The Morphogenesis of the Notochord in Amphibia

In spite of the great progress of experimental embryology in recent years, we still know rather little about the physical processes by which the various parts of the egg become moulded into definitely shaped organs. It is obvious, however, that this phenomenon of morphogenesis is one of the most important and characteristic aspects of development, and further knowledge about it is much to be desired. The investigations to be reported here represent an attempt to obtain further understanding of such processes as they affect the Amphibian notochord. This organ was chosen because the final shape which is attained is of great simplicity, being no more than an unbranched cylindrical rod. Moreover, casual inspection shows that rather considerable changes in cell shape occur during the formation of this rod, and it appeared possible that these changes in the constituent cells would be found to be directly related to the morphogenesis of the organ as a whole.

The embryos used were those of Axolotl (Ambystoma tigrinum), Triturus alpestris (imported from Belgium and Switzerland), and locally caught Triturus palmatus.

Material for histological examination was fixed in Bouin’s fluid and cleared through xylol or methyl benzoate; small fragments of tissue can be handled quite satisfactorily in xylol provided they are left in the fluid for not more than a few minutes. Most of the material was stained in Delafield’s haematoxylin, but some preparations stained in Haidenhain’s Azan have also been examined. In a third series the dorsal roof of the embryo was cut off and spread out flat under a strip of coverslip in a Petri dish containing Smith’s Formol-Bichromate fixative. The ends of the piece of coverslip were supported by two other fragments of glass so that it slightly compressed but did not crush the tissue. After fixation overnight in this position the tissues remained flat, so that accurate horizontal sections of the whole length of the archenteron roof could be obtained. These were much easier to interpret than the usual mixture of transverse, oblique, and horizontal views which results from cutting the curved dorsal layers in the intact embryo. Most of these flattened preparations were stained with Mallory’s triple stain by the modified procedure of Cason (1950).

Fragments of notochord tissue were cultivated in a number of different salines. These were all based on ‘normal Holtfreter saline’, which was made up in the usual way, with the pH adjusted to a value of 7 by means of phosphate buffers (2 3 per cent. KH2PO4 with 1·5 per cent. K2HPO4). In one series of modified salines the pH was altered by the addition of sodium phosphate buffers as shown in Table 1; the pH values of these were measured by Dr. H. G. Callan on a Cambridge pH meter. Other salines were prepared by the subtraction or addition of calcium ions in solution c, and solutions with altered surface tension were made by the addition of sodium lauryl sulphonate to solution c; these are mentioned in more detail in the text.

We should like to express our gratitude to Dr. (now Professor) H. G. Callan for assistance with the pH measurements.

During the process of invagination the presumptive notochord cells go through the well-known ‘flask-shaped’ stage which has been described by Waddington (1940, 1942), Holtfreter (1943), and others. After they have passed through the blastopore and moved forward to form the roof of the archenteron, they return to a quasi-spherical shape and adhere to one another rather weakly, so as to constitute a loose tissue (Text-fig. 1a). The definitive notochord begins to be formed in the late gastrula by the separation from the archenteron roof of a mid-dorsal strand of tissue. The cells of this arise partly by delamination, that is by active movement out of the general layer of archenteron roof, but to some extent they are formed by mitosis. They are at first more or less spherical, and only loosely attached to one another, forming an elongated strand of tissue two or three cells wide. Soon the cells begin to cohere more strongly, so that it looks as if the strand were being compressed from all sides. The cells alter towards a polygonal shape in longitudinal section (Text-fig. 1b), while the notochordal strand as a whole assumes a roughly circular outline in transverse section.

By the early neural plate stage (Harrison’s stage 14) the notochordal strand in Triturus has already become a definite rod, distinctly separated from the rest of the archenteron roof mesoderm. In this the intercellular membranes are very difficult to see with most stains, but they can be detected in sections as clear regions from which yolk granules are absent (Text-fig. lc). A similar condition is reached at a slightly later stage (about H. 17) in Ambystoma, and in this form the membranes are clearer. In both species the cells are more or less equidimensional polyhedra. At the same time, the presence of the notochordal sheath can first be recognized, as a very thin membrane staining with Aniline Blue.

In the later stages of neurulation rather rapid alterations occur in the shape of the chordal cells. There is in the first place a considerable increase in number; in an early tail-bud stage (H. 23) there are about 50 per cent, more cells per unit length of chorda than in the mid-neurula stage (H. 17). This increase is presumably due to cell division, although mitoses are rarely seen in the sections. The cells are arranged in a highly characteristic way, which may be compared to a pile of coins, many of the cells being reduced to flat disks which occupy the whole thickness of the notochord rod, though some do not extend right across it (Text-fig. 1d).

The first stage in the transition to this condition is an increase in the vertical (dorso-ventral) dimensions of the cells, which in sagittal section take an appearance very similar to those of the overlying neural groove. This causes the whole notochord to become oval in cross-section, with the long axis vertical. At the same time the nuclei become extremely flattened anterior-posteriorly. Soon after this the notochord cells increase in width in the medio-lateral direction, so that the cross-section of the notochord becomes circular once more, with most cells extending right across it. The cell membranes become thicker during this process, and stain more deeply, particularly in dyes such as Aniline Blue, and the notochord sheath is also becoming more obvious. The cells are still nearly filled with yolk granules, although these are beginning to disappear in the centre of each cell in the neighbourhood of the nucleus.

The flattened disk-shaped cells of the early tail-bud embryo gradually expand again into a polyhedral shape during the development to the late tail-bud stage. The expansion is brought about by a vacuolization, the cytoplasmic contents of the cells eventually disappearing almost completely except in the immediate neighbourhood of the nucleus. The expansion and vacuolization begins in the centre of the rod, and as the expanding cells move relative to one another, a more or less continuous layer of unexpanded cells, still containing plentiful internal cytoplasm, is formed at the surface of the notochord; this is known as the notochordal epithelium (Text-fig. 1e). It is never completely continuous over the whole surface, and as the expansion and vacuolization continue it gradually disappears so that in the fully formed larval notochord all the cells are vacuolar and polyhedral in shape, although the more superficial ones may remain somewhat smaller than those more deeply placed (Text-figs. 1f, 1g). No mitoses can be seen during the expansion of the cells, and the enlargement of the chorda seems to involve no increase in cell number. It is accompanied by the deposition of the chordal sheath, the formation of which will be discussed in another paper (Mookerjee, 1953).

It is well known from the work of Holtfreter (1938, 1944) and others that pieces of presumptive notochord can develop into this tissue after isolation in normal Holtfreter saline. The shape of the chordal masses is, however, always extremely abnormal. When tissue is isolated from young gastrulae, somewhat elongated notochords may be produced, but in these the elongation occurs in the first day or two after isolation, and does not continue thereafter. It seems probable that it is brought about by the continuance of gastrulation movements in the isolate, rather than by the processes which normally cause the elongation of the chorda in post-neurula stages. This suggestion is confirmed by the fact that when the isolates are made from late gastrulae or neurulae the notochords are normally not elongated, but form rounded masses. If the isolation has been made before the deposition of any noteworthy amount of chordal sheath, the cells become chaotically arranged. When notochords are taken from tail-bud stages, in which the sheath is present but still fairly thin, the whole isolate will swell as the individual cells enlarge, but it does not increase in length to any great extent. It seems clear, then, that the normal elongation is dependent on the enlarging cells being confined within a chordal sheath which increases in thickness pari passu with the swelling of the cells.

Fragments of chordal tissue from embryos of various ages have been isolated into the salines mentioned in Table 1. It has been shown by Holtfreter (1946) that alkaline salines reduce the cohesion between Amphibian embryonic cells, probably by bringing about a hydration of the cell contents. Since during the development of the chorda the cells become ever more closely applied to one another, until they are packed tightly together in the ‘pile of coins’ stage, it might be expected that their resistance to the disaggregating influence of alkalinity would gradually increase. This was in fact the case. Notochords from later stages were removed from the sheath before being placed in the solutions.

Notochord cells of all stages almost instantaneously burst and disintegrate.

Fragments from stages up to the late neurula begin to fall apart within a few minutes, and are usually completely disaggregated within 10-12 minutes. Early tail-bud chorda cells behave somewhat similarly, but by the late tail-bud the chorda has become more resistant and may require about 20 minutes for complete disaggregation.

The disaggregated cells in this solution show marked hyaline protuberances and exhibit a wide variety of movements (see below). It is clear, however, that prolonged exposure to the solution is highly deleterious. The cell wall may become permeable, and the cell contents gradually escape through it, or it may break down entirely so that the cell disintegrates. Continuous culture of chorda cells in the medium is impossible.

Presumptive notochord cells from late gastrulae or neurulae of Axolotl remain firmly attached to one another for the first 24 hours in this solution, but thereafter they begin to round up into spheres, and gradually fall apart, usually becoming completely separated by 48 hours after isolation. In fragments from tailbud stages of the same species there is some stretching of the cells and a lessening of their association, but they do not as a rule separate completely.

Tissue from Triturus palmatus is less resistant. Notochord cells from the late gastrula have completely dissociated in 4-6 hours, and there is not much difference in isolates from the neurula. Isolates from the early tail-bud may take as long as 30 hours for complete disaggregation, and tissue from late tail-bud stages can survive indefinitely without separation of the cells.

Separation of the cells is rather slower. In Axolotl material from late gastrulae or neurulae the cells begin to round up after about 30 hours, and have usually fallen apart in 55-65 hours, although small groups may remain coherent even after 4 or 5 days. Early tail-bud tissue of the same species shows a slight tendency for weakening of cellular attachment, but the cells do not usually become separate; late tail-bud material is apparently quite unaffected. Triturus gastrula and neurula tissue begins to dissociate after a few hours, even at pH 7-31, and the fragments are broken down into their component cells after about 24-36 hours. In this species early tail-bud material also becomes completely disaggregated, the cells swelling and bursting the chordal sheath in about 6 hours, becoming completely dissociated after about 36 hours. Late tail-bud fragments also exhibit swelling of the cells, but these do not become completely separate.

This is considered the ‘normal’ culture medium. The cells from all stages, of both Axolotl and Triturus, show no sign of disaggregation.

This is the mildest acidic solution used. Notochord cells differentiate perfectly in it.

Chordal cells can differentiate moderately well in this solution, becoming converted into a fairly normal polygonal shape. Both this solution and the last are very effective in bringing about the re-aggregation of cells which have been dissociated in alkaline solutions.

Some isolated notochord cells succeed in differentiating into a polyhedral shape, but about 70 per cent, of them remain spherical and un vacuolated. Apparently the cytoplasm is solidified at this very acid pH and differentiation prevented.

The cells remain strongly attached to one another, and look healthy under the binocular dissecting microscope, but no differentiation occurs.

Summarizing these results, it may be seen that the chordal cells tend to become dissociated at any pH greater than 7 3. Triturus material is always more affected by a given treatment than Axolotl, and in both species younger tissue is more easily dissociated than older. It should be noted that all these experiments were carried out with the cells on a glass surface, to which they readily adhere; there is some evidence (Deuchar, unpublished) that disaggregation of tissues takes place much less readily when the tissue lies on agar.

In view of the well-known importance of calcium for the maintenance of intercellular membranes in many forms of marine eggs, modifications of solution c(pH 8 2) were prepared in which calcium was either omitted entirely, or added in twice, four times, or eight times the normal concentration. (In the last two solutions some precipitation occurred.) Presumptive notochord cells isolated in these modified solutions disaggregated just as readily as in the unmodified solution of the same pH, although in the higher concentrations of calcium there was some suggestion that hyaline bulges were less often formed by the isolated cells.

Solution c was also modified by the addition of the detergent sodium lauryl sulphonate. In this way the surface tension, which in the original solution was measured (by the height of the column in a capillary) as 70-3 dynes per cm., was lowered in one case to 66·8 dynes and in another to 56·5 dynes per cm. Neither of these solutions caused any noticeable change in the behaviour of isolated fragments of notochord.

Cells from the chordal region of late gastrulae and neurulae, when isolated from one another by alkaline solutions, exhibit the types of mobility which have been well described by Holtfreter (1946). It is easiest to study these movements if the chorda cells are dissociated in pH 9 6 and then transferred to pH 7·6, when they remain viable and mobile for long periods.

The cell movements are associated with the formation of hyaline bulges, in which the external surface (plasmalemma) is lifted off from the underlying plasmagel and separated from it by a fluid ectoplasmic layer.

The present observations on the movements of chordal cells fully confirmed Holtfreter’s account of the behaviour of early embryonic cells in general, and there is little which it is necessary to add to his description. It is worth noting that in some cases in which a cell was ruptured so that the whole endoplasm escaped, it could be seen that the isolated cell-wall continued to exhibit bending and flowing movements for many minutes before finally disintegrating; similarly, isolated masses of cytoplasm from which the cell membrane had been lost may remain coherent and show gliding and rotary movements for a similar period.

If cells, isolated by alkali, are transferred to normal Holtfreter saline, the hyaline bulges disappear and the plasmalemma (cell membrane) becomes once more closely adherent to the internal cytoplasm. If several such cells lie near together, they will re-aggregate into a coherent tissue. If, however, such a cell is quite isolated, tissue formation is impossible, but nevertheless such a cell may proceed with its histological differentiation. That is to say, it becomes swollen, vacuolated, and polyhedral. As might be expected, differentiation of this kind is observed more frequently, and progresses more normally, when the cells are taken from older neurulae than when they come from the archenteron roof of a late gastrula. But it is clear that even from an early stage, single individual cells are capable of carrying out a more or less normal histogenesis. Such isolated cells do not, however, attain as large a size as they would do within the embryo.

Examination of developing notochords has shown that the most noticeable changes in the early period (from late gastrula till early tail-bud) can be summarized as an ever-increasing closeness of contact between the cells. These are at first more or less spherical and only touch one another at a few points. By the neural plate stage they have become polyhedral, and fit together so that any one cell is in direct contact with other cells over the greater part of its surface. Further closeness of contact could only be achieved by increasing the proportion of surface to volume of the individual cells; and this is what actually occurs, the cells departing markedly from equi-dimensionality. Such a departure, with the concommitant relative increase in surface, might be achieved by each cell becoming very long and narrow, assuming the shape of an elongated spindle. If the chorda cells were to develop in that way, they could only form a chorda of normal shape if they became arranged parallel to one another in a bundle, with the long axes of the cells lying in the direction of the chordal axis. In such an arrangement the cells in the middle of the bundle would be in contact all round with other cells, but those lying at the surface of the bundle would have a considerable part of their surface exposed to the ambient intercellular medium.

There would thus be considerable differences between the superficial and more deeply lying cells, and one might on general grounds suspect that such an arrangement would be unstable. Such considerations may provide some explanation for the fact that the developing chordal cells attain an increase in the proportion of surface to volume by adopting an alternative type of shape. They become exceedingly flat disks, each of which occupies the whole, or nearly the whole, cross-sectional area of the chorda. Thus each cell has only a small part of its surface (the rim of the disk) exposed to the ambient medium, while all the rest is in contact with other cells; and all the cells are very similarly placed in this respect. The arrangement would seem to be the optimum for ensuring the greatest possible cell-to-cell contact together with similarity between all cells.

After the ‘pile of coins’ stage, differentiation proceeds by the expansion and vacuolization of the cells, which again assume a polyhedral shape. It is perhaps possible that these changes also should be regarded as fundamentally a matter of increasing the proportion of cell-surface; but now this goes in such a way that the surface is too large for the amount of cytoplasm available, so that vacuoles are formed and the cell gradually transformed into what is little more than a large bag of cell-membrane filled with a clear sap.

During this stage another constraint has entered into the system, namely that provided by the notochord sheath, which is deposited around the developing chorda. It will be shown in another paper (Mookerjee, 1953) that the chorda is compressed by this sheath, and impeded by it from expanding in diameter, so that the greater part of the cell expansion becomes exhibited as a growth in length of the organ as a whole. It is presumably the constriction by the sheath which makes it impossible for the enlarging cells to retain their arrangement in an orderly column, and forces them once more to become irregular polyhedra.

The increasing closeness of contact between the developing chorda cells is reflected in their behaviour when exposed to alkaline solutions. Whereas the archenteron roof of the late gastrula is rapidly disaggregated, it requires a longer and longer exposure to bring about complete separation of the cells from older stages up to that of the ‘pile of coins’; and thereafter the resistance to dissociation is still stronger. It is, however, not clear whether these facts are evidence that the cells are gradually becoming more and more sticky with respect to one another, or whether they depend merely on the increasing area of contact between cells whose stickiness remains unaltered.

It is noteworthy that Triturus tissue is more easily disaggregated than that of Axolotl, although, as Holtfreter (1945) has shown, the presumptive ectoderm of the latter shows considerable sensitivity to abnormal salines, being easily stimulated to develop into neural tissue.

The experiments reported here have done little to clarify the nature of the cohesive forces between the notochord cells. The concentration of calcium ions in the medium seems to be without effect on them. The surface-active agent sodium lauryl sulphonate was also ineffective. Holtfreter, who, like several other authors, has laid great stress on the importance of the cell surface in morphogenesis, has discussed such phenomena in terms of surface tension (e.g. Holtfreter, 1944). There is no doubt, however, that the surface membrane of embryonic amphibian cells possesses a certain rigidity, and the concept of surface tension cannot be applied to it in any strict sense. Stableford (1949) has also presented evidence that surface tension as such is not an important force during early morphogenesis. At present, however, one cannot specify any more precisely the nature of the forces at work.

It is important to realize that, although the morphogenesis of the notochord can be very largely accounted for in terms of an increasing closeness of surface contact between the component cells, this process is not at all essential for cellular differentiation. This is demonstrated by the fact that single isolated presumptive chorda cells can continue their normal histogenesis and finish up as large vacuolated polyhedral cells, which differ from normal chorda cells only by a certain falling short in size. Such cells were, of course, determined, at least to some extent, at the time of isolation; and the present experiments give no information on the mechanism of this initial step. But the fact that differentiation can occur in isolation shows conclusively that the continuance of histological differentiation does not depend on any sort of interaction between contiguous cells. Thus it becomes rather implausible to invoke, as a possible explanation of it, a progressive immobilization of certain molecular species on the cell surface as has been suggested by Weiss (1947,1949).