue only rarely formed (cf. Okada & Hama’s findings, 1945), and it was never organized into a series of somites, even though notochord was often present. Normal somites were, however, formed by implanted organizer tissue, and their pairing was dependent on the presence of the notochord.

The experiments of series 5 and 6 confirmed the observations of many other workers that the conditions of implantation control the development of a normal symmetrical embryonic axis. But the superior differentiation of lateral secondary axes as compared with ventral ones was contrary to Lopaschov’s findings (1941). He obtained the best inductions by ventral implants. However, since his grafts were made into late neurulae, they were unaffected by gastrulation and neurulation movements of the host tissues. In the present experiments, on the other hand, where the host was an early gastrula, a graft which came to lie laterally would be assisted in spreading forwards by the similar movements taking place in the host’s mesoderm. A ventral implant would, however, tend to be carried towards the posterior by the rotation of the invaginating endoderm (Nicholas, 1945), and the cells of the gut floor would then pile up on top of it. It is clear that under these conditions the implant could not spread forwards or induce as successfully as a lateral implant, so that the present results are readily explained. Mangold (1929b) and Spemann (1931b) also observed a correlation between the success of inductions and the degree to which the graft had been able to carry out normal morphogenetic movements unimpeded by host tissues.

There are indications that in contrast to the impeding action of ventral ectoderm, dorsal endoderm can assist the individuation of a graft. For in series 5, where the host dorsal lip had been removed, two successfully implanted organizer reaggregates differentiated much better than any of the explants, forming paired somites as well as a notochord. It had also been noticed in series 5, that all the well-differentiated secondary axes were associated with a secondary archenteron. It is unknown whether the endoderm exerts any control over the differentiation of other tissues in amphibia. Although observations have been made on embryos lacking endoderm by Mangold (1936) and Nieuwkoop (1946), and on endoderm isolates in vitro by Stableford (1947) and Nicholas (1945), none of these studies has given any evidence that the endoderm acts as an inductor in the physiological sense. It remains possible, however, that it plays a purely mechanical role, since its active forward and lateral immigration would assist the spreading of the dorsal mesoderm in contact with it.

The morphological observations in the present work, then, show a striking ability of the disaggregated mesoderm to reorganize itself, to differentiate and to induce various types of nervous tissue. They also point to the importance of the endoderm in amphibian development. This tissue, which occupies more than half the volume of the early amphibian embryo, has received scant attention in the literature. It is evidently concerned in the formation of the infundibulum, as well as possibly controlling the individuation of the mesoderm.

To return to the quantitative data obtained in this work: it was remarked in the Introduction that if Holtfreter’s observations on the inducing powers of the reaggregated organizer were confirmed, it would be difficult to maintain any theory of regional differentiation which depended upon a property of the organizer tissue as a whole. However, a closer examination of the data from the explants shows that they are explicable on a simple ‘gradient’ theory.

In series 4, the appearance of significantly different numbers of cephalic and trunk inductions in explants of different thirds of the archenteron roof has been taken as evidence for regional specificity in the inductor. On this criterion, untreated thirds of the archenteron roof explanted singly showed no regional specificity but when three similar thirds of the archenteron roof were combined in each explant, the controls showed regional specificity while the experimental (disaggregated) series did not.

Text-fig. 3 depicts the simplest imaginable gradient system within the archenteron roof. A single evocator is assumed to be responsible for all types of neural induction, and its concentration is shown as a linear gradient with the highest point anterior. The concentration T represents the threshold for induction of brain: normally reached only in the anterior parts of the organizer. Though one could assume that a number of thresholds exists, each responsible for a certain level of neural induction along the anterior-posterior axis, it is only the brain level which is morphologically distinct and, therefore, of interest to us. Now since at the late gastrula stage regional differentiation in the mesoderm is still labile, and anterior pieces may in certain circumstances induce trunk nervous system (Spemann, 19316; Hall, 1937; etc.), it is necessary to assume that the threshold T has not yet been reached at this stage. Alternatively, one would have to suggest that transplanted tissue sometimes suffers a loss of evocator so that the concentration falls below T. This would be comparable to the lowering of morphogenetic potential’ invoked by Dalcq (1940) to explain the subnormal differentiation of some organizer transplants.

To account for the absence of regional specificity in the series 4 (a) explants (single thirds), it could be suggested that the anterior third of the organizer normally depends upon the other two-thirds for supplies of essential substrates. We have just seen that the lability of anterior mesoderm requires the assumption that the evocator concentration for one reason or another lies below T immediately after it has been isolated. It is possible that the isolated anterior third is sometimes unable to synthesize sufficient evocator to induce a brain, while an isolated posterior third may, on the other hand, be capable of synthesizing more than its normal concentration of evocator because it is no longer giving up substrates to anterior mesoderm. This will result in brain inductions by some posterior thirds (cf. Table 4). These two effects together may have been responsible for reducing the observed regional differences between the two extremities of the archenteron roof below the level of statistical significance, in series 4 (a).

The appearance of regional specificity when three archenteron roof thirds are combined, as in series 4 (b), is explicable if two more anterior thirds provide either additional evocator or additional substrates to the original anterior piece, enabling the threshold concentration T to be reached. The final evocator concentration in middle and posterior thirds must also be higher when three pieces are combined than in single pieces, but concentration differences between these and the anterior explants could still remain large enough for significantly more of the latter to give brain inductions. It is interesting to note that if any three numerical values are assigned to the evocator concentrations of the three organizer regions, and these are added when three similar thirds are combined, arithmetical differences are bound to be greater between groups of three thirds than between single thirds, when different regions are compared. So theoretically, a very simple numerical explanation could be given for the appearance of regional specificity in series 4 (b) when it was not apparent in (a).

It has been noticed by other workers that the nature of neural inductions depends on the size of the inducing tissue. Lopaschov (1941) obtained variation in the type of neural tissue induced, by explanting different numbers of dorsal lips. Dalcq & Lallier (1948) report that large implants give more complete inductions than small ones. On the other hand, Marx (1925) was unable to find any differences in the inductions by large and by small implants: it is clear, then, that further work needs to be carried out before a decision can be reached as to the significance of organizer size in regional determination.

Disaggregation of the cells of the archenteron roof, followed by their reaggregation in random order, would level out the gradient of Fig. 3. In series 2, 3 and 5, where all the cells were recombined as one piece, the resultant evocator concentration could be represented as a horizontal line at the average value of the original gradient. However, since according to the data the regional properties of the tissue appear unchanged, it is necessary to postulate that the gradient is re-established after explantation. The substrates required for this process should be available, since all the material from the original archenteron roof is present. What determines the anterior end of the reaggregated tissue, however, remains unexplained.

A different situation will arise when separate thirds of the archenteron roof are disaggregated, for if the concentration gradient is levelled out to an average value in each region, there will be much sharper differences in evocator concentration between adjacent boundaries of the different regions (Text-fig. 4) than there were before disaggregation. At first sight, then, it seems that regional differences should be increased, and not decreased as in series 4 (c). However, the recognition of these differences depends upon the number of explants in which the threshold concentration (T) for brain induction is reached. It was observed previously that disaggregated anterior and middle thirds gave much fewer cephalic inductions than controls: this suggests that in many of these explants the average concentration of evocator after flattening of the gradient lay well below T. This would mean that the original concentration gradient had been steep. No such reduction of cephalic inductions occurred in disaggregated posterior thirds, however, suggesting that the gradient in this region had not been very steep. So the curve of Text-fig. 5 probably gives a truer picture of the type of gradient which may be responsible for regional differences. It is consistent with Waddington & Yao’s hypothesis (1950) that liberation of evocator begins in the presumptive anterior mesoderm of the early gastrula and continues throughout gastrulation. The concentration would in this case always be higher in the anterior mesoderm, where it would also increase proportionately more quickly: a process which would result in the logarithmic curve of Text-fig. 5.

It can be concluded, then, that all the results obtained in the present work are explicable on the assumption that regional differentiation is controlled by a single gradient in the organizer tissue. It is, however, very unlikely that such a simple system ever exists in the normal embryo. Any biological system embodies a large number of biochemical processes carried on in close relationship to one another, and a change in one process indirectly affects all the rest. It would not be possible, therefore, to build up a gradient in the concentration of any one substance without affecting the concentration of a number of others, whether these are referred to as ‘evocators’ or not. The only satisfactory explanation of regional differentiation possible at present is one which assumes the existence of a number of inter-related gradients. Spiegelman (1945) has recently given an interesting implementation of this idea, showing mathematically that differentiation could be controlled and limited by the competitive interaction of several biochemical processes. In any case, as Waddington has pointed out (1950), it is clear that one must allow the existence of a number of different specific evocators in order to explain the apparent specificity of some foreign tissues in inducing only certain types of nervous tissue (Chuang, 1939; Toivonen, 1949; Yamada, 1950).

Although the present results have given no grounds for rejecting the popular idea that gradients are responsible for regional differentiation, it has emphasized the considerable regulatory powers which must exist in the organizer cells, in order to re-establish such gradients after a complete rearrangement of its cells.

I wish to thank Prof. C. H. Waddington, F.R.S., for suggesting this problem, for supervising the work, and for his helpful criticism of the manuscript. I am grateful also for laboratory facilities enjoyed at the Institute of Animal Genetics, Edinburgh.

In addition, I should like to thank Mr D. Roberts and Mr D. Pinkney for help in preparing the illustrations.

The work was carried out during the tenure of a research studentship from the Agricultural Research Council, which I also gratefully acknowledge.