The Reorganization of Tissue Masses of Cordylophora Lacustris and the Effect of Oral Cone Grafts, with Supplementary Observations on Obelia Gelatinosa

The remarkable capacity of certain sponges and hydroids to reorganize themselves after complete dissociation into a suspension of separated cells has been taken less advantage of than might have been expected for investigating cell differentiation and the factors controlling organization. The comparative simplicity of structure of these animals should theoretically make them more suitable for such investigations than the more complicated forms which have so far been employed by workers in this field. The marine sponges and hydroids which will readily form reconstitution masses are not, however, suitable for extensive and prolonged experiments, since they are mostly seasonal in their occurrence and are not easily kept for long periods in the laboratory. In addition it is necessary to observe the strictest precautions against contamination of the water if the cell masses are to remain healthy, and to produce hydranths in even a small percentage of cases (Sponges: Wilson, 1907; Huxley, 1911, 1921 ; Hydroids: Wilson, 1911 ; De Morgan & Drew, 1914; Hargitt, 1915; Okada, 1927; Child, 1928). It is therefore not easy to obtain a constant supply of material from these animals, the behaviour of which under standard conditions can be predicted with any degree of certainty, which is an essential requisite for any systematic experimentation. By using the estuarine gymnoblastic hydroid Cordylophora lacustris Allman which is abundant at most times of the year, and, like most brackish water animals, is hardier than its marine relatives, we have overcome these difficulties entirely.

Goetsch has reported that regeneration of hydranths occurs from pieces of Cordylophora coenosarc expressed from the perisarc tube (Goetsch, 1929). For our purpose it was found sufficient to chop these pieces into minute fragments from aggregates of which typical reconstitution masses were obtained.¹ This chopping method has been found successful with Hydra (Papenfuss, 1934). It has great advantages in that pieces of tissue can be freed from any debris before they are cut up, and masses of any size can be made without waste of tissue.

Previous workers have confined themselves to observations of the gross and histological changes (as seen in sections) occurring during normal reconstitution. Such observations can give no certain indications either of the changes undergone by particular cells or of the factors determining these changes. In these experiments we have endeavoured to apply the two main methods of attack employed by the experimental embryologists, namely: (a) isolation of one type of tissue from another, and (b) artificial control of organization by means of grafts. The first problem of importance which can be investigated by the method of tissue isolation concerns the fate of the two cell layers during dissociation and reconstitution. Both Wilson (1911) with Eudendrium and Pennaria, Okada (1927) and Child (1928) with Corymorpha have concluded that complete dedifferentiation occurs in the mass so that the cells of the original two layers become capable of forming cells of either layer in the reorganized structure. Whether or not cells previously well differentiated in one direction can retrace their steps and proceed in another is a problem of great general interest, more certain information on which can be obtained by experimental isolation of tissues one from the other.

We have found that separation of the two layers is relatively simple in Cordy-lophora. The results of the experiments recorded below suggest that such a change of cell type does not occur during reconstitution of the mass. They are significant also in view of the generally supposed role of the interstitial cells in Coelenterates as a reserve of totipotent cells from which new tissues are formed during repair and regeneration. The evidence for this supposition has been derived from examination of sections and from experiments on the effect of × rays which destroy the interstitial cells together with the power to regenerate. (Zawarsin, 1929 ; Strelin, 1929; Stolte, 1936.)

We have corroborated our conclusions on this subject by a few experiments on another brackish water but calyptoblastic hydroid, Obelia gelatinosa.

Experimental work upon the nature of the organizing influences at work during development and regeneration has been focused mainly upon the vertebrates. But many demonstrations have been given of the fact that such influences are at work in invertebrates and are localized in certain regions which thus control the development of other parts (see Huxley & de Beer, 1934, chap. vi). Child and his coworkers have clearly shown that in animals such as hydroids and planarians, whose organs are arranged in a relatively simple manner upon a main polar axis, the more basal regions are dependent for their appearance and development upon the presence of the more apical structures. The apical end functions therefore in a sense as an “organizer” determining the development of the rest of the animal. As a result of a large number of experiments demonstrating the existence of a gradient of susceptibility to toxic agents, of acclimatization to narcotics, of rate of oxygen consumption and reduction of dyes, and of electrical potential, Child has concluded that this organizing influence is entirely non-specific and is in the nature of a gradient of metabolic rate emanating from the apical end. This conclusion has been further supported by experiments in which the development of organs has been suppressed or the polarity reversed, the results being interpreted as due to artificial suppression or reversal of this metabolic gradient (Child, 1929a). Recent work on the vertebrate primary organizer has shown that at least in the vertebrates such a relatively simple explanation cannot be sufficient, since one part of its action (namely “evocation” or the induction of histological differentiation without acquisition of polarity), is purely chemical in nature and can function independently of the metabolic processes occurring in the organizing tissue. It seems, however, that the regional character or polarity of the induced structures can be determined only in the presence of a living organizer (Waddington & Schmidt, 1933; Heatley, Waddington & Needham, 1937). This aspect of the organizer’s action might therefore theoretically be dependent upon the setting up of gradients of metabolic activity.

Though it cannot be denied that physiological gradients of some kind exist which bear a certain relation to the structural features of an organism, Child’s theory has been criticized in many quarters on several grounds. The evidence is insufficient to warrant the conclusion that they are primarily gradients of oxidative metabolism (Needham, 1931, 3, 3.8). In cases where more than one gradient is present at once, as at the front and hind ends of annelids, the form of organization at each end is quite different (Huxley & De Beer, 1934, ch. VIII, 309). Spemann has recently criticized the theory on similar grounds and points out that gradients must presumably be continuous whereas the development of structures on an axis is discontinuous. He also discusses the important series of experiments in which embryos were subjected to temperature gradients in various directions, concluding that the results show definitely that the early embryo possesses an inherent structural polarity independent of gradients of metabolic rate; though such a temperature gradient will naturally influence rates of growth (Spemann, 1936, pp. 207–23). It is at least difficult to see how such a gradient could be maintained automatically and how any such simple, non-specific factor could play more than a minor role in the determination of structural patterns which differ so widely in different organisms, and it is possible that the gradients demonstrated by Child are more the result than the cause of organization. Nevertheless it is the function of hypotheses to stimulate research, and Child’s conception has not only proved extremely valuable in this respect, but, when certain assumptions are made, can theoretically account for more of the known facts of normal and abnormal development than any other theory. The possibility of a chemical mechanism analogous to the vertebrate evocator has not been investigated in the invertebrate animals, though it has been shown that the tissues of Hydra behave as though they contain the necessary evocator principle when grafted into amphibian embryos (Waddington & Wolsky, 1936). It is, however, hoped that the work here described will form a basis for further experiments from which some conclusions on this question can be formed.

We are here concerned mainly with the action of the most apical structure (the oral cone) which we have found to determine the development of the more basal structures of the hydranth when grafted into a Cordylophora mass. Similar effects have been produced by grafts of apical structures into various positions on the side of the hydranth in Hydra (Browne, 1909; Mutz, 1930), in Corymorpha (Child, 1929b, 1932, 1935) as also in planarians (Santos, 1929, 1931; Gebhart, 1926); but in all these experiments the possibility that the graft itself was supplying the tissues of the supposedly induced structures was not entirely excluded. In most cases they were obviously capable of doing so. The oral cone of Cordylophora, however, is incapable of regenerating other structures when isolated and we can thus be sure that we are dealing with a case of pure induction. Our main line of attack upon the problem of the nature of this inducing influence has followed from the discovery that certain conditions could be produced which would inhibit self reorganization but which could be overcome by insertion of a graft.

Cordylophora lacustris Allman (Pl. I, fig. 1) was obtained at exceptionally low tides, attached to Ryton Bridge on the River Tyne. It was found with hydranths at all times of the year except in a period roughly between. February and April, when the hydranths had disappeared. But the stolons during this period would produce hydranths after about 10 days in the warmer conditions of the laboratory. In May and June 1936 and 1937 gonophores were abundant but at other times they were scarce or absent (Pl. I, fig. 2). It seems then that sexual reproduction is confined to the period of spring growth. It was found that removed hydranths could be regenerated in salinities ranging from 20 to 90% sea water. The stock colonies were kept in aerated 50% sea water and all experiments were conducted in a medium isotonic with this. It appears that under these conditions they can be kept indefinitely, the only limit to life being reduction due to starvation. One batch was still in a healthy condition after 4 months.

We have also done a few preliminary experiments with another brackish water hydroid, Obelia gelatinosa, found attached to driftwood in the Blyth estuary, but we have been unable to keep, these in a healthy condition for more than a few weeks.

In order to make reconstitution masses the coenosarc tissue was squeezed out of the perisarc tube under a binocular microscope by means of small cataract knives. A suitable quantity of this was then transferred to a watch-glass and was chopped up into small fragments of 0·1–0·2 mm. in diameter (Pl. I, fig. 3). These fragments were then heaped together with a needle so that they were all in contact, and within about half an hour they had adhered together. After a variable period of time, averaging about 12 hr., such an aggregate would have rounded up into a spherical mass (Pl. I, fig. 4).

The protruding oral cone (Pl. I, fig. 1) is easily removed by a transverse cut. It was found that the distal two-thirds of the cone when isolated would remain unchanged for several weeks (Pl. I, fig. 7) but that if the cut was made closer to the first ring of tentacles there was a tendency for a single tentacle to develop at the base of the cone (Pl. I, fig. 7x). In order to show that the inability of these two-third oral cones to regenerate in isolation was not in some way due to their small size, pieces of similar dimensions were cut from the portion of the hydranth just basal to the tentacles. In every case tentacles were developed in a few days (Pl. I, fig. 8).

A two-third oral cone could be grafted on to the mass described above by the simple process of heaping the tissue fragments on top of the cone in a watch-glass. The cone tended to adhere by one end and the result when successful was a spherical mass with an oral cone protruding from it (Pl. I, fig. 10). From the rounded free end of the graft we concluded that in almost every case the cone was attached by its cut proximal end. It was difficult, however, to be certain of this and in one experiment we had proof that the attachment was in the reverse direction (p. 313). The percentage of successes in each set of about six trials was extraordinarily variable but the total average was roughly 50 %.

In a few experiments the distal half of a young gonophore was used as the graft. This can only be attached by the proximal end as it is elsewhere surrounded by perisarc.

When the coenosarc was squeezed out of the perisarc tube it was found that the two layers tended to fall apart, the endoderm taking the form of globular or wormlike lumps and the ectoderm being distinguishable as sheet-like masses of tissue. But the layers were more easily separated if the colonies were first stained for about 24 hr. in 50 % sea water containing a trace of neutral red sufficient to give the water a barely visible colouring. They were then left for about 2 days in, plain 50 % sea water with the result that the neutral red collected in granules in the endoderm, making this layer easily distinguishable from the unstained ectoderm. By careful manipulation with needles the two were separated and masses were made from each. As will be explained later, it was not easy to be certain that the ectoderm was entirely free from adhering endoderm cells. But the method served its purpose. This technique was also successful with Obelia gelatinosa.

In order to avoid any significant alteration in concentration of the medium due to evaporation it was necessary to keep the masses in a moderate volume of solution. At the same time it was important that they should remain in the centre of the vessel so that they could be viewed at intervals under the binocular microscope. For this purpose a deep watch-glass was supported under the solution in a small crystallizing dish (5 cm. diam. × 3 cm. deep) by means of a glass ring, the whole being covered with a Petri dish lidpl (Text-fig. 1). About 30 c.c. of solution were used in each case.

Except in the low temperature experiments the vessels were kept in an air thermostat working between 16 and 17° C.

It was obviously impossible to examine the masses at any but irregular intervals during the course of the 3 or 4 days taken for reorganization. The figures given for average times taken to reach a certain stage are probably correct to the nearest 6 hr. We have therefore not attempted to apply any statistical analysis to these figures and the average times taken to reach a given stage under two different conditions must be widely different if any significance is to be given to the comparison.

In all cases Zehker-with-formic acid was used as fixative. This penetrates rapidly and induces very little contraction of hydranth tissues. The celloidin-paraffin embedding method was employed which in previous experience with Hydra was found to prevent much of the cell breakage resulting from pure paraffin impregnation. The stains used for general examination were Delafield’s or iron haematoxylin with eosin or orange G as counterstain. For detection of interstitial cells we have used Giemsa’s eosin and azur stain which has been found successful with Hydra (Kanajew, 1930). Examination of the cell inclusions in the coenosarc tissue has been done by means of Sudan IV for fat, Best’s carmine for glycogen and Feulgen’s method for chromatin (Gatenby, 1937).

A Leica camera and microphotographie attachment was used. Whole specimens were photographed in the experimental dishes under a 2 in. objective lens. All the figures in Pl. I are therefore to the same scale.

Sections of the coenosarc in its normal position in the perisarc tube showed that the two layers are approximately of the same thickness and separated by a well-defined mesogloea (Pl. II, fig. 31). We have not as yet made à thorough examination of the structure of the cells which compose these two layers, but the more obvious histological features are worth recording. It is evident that the tissues are less differentiated than those of the hydranths. When the coenosarc was squeezed out under a cover-slip the cells came apart easily and rounded up. Hydranth tissue, on the other hand, was with difficulty broken up and muscle fibres could be seen in squeezed preparations, which were not visible in the coenosarc. The large cells which comprise the bulk of both layers have large nuclei, are vacuolated, and contain included granules which are particularly abundant towards the free ends of the endoderm cells. Fat globules were obvious as a result of staining with Sudan IV, a positive result was given by some of the granules with Best’s carmine stain for glycogen (comparison being made with control sections digested with saliva), and a large proportion of the granules gave a positive reaction to Feulgen’s stain for chromatin. From the irregular shape and arrangement of the last it was obvious that they were not normal nuclei but were perhaps derived from nuclear degeneration. The coenosarc tissue therefore contains a reserve of food materials in the form of included granules which are relatively scarce in hydranths which have not been fed for several weeks. By this means the colony can presumably tide over the winter periods when the hydranths degenerate and feeding stops, the reserves being drawn upon for maintenance and for subsequent regeneration. Immature nematocysts staining readily with orange G or eosin are distinguishable in the ectoderm.

It was important to discover whether cells are present which correspond in structure with those described by others as “interstitial” cells. These have been described in Hydra as small, often spindle-shaped, non-vacuolated cells with small nuclei containing well-defined nucleoli (Kanajew, 1930; Stolte, 1936). After staining by Giemsa’s method such cells were found in the ectoderm but not in the endoderm of the coenosarc (Pl. II, fig. 32).

The usual behaviour of a mass prepared in 50 % sea water and cultivated in the manner described above is illustrated by the typical case shown in Pl. I, figs. 3–6, which are photographs of different stages passed through by the same mass. The apparent progressive increase of volume was due to the development of an internal cavity after the mass had rounded up (see next section). After about 2 days, one or more hollow outgrowths usually appeared. On pricking a mass in this condition with a needle it temporarily collapsed, but later regained its former rigidity. A considerable internal hydrostatic pressure had been developed which perhaps plays some part in the production of the outgrowths. These might appear at any point on the mass except where it was in contact with the substratum. In all cases their appearance was at first similar. Some might continue to grow in length without further differentiation and in some cases, but not in all, they became attached to the substratum. One and rarely two or three began to differentiate into hydranths, the rudiments of the tentacles and oral cone apparently developing simultaneously (Pl. I, fig. 5). In one case the mass, having already produced two hydranths, which were later withdrawn, extended out in contact with the glass as a tubular stolon from which three more hydranths grew vertically upwards. Within the first 48 hr. it was usually possible to observe the presence of a covering of perisarc except over the regions which were actively growing out. This was not, however, easy to detect unless, as often happened, the mass had shrunk away from the surrounding perisarc.

The following is a summary of these experiments :

Some masses therefore produced no growths and the time taken to reach a given degree of development was very variable. Both these results may perhaps be explained in part as due to varying proportions of the two cell layers incorporated in the mass (see p. 312). Though there was considerable variation in the size of the original mass, there was no apparent relation between this and the result obtained. No experiments were done with masses either large enough or small enough to give results which consistently differed from the above.¹ An observation of significance in relation to other work is that unorganized outgrowths can develop which are not in contact with the substratum and in several cases it was observed that hydranths developed from processes which were lying along though not attached to the glass.

The internal changes which occur in a mass during the first 50 hr. are illustrated by the sections in Pl. II, figs. 33–35. Within the first 20 hr. (Pl. II, fig. 33) an outer wall of cells has become partly separated by irregular cavities from an inner mass. A higher power view of the inner mass shows clearly (Pl. II, fig. 36) that the cells composing it are similar to those which have already formed the endoderm layer, are distinct from one another and do not form a syncytium as described by Wilson in reconstitution masses of Eudendrium and Pennaria (Wilson, 1911). The surrounding wall of cells is already at this stage composed of two layers, an outer thin and an inner thick layer—the future ectoderm and endoderm. The cells of the latter are already of the columnar form typical of coenosarc endoderm. Pl. II, figs. 34, 35, show the subsequent enlargement of the cavity and decrease of the inner mass which is in contact at one point (not shown) with the endoderm and is presumably gradually incorporated into it.

The thickness of the ectoderm in the early stages varied greatly between different masses and the experiments described in the next section suggest that this is due to the relative amounts of the two layers originally incorporated into the mass. Pl. III, fig. 37, shows one in which the ectoderm is relatively thick at an early stage. During the swelling and subsequently, the ectoderm does not become appreciably thinner, and in masses in which it is very thin at the start it must increase in thickness. After the ectoderm is first separated its cells must therefore increase in number, but it is not obvious whether this is accomplished by cell division within the ectoderm with absorption of nutriment from the inner layer for synthesis of new cytoplasm, or whether cells migrate outwards into the ectoderm. We were unable to observe any signs of mitotic division in either layer at any stage of reconstitution. The final condition of the two layers in a mass which was already producing a hydranth is shown in Pl. III, fig. 39.

At a very early stage interstitial cells were seen to be abundant in the newly formed ectoderm (Pl. Ill, fig. 38) but we were unable to detect any in the endoderm or inner mass. This suggests what is further corroborated in the next section that the new ectoderm is derived from the cells of the original ectoderm.

Masses were made from pure endoderm and ectoderm after staining with neutral red as described in the section on methods.

Three masses made from pure endoderm rounded up within a few hours. The cells however did not remain aggregated for more than about 45 hr. and then began to fall apart. Sections showed the masses to be solid with no signs of an outer layer or internal cavity (Pl. III, fig. 40). Control masses made from the same stained material but with both layers included behaved in the normal manner and formed two layers and an internal cavity within 46 hr. (Pl. III, fig. 37).

The behaviour of isolated ectoderm tissue was distinctive from the beginning of cultivation. It rounded up very quickly, usually into lumpy masses (Pl. I, fig. 9). These were translucent and therefore probably hollow as soon as formed. Four such masses were made at different times and from different material. Three remained unchanged though apparently quite healthy and were fixed after varying periods from 96 to 190 hr. Sections of two of these are seen in Pl. III, figs. 41, 42. Fig. 41 shows a hollow structure surrounded by a thick outer layer of cells. Just under this is a very thin layer containing nuclei and granules. Whatever the nature of this inner layer it was certainly not composed of typical columnar endoderm cells such as are seen in Pl. II, fig. 35 and Pl. III, fig. 37. Fig. 42 again shows a hollow vesicle but this time surrounded by a rather irregular layer which in places appears to be two cells deep. But the dividing line between the two portions when present is less marked than in a normal mixed mass, and the inner cells are totally unlike typical endoderm. The fourth mass produced a small outgrowth after about 125 hr. which later developed into a minute hydranth. Here two well-defined layers had been formed, the inner one being composed of typical columnar endoderm cells (Pl. III, fig. 43). We have concluded that in this last experiment a small amount of endoderm must have been included, with the result that the production of a small hydranth was made possible though it appeared much later than from a normal mixed mass.

These experiments indicate that for reconstitution to occur both original layers must be incorporated into the mass and that the new layers are formed from the original ones. When the endoderm tissue is scarce reconstitution can occur but is much delayed. It seems now probable that masses showing a very thin ectoderm relative to endoderm (e.g. Pl. II, figs. 33–35) are due to inclusion of a disproportionately small quantity of ectoderm. We have found in fact that when the coenosarc is squeezed out of the tube some of the endoderm comes out first followed by the ectoderm, some of which remains in the tube and is difficult to dislodge. It is thus possible that the great variations in time taken for reconstitution, are partly due to varying proportions of the original two layers present in the mass.

All the sections of isolated ectoderm show abundant interstitial cells. At least under these conditions therefore the interstitial cells are unable to give rise to endoderm. It should also be noted that isolated endoderm is unable to hold together or to become hollow. In the presence of ectoderm surrounding it the mass is held together and a cavity appears. Since isolated ectoderm will form hollow vesicles it seems that the appearance and swelling of the inner cavity is somehow due to the action of the ectoderm layer.

Two-thirds oral cones were attached to masses in the manner already described (p. 307, Pl. I, fig. 10). During the second 24 hr. such a graft began to induce a tubular outgrowth on the end of which the cone remained attached (Pl. I, fig. 11). Subsequently a ring of tentacle buds began to appear just proximal to the graft, and it was usual that the tube in this region became slightly expanded (Pl. I, fig. 12). From the difference in opacity of the host and graft tissues at this stage, it appeared that the tentacle buds arose solely from the host tissue, but that this was necessarily the case had been previously shown by the demonstration that two-thirds oral cones do not possess potentialities for tentacle production (p. 307). Later, more tentacles were developed proximal to the first ring and the hydranth began to enlarge, but it was more than a week before the normal proportion was attained between the size of the hydranth and that of the oral cone. This too indicates that the grafted cone was supplying no tissue to the reorganized structures.

Summary of results:

Except for one example, in which an extra one appeared, all of the nine induced hydranths were the only hydranths which grew from the mass. One can conclude therefore that a cone graft will determine the position of the hydranth but that the dominating influence of the cone may occasionally fail to spread through the whole mass, with the result that another hydranth develops elsewhere. In the single case in which this was observed the extra hydranth grew out from a point opposite to the induced hydranth and the growth from which it developed did not appear until after tentacle buds had appeared on the induced outgrowth. This mass was not significantly larger than some others which produced no extra growths.

In order to determine whether the polarity of the graft was of any importance, several attempts were made to induce the cone to attach itself to the host by its distal end, and to ensure that this result could be recognized if it occurred. We have as yet found no means of controlling the orientation of the graft, but a method for recognizing a reversed graft was discovered as a result of a single experiment. In this case as much of the cone was used for grafting as could be cut from the hydranth without including the tentacle region. This cone was originally attached to the mass by a very small surface at one end, and within 48 hr. one tentacle bud had developed at the now distal end of the cone. In the same period the graft induced an outgrowth on which the usual ring of tentacle buds appeared but much later than the single bud on the free end of the graft (Pl. I, fig. 13). An isolated whole cone has been shown to produce a tentacle at its proximal end (Pl. I, fig. 7 x) and there can be no doubt that the cone in this experiment had been grafted in the reverse position. From this we must conclude that the orientation of the graft is immaterial in relation to its power to induce an outgrowth and hydranth. This is further corroborated by another later experiment (p. 317).

During May 1937, when young gonophores were obtainable, a few masses were made by chopping several and heaping the fragments together. Of four masses made in this way, all failed to round up completely and remained lumpy. Three showed no sign of further change after 144 hr. cultivation, but one produced several outgrowths after 70 hr. and the rudiments of a small hydranth after 120 hr.

Four of seven attempts to graft distal halves of young gonophore buds on tc coenosarc masses were successful. In none of these four cases was a hydranth induced at the position of the graft after 144 hr. cultivation. In two the grafts were gradually absorbed and hydranths appeared elsewhere; in one, absorption of the graft occurred without production of a hydranth ; and in one case the graft, which was larger relative to the mass than in the others, seemed to grow in size by drawing into itself some of the host tissue (Pl. I, fig. 14).

In view of the fact that oral cone grafts under normal conditions invariably induce hydranths, we can reasonably conclude that gonophore grafts are unable to do so. The last two examples described above suggest the possibility that they may have an inhibitory effect on hydranth formation under certain conditions. This possibility will be more fully investigated when suitable material is again available.

The purpose of these experiments was to discover whether the self organizing capacity of the mass and the inducing capacity of the oral cone were equally affected by inhibitory agents, and whether different types of inhibitors would produce different results in this respect. The first conditions to be tried were ionic deficiencies, we then used cyanide and phenyl urethane as inhibitors acting on different phases of the respiratory cycle (Meldrum, 1934) and lastly, low temperature which presumably acts as a general non-specific inhibitor. Artificial sea water deficient in a given ion was made by the method described elsewhere (Beadle, 1934).

Calcium-free 50 % sea water was originally tried as a possible means of separating the cells completely. The result, however, was unexpected. The tissue fragments adhered together and formed hollow twolayered spherical masses as rapidly as in the normal medium. Sections showed them to be of normal structure. Sometimes these masses would produce unorganized tubular outgrowths but would develop no further (Pl. I, fig. 15). In other cases they would remain spherical. The majority remained intact for long periods without disintegration, one specimen being observed for 15 days, after which time hydranths developed on return to normal 50% sea water. The inhibitory effect was not, however, always reversible after long exposure to calcium-free water and we are not as yet able to suggest any reason for individual differences in this respect.

Summary of results:

A few experiments were done to determine the effect of small concentrations of calcium. 50% sea water contains about 0·0055 M Ca. A normal hydranth was developed from a mass cultivated in sea water containing 0·0013 M Ca. A series of concentrations below this was tried, the result being that the rate of reconstitution was progressively reduced and the resulting hydranth was more and more malformed, as the concentration of calcium was lowered. The lowest concentration tried was 0·000066 M Ca. In this the first tentacle bud appeared only after about 140 hr. After 160 hr. two small tentacles had developed which showed no further progress during the following 2 days. A certain degree of differentiation is therefore possible when the calcium content of the medium is reduced to almost 1/100 of the normal, the resulting concentration of calcium being equivalent to that of a moderately soft fresh water.

The absence of calcium had no obvious effect on the ability of the oral cone to become attached to the mass. In every case in which there was no disintegration the graft induced a tubular outgrowth and finally tentacles. But the process was slow and the resulting tentacles were few in number and never attained the full length characteristic of those cultivated in normal 50% sea water (Pl. I, fig. 16).

Summary of results:

This large number of failures is probably attributable to some accident resulting in contamination, since seven of the ten occurred in a single experiment. We can thus conclude that to some extent the oral cone is able to overcome the inhibiting effect of lack of calcium.

A similar differential inhibitory effect was obtained with magnesium-free 50 % sea water. Four plain masses cultivated for 200–300 hr. produced outgrowths but no tentacles (Pl. I, fig. 17). Three engrafted oral cones induced outgrowths and tentacle buds within 200 hr. (Pl. I, fig. 18).

Aggregation of the fragments and rounding up of the mass occurred in the absence of potassium, but of five masses none showed any signs of further development even after more than 200 hr. Two had partially disintegrated. Ten oral cones grafted on to masses induced no outgrowths and in all cases eventually disappeared, presumably having been absorbed into their hosts. It appears therefore that the presence of potassium is a necessary condition for the organizing action of the oral cone, which can, however, function in the absence of calcium or magnesium.

In this the fragments would not stick together and within 12–24 hr. complete separation of the cells had occurred. But return to normal 50 % sea water did not result in subsequent aggregation (five examples in all).

On exposing plain masses and masses with engrafted oral cones (prepared 20–24 hr. previously in normal 50% sea water) to different concentrations of sodium cyanide from M/10,000 to M/100,000 it was found that a range of effects was produced from inhibition of tentacle formation in both plain and grafted masses in the highest, to partial inhibition in both in the lowest concentration. Between these two extremes evidence of a differential effect was found. In these concentrations sodium cyanide was found to produce no significant effect on the pH of the water. In the experiments described below the masses were made in normal 50 % sea water and after 24 hr. were transferred to the cyanide solution. The times stated include this initial period.

• M/10,000 NaCN. In all of the seven plain masses cultivated in this, further development was completely inhibited, five being kept for 200 hr. Sections showed no apparent abnormality and two on return to normal 50% sea water produced hydranths. In two of five masses with engrafted oral cones the graft was absorbed and the mass remained spherical, in two others the graft induced a tubular outgrowth but no tentacle buds were formed even after 300 hr. In another a minute tentacle bud appeared on the induced tubular outgrowth after 230 hr., but this developed no further during a subsequent 100 hr. (Pl. I, fig. 19).

• M/10,000 NaCN. In this concentration we found the maximum differential effect. Four plain masses were cultivated for 130-160 hr. One remained spherical, and the other three produced unorganized outgrowths but no signs of tentacles. On the other hand three oral cone grafts all induced outgrowths in the same medium and in two of these tentacle buds appeared within 130 hr. (Pl. I, fig. 20).

• M/100,000 NaCN. This was found insufficient to inhibit tentacle formation even in plain masses, though the process was much retarded. The two masses so cultivated produced outgrowths in 50 hr. and tentacle buds in 130 hr. Even in this low concentration, however, a certain differential effect could be detected in that an engrafted oral cone induced tentacle buds in less than half that time (60 hr.).

As with cyanide a range of concentrations was discovered which would exert a differential inhibition on plain masses and on masses with engrafted oral cones, the maximum difference being found with M/3000. As before they were made in normal 50 % sea water and after 24 hr. transferred to urethane in 50 % sea water. This initial period is included in the times stated.

• M/2000 phenyl urethane. Three plain masses showed no signs of further development during 180 hr. cultivation (Pl. I, fig. 21 a). Two engrafted oral cones, however, induced tubular outgrowths within 40–60 hr. (Pl. I, fig. 21 b), but no tentacle buds appeared with 180 hr. cultivation.

• M/2500 phenyl urethane. In this there was apparently no differential effect, since three plain masses produced only unorganized outgrowths and an, engrafted oral cone induced an outgrowth but no tentacle buds. All were cultivated 168 hr.

• M/3000 phenyl urethane. Here the differential effect was maximal. Eight plain masses were cultivated for about 150 hr. All of these produced outgrowths (Pl. I, fig. 22 a) and one grew small tentacle buds after 120 hr. Five engrafted oral cones, however, all induced outgrowths and tentacle buds within the same period (Pl. I, fig. 226).

• M/4000 and M/10,000 phenyl urethane. In both these concentrations plain masses produced outgrowths and tentacle buds within 90 hr. (three in each solution).

To discover whether a low temperature would act similarly as a differential inhibiting agent was found to be a more difficult task. It soon became obvious after several trials that the range of temperature which was likely to produce this effect was very small. The thermostat used has been already described (Beadle & Booth, 1937). The temperatures quoted were maintained ±0·1° C. Before placing in the thermostat the masses were made and kept for 24 hr. at room temperature. This period is included in the times given.

• 7·6° C. for 212 hr. Though, as might be expected, development was much retarded, this temperature was not low enough to give the desired result. Two plain masses produced outgrowths on which tentacle buds appeared after 185 hr., and these were also induced by engrafted oral cones in a similar period.

• 7·6° C. 0·72 hr. ; 6·0° C. 72–192 hr. This treatment gave some indication of a differential effect. Of two plain masses which both produced outgrowths one also showed tentacle buds after 192 hr. Three engrafted oral cones all induced tentacles within 144 hr.

• 5·0° C. 0–192 hr.; 5·5° C. 192–394 hr. These conditions resulted in an obvious differential inhibition of plain and engrafted masses. Of six plain masses all remained spherical except one which produced a single outgrowth but no tentacle buds (Pl. I, fig. 23). Of four engrafted oral cones two induced tentacle buds within 300 hr. (Pl. I, figs. 24, 25). In the second of these (fig. 25) the oral cone had previously been chopped into several fragments which were then reunited in a haphazard fashion and used as the graft. This result is of interest in that it corroborates the conclusion from the experiment described on p. 313 that the action of the cone is independent of its structural polarity. In another case the cone was accidentally removed after 192 hr. before any tentacles had appeared; the induced outgrowth remained but developed no further. The fourth graft induced an outgrowth after 68 hr. which never showed tentacle, buds. The graft, however, was attached to the mass by a very small neck (Pl. I, fig. 26) which might be expected to prevent the cone from exerting its full effect especially under inhibitory conditions.

The above experiments with inhibitory agents suggest that the oral cone acts as a kind of accelerator of processes leading to organization, which, however, occur under normal conditions in a plain mass. Under certain conditions (e.g. calcium and magnesium deficiency, M/30,000 NaCN, phenyl urethane, 5·5–6·5°C.) the capacity of the mass to become organized is not lacking, but the stimulus to do so is inhibited. This stimulus can be supplied by the grafted cone in an intensity greater than is normally present in the plain mass since it requires a greater concentration of inhibitor to check it.

In all cases normal 50% sea water at 16–17° C. was used.

Two of these were prepared by the usual method. Instead of rounding up into spherical masses they spread out as a sheet of tissue on the substratum. One, a very small mass, produced one outgrowth after 42 hr. which after 66 hr. had formed a hydranth. The other was much larger and had produced three hydranths after 42 hr. (Pl. I, fig. 27).

The neutral red staining method was found successful for separating the two layers.

One mass of isolated ectoderm spread out as an irregularly shaped flat sheet (Pl. I, fig. 28) but produced no growths after 55 hr. It later disintegrated. Two masses of isolated endoderm rounded up, but after 48 hr. started to disintegrate.

About equal quantities of coenosarc tissue from each species were chopped up into very small fragments and mixed thoroughly together. In the resulting mass after 18 hr. it was found that the two types of tissue had taken up distinct positions though they were fastened together (Pl. I, fig. 29). The centre of the mass was formed of Obelia tissue on two edges of which were embedded spherical masses of Cordylophora tissue. These three portions then proceeded to develop independently, each eventually producing a hydranth of normal appearance (Pl. I, fig. 30).

Several attempts were made to graft oral cones of Obelia and Cordylophora on to masses of the other species, but with no success. Obelia oral cone is unsuitable as a graft, since it is extremely active and moves too rapidly to become attached. This difficulty was fortunately not found with Cordylophora though we could not induce the cone to attach itself to Obelia tissue.

The problem of the origin of the differentiated tissues during regeneration and reconstitution is fraught with practical difficulties. Examination of sections is usually unsatisfactory, since it is impossible to be certain of the identity of cells from one stage to another, and even where it is possible to observe the behaviour of one type of tissue in isolation we cannot be sure how much light such observations will throw upon the potentialities of the tissue in its normal environment in contact with the rest of the organism. The differentiated tissues may theoretically arise (i) from cells of another type which have dedifferentiated and become pleuripotent (metaplasia); (ii) from cells of the same type which are structurally dedifferentiated, but still unipotent, or (iii) from a reserve of indifferent cells which have never previously been differentiated. There have been several reported cases of supposed metaplasia, but as Gray has said : “In nearly every case it is exceedingly difficult to make certain that the new tissue has actually arisen from cells which were an integral part of the old tissue, and has not developed from undifferentiated or quiescent cells present in the original tissue in comparatively small numbers.” (Gray, 1931, p. 297.) In fact, most investigators of invertebrate regeneration incline to the view that reserves of undifferentiated totipotent cells play a large part in the formation of the regenerated tissues, and in the hydroids this function is generally considered to be performed by the interstitial cells (Stolte, 1936). But here again the evidence is not conclusive and Gray’s criticism might reasonably be applied in the reverse direction.

There seems no reason to doubt that the course of events during the reconstitution of Cordylophora masses is the same as that already observed in Eudendrium and Pennaria (Wilson, 1911), in Antennularia (De Morgan & Drew, 1914), in Podocoryne (Hargitt, 1915) and in Corymorpha (Okada, 1927; Child, 1928). We cannot, however, agree with the conclusions of these workers as regards the origin of the two-cell layers. Wilson considered that the cells on dissociation become completely dedifferentiated and unite together to form a syncytial mass. From this “totipotent” aggregate of cytoplasm and nuclei there later separate the cells of the subsequent two layers. The cells of each original layer thus lose their identity and the nuclei and cytoplasm are subsequently capable of becoming redifferentiated into the cells of either layer. Our sections of Cordylophora masses show that no syncytium is formed and that the cells remain distinct. This apparent contradiction may perhaps be explained by a difference in embedding methods. From our experience it seems very probable that the ordinary paraffin method used by Wilson might so damage the cell walls as to give the appearance of a syncytium and that this is avoided by employing the celloidin-paraffin technique. Okada did not suggest the formation of a syncytium in Corymorpha masses but concluded that redifferentiation occurs from a mass of cells reduced to an embryonic (i.e. totipotent) condition (Okada, 1927). Child considered that some of the masses prepared by him from dissociated cells of Corymorpha, which produced hydranths were so small as to consist in all probability of cells from one layer (Child, 1928), but in the absence of any published histological details this conclusion must be considered as unjustified.

In sponge reconstitution masses there is no evidence of a complete reduction to a totipotent condition. In fact most workers agree that collar cells are formed from previous collar cells and that at least some of the other tissues are formed from previously undifferentiated mesenchyme cells (Wilson & Penney, 1930). Huxley’s experiments with Sycon are consistent with these conclusions. He found that masses made with an excess of either collar or “dermal” cells could not develop normally (Huxley, 1921).

Our experiments with Cordylophora in which masses were made from isolated ectoderm and endoderm show that a change from one type of tissue to another does not occur, at least under these conditions. This does not of course necessarily exclude the possibility that some metaplasia of this kind may occur during normal reconstitution when both original layers are present. It is at least certain that isolated ectoderm does not undergo differentiation into the structures found in the hydranth ectoderm, such as muscle fibres and nematocyst batteries which would occur were it underlain by endoderm, which might thus be considered to act as an “organizer”. It is hoped that some interesting results in this connexion may be obtained by inserting various types of tissue into ectoderm vesicles. It is of interest to note that isolated ectoderm rapidly rounds up and forms a hollow vesicle due to internal accumulation of fluid. This occurs normally in a mixed mass but not in masses consisting of pure endoderm. The formation of the internal cavity is thus determined by the ectoderm and can occur in the absence of endoderm. The ectoderm is therefore in some way able to perform osmotic work resulting in the internal accumulation of fluid.

In ectoderm masses the cells which correspond in structure and position to those described by others as “interstitial” cells are not capable of giving rise to endoderm. There is in fact no indication from our sections that these cells play any significant part in the reconstitution of normal mixed masses. There is no sign of any special activity (cell division, migration, etc.) on their part during the process. They are not found in abnormal numbers in the outgrowths which subsequently produce hydranths. On the other hand we have found a large accumulation of interstitial cells in the developing gonophores where they appear to develop into germ cells. This of course is the commonly accepted view of the origin of the germ cells in Hydra, which we would support from our observations of Hydra gonads sectioned and stained with Giemsa. Kanajew in his most recent work on Hydra regeneration (Kanajew, 1930) has concluded that the interstitial cells play no obvious part in normal regeneration except perhaps in the formation of new nematocysts.. Each new cell layer is formed from cells previously situated in that layer. This would presumably involve some dedifferentiation and redifferentiation but not metaplasia between ectodermal and endodermal cell types. Though these conclusions differ from those of most previous workers (see Stolte, 1936) the above experiments on Cordylophora certainly support them. Mattes has also suggested that the interstitial cells of Hydra are unimportant in normal regeneration but that they play a more significant part in growth after feeding (Mattes, 1925). We have not investigated the histology of growth in Cordylophora. Evidence which is often quoted in support of the totipotent interstitial cell theory is derived from experiments of Zawarsin and Strelin who found that doses of × rays sufficient to inhibit regeneration of lost parts in Hydra also destroyed the interstitial cells (Zawarsin, 1929; Strelin, 1929). But to conclude from this that the power to regenerate is therefore dependent upon the presence of interstitial cells is surely unjustified. The inability to regenerate might equally well be due to some inhibiting effect of irradiation upon other cells, the destruction of interstitial cells being an additional result. It might even be due to inhibition of self-organizing capacity which might be overcome by engrafting organized tissue not subjected to × rays. We propose to attempt such an experiment with Cordylophora.

With regard to the process of self organization in a mass under normal conditions we have found that tubular outgrowths which later produce hydranths will develop from any portion of the mass with the exception of that part actually touching the substratum. Child concluded that the main factor determining the polarity of the hydranth developing from a Corymorpha mass was a metabolic gradient set up in relation to the environment and substratum. That is to say that the free surface becomes the most active end of the gradient (perhaps as a result of being more exposed for gas exchange) and thus determines, the position of the new apical end (Child, 1928). On this assumption it might be expected that the hydranths would tend to appear from the surface of the mass most removed frojn the substratum. With Cordylophora masses this was certainly not the case. The majority of the hydranths grew out from the side and more or less parallel with the substratum. Very few grew vertically from the upper surface. In several cases a hydranth-producing growth appeared from a point extremely close to the substratum and extended out so that the hydranth was almost in contact with the glass. The two sheet-like masses prepared from Obelia tissue both produced hydranths from the edge in contact with the substratum. These results do not support the contention that the polarity of the regenerated hydranth is solely determined by a metabolic gradient set up in relation to the external environment and substratum. It is perhaps possible that a local variation in the structure of the surface, such as for instance a patch of ectoderm thinner than the rest, might serve as a focus from which such a gradient might emanate.

Goetsch concluded from his experiments on the regeneration of pieces of Cordylophora stem that an outgrowth from either end would normally develop into a hydranth unless it came into contact with the substratum, which inhibited this and induced the formation of the glandular structure by means of which it became attached (Goetsch, 1929). Our observations, however, show that even in a normal medium free outgrowths which have not come into contact with the substratum may yet fail to produce hydranths. Such outgrowths were formed from masses which at the same time produced hydranths. It may be that the presence of a hydranth or of organizing processes leading to the formation of a hydranth can sometimes inhibit the organization of other outgrowths from the same mass. Organization of outgrowths which would certainly form hydranths under normal conditions is inhibited by calcium and magnesium deficiency, by low temperature, and by certain concentrations of cyanide and phenyl urethane. The evidence therefore suggests that a stolon is an inhibited hydranth and is not a unipotent structure as has been claimed by Burt for the foot of Pelmatohydra (Burt, 1934). Kanajew’s experiments with Pelmatohydra, however, indicated that the foot can form trunk when under the organizing influence of a hypostome graft (oral cone plus ring of tentacles). But the graft was stained with methylene blue in order to demonstrate that it was supplying little or no tissue to the new trunk and it is perhaps open to doubt whether the final distribution of the dye was a sure indication of the position of tissues derived from the hypostome (Kanajew, 1930).

Our experiments have shown that an oral cone graft will induce the formation of an outgrowth at its base and that this will become organized into a hydranth in relation to the grafted cone. The graft supplies no tissue to the induced structure, its action is not dependent upon its orientation and it can function even after having been chopped up and reunited in a haphazard fashion before engrafting. In view of these facts and since a mass under normal conditions will form a hydranth at some point without the aid of a graft it seems likely that the action of the grafted cone is in the nature of a non-specific stimulus which releases organizing potentialities already present in the mass. Two possibilities at once suggest themselves: (a) the diffusion of some chemical substance which is the basis of the “evocator” action of the vertebrate organizer; (b) the establishment of some form of metabolic gradient with its peak in the graft in accordance with the theories of Child and particularly with his work on lateral stem grafts in Corymorpha (Child, 1929, 1932, 1935). We have so far unfortunately been unsuccessful in our attempts to engraft killed oral cdnes ; but the failure of such grafts to induce hydranths would not necessarily preclude the possibility that the normal inducing action is due to the production of a substance, because the graft is applied externally and most of its surface is exposed to the medium which would offer an easy path for diffusion. A test for the possibility (b) has also failed; pieces of coenosarc tissue engrafted into masses were previously stained with various respiration stimulating oxidationreduction dyes such as pyocyanine and cresyl blue. The dyes failed to remain in one place and diffused through the mass.

The discovery that a variety of different kinds of inhibiting agents can act differentially upon the self-organizing capacity of the mass and the inducing capacity of the cone graft seems likely to be of importance. But it is not at present obvious how this phenomenon is to be interpreted. It appears that in a plain mass the development of an organization centre can be inhibited by a certain concentration or intensity of an inhibitory agent which will still permit unorganized outgrowths, but when an organization centre in the form of an engrafted oral cone is introduced a greater intensity of inhibiting agent is required to prevent organization. In other words the process leading to the establishment of an organization centre is more sensitive to inhibiting agents than is the organizing action of the centre once it has been established. If we assume, what is by no means justified, that all these inhibiting agents (calcium and magnesium deficiency, cyanide, phenyl urethane, and cold) produce this effect by virtue of an inhibition of metabolic rate, and if we agree with Child that the organization of a hydranth is due solely to the setting up of a metabolic gradient by environmental conditions in the plain mass or by the grafted cone, how can we explain the fact that a certain intensity of inhibiting agent will prevent the establishment of a suitable gradient by environmental stimulation, but is insufficient to restrain an identical action on the part of the grafted cone? On the same hypothesis the inability of a young gonophore graft to induce tentacle formation should be due solely to its low metabolic activity, but in the absence of any positive evidence on this point, it is difficult to imagine that a structure containing a large number of rapidly differentiating cells should have a lower metabolic activity than a fully developed and histologically stable oral cone.

Child’s interpretation of the inhibiting effects of narcotics upon regeneration is that they inhibit the metabolism of the apical (organizing) structure more than that of the more basal regions. In the present instance the reverse would seem to be a more reasonable conclusion. But until we know more about the nature of the effect of these agents upon the metabolism of the mass and cone, further speculation is obviously worthless. It is possible that a micro-respirometer might be made to yield some information on this point.

There remains yet another possible explanation of the action of the engrafted cone. Child found that new hydranths would develop from various points on the stem of Corymorpha merely as the result of lateral cuts. The more proximal the point the more severe had the cut to be in order to produce a hydranth (Child, 1929b). Huxley & Gross suggested from their experiments on Sabella that the reorganization of isolated abdominal segments might be due to some stimulating action from the anterior cut surface causing a migration forward of partially determined cells and not to an “organizer” action on the part of a differentiated head end (Huxley & Gross, 1935). The action of the engrafted cone might thus be to keep open a wound, which might be performed by any piece of well differentiated tissue. It was with this in view that distal pieces of gonophores were used as grafts. The negative results obtained, so far as they go, seem to dispose of this possibility.

We are indebted to Dr J. S. Huxley and to Dr Frank Dickens for the interest they have taken in this work.

Living specimens photographed through No. 10 eyepiece and 2 in. objective lens.

Fig. 27. Reconstituted mass after 42 hr.

Fig. 28. Pure ectoderm mass after 55 hr.

Fig. 29. Mixed Cordylophora and Obdia mass after 18 hr.

Fig. 30. Same after 90 hr.