ABSTRACT
Grafts consisting of area opaca ectoderm, presumptive epidermis, presumptive neural tissue, or presumptive mesoderm (axial or side-plate), were transplanted to a position immediately under the primitive streak of chick blastoderms in the primitive streak stage.
The grafts, though they sometimes remained as a non-neural epithelium, were usually neurally induced, contemporaneously and co-extensively with the neural induction of the host. The graft-derived neural tissue is often much thicker than the host neural tissue, and though usually forming an autonomous structure, it is frequently arranged with a high degree of symmetry relative to the host. If, however, the graft remains for a long time uninduced, lying in the host mesenchyme, it tends to break up into mesenchyme itself.
It is probable that all parts of the epiblast, whatever their presumptive fate, are competent to form neural tissue, provided their epithelial structure is maintained ; and it is notable that this is true of presumptive mesoderm.
The dorso-ventral polarity of the grafts is maintained whatever their orientation in the host; but the irreversible determination of this polarity was probably not tested. The antero-posterior polarity of the grafts was without effect on their differentiation.
An elongation of the graft along the antero-posterior axis of the host usually occurred. It was often very marked, and generally consisted in the posterior end of the graft accompanying the host primitive node as it moved backwards. It is believed to be due to the induction of an active movement in the graft itself.
The host is considerably modified by the presence of the graft. In particular, the head-fold is usually suppressed. The formation of the foregut is frequently upset, but the closed foregut shows a considerable power of regulation.
I. INTRODUCTION
One of the most interesting advances in the experimental embryology of the Amphibia was the discovery by O. Mangold (1924) that when presumptive epidermis or neural tissue from a Triton gastrula is grafted into another gastrula in such a way as to be included in the invaginated organizer material, it is induced to form mesodermal or endodermal derivatives, which are built into the normal organs of the host. Later, Bytinski-Salz (1929), using a rather different technique, showed that besides the complementary induction2 found by Mangold, a certain amount of autonomous induction of neural tissue from the graft could occur. The experiments which are the subject of this paper were undertaken to discover whether, in spite of the difficulty of distinguishing host from graft tissues, similar results could be obtained in the chick. Blastoderms in the primitive streak stage were used, since this stage corresponds roughly to the amphibian gastrula. In many of the experiments of Mangold and of Bytinski-Salz the graft eventually lay between the axial chorda-mesoderm and the endoderm ; and this can be imitated in the chick by placing the graft into the host immediately under its primitive streak (which Waddington has shown to be the organizer), that is, between primitive streak and endoderm. Grafts from all regions of the epiblast, excluding the primitive streak, were used. The behaviour of grafts of the primitive streak is the subject of another paper (Abercrombie & Waddington, 1937).
II. METHOD
The technique used was the well-known one of cultivation in vitro, as used and described for chick embryos by Waddington (1932). The graft was inserted through a slit in the endoderm of the host. It is important not to remove the endoderm of the graft until the last possible moment, in order to minimize the rolling up of the graft. There was considerable variation in the size and shape of the graft used ; but in width (that is, across the long axis of the host) the most usual size was about half the width of the area pellucida, and in length grafts were usually rather less than this. The cultures were grown for about 24 hours after the operation, at 38.5°; in this time they developed at the most about a dozen somites. They were fixed in Bouin, stained as whole mounts in Delafield’s haematoxylin, cleared and examined in cedar-wood oil; then they were embedded in paraffin, sectioned at 10p (some at 15 µ), and stained again in Delafield’s haematoxylin. The work is based on eighty-eight specimens.
III. DIFFERENTIATION OF THE GRAFT
The characteristic result obtained by Mangold (1924), the complementary induction of the graft tissues, and any other kind of incorporation,1 was rare in these experiments. In a few the graft remained little changed; but in the majority it was greatly influenced by the host. It usually underwent induction to an autonomous sheet of neural tissue, and this was often accompanied by a marked elongation; occasionally it became broken up into a mass of mesenchyme, and in these cases some degree of incorporation may be said to have occurred. These results are described in detail and discussed below.
The behaviour of the grafts was the same whatever their place of origin in the donor. The precise presumptive tissues which the various grafts represented is at present uncertain. The maps of presumptive tissue regions put forward by Gräper (1929) and by Wetzel (1929), which were combined by Waddington (1932), have seemed to fit in best with much subsequent work on the chick, particularly that of Waddington and that of the chorio-allantoic grafting school (see Rawles, 1936). The map recently given by Pasteels (1936) is in substantial agreement with Waddington’s, and supplies an obvious correction with respect to the anterior extra-embryonic mesoderm. On the basis of Waddington’s scheme, grafts taken from close beside the primitive streak (seven specimens) would probably contain presumptive mesoderm (both somitic and lateral plate) and sometimes presumptive neural plate. Grafts taken from close in front of primitive streaks of various lengths (eight specimens) would include presumptive neural material. Grafts from the side of the area pellucida away from the primitive streak (seventeen specimens), and those from the extreme anterior or antero-lateral edge of the area pellucida (thirty-four specimens) would be presumptive extra-embryonic (sometimes embryonic) epidermis. Finally some grafts (sixteen specimens) were taken from the area opaca, from places varying from the edge of the aréa pellucida to the outside edge of the blastoderm. It is doubtful whether presumptive notochord was ever included. Even if quite different schemes of presumptive areas are later adopted, it is probable that all presumptive areas were included in the grafts, and all behaved alike.
(a) Specimens showing little effect of the host on the graft
In a few cases the graft is apparently almost unaffected by the host ; it forms an epithelium of a histology similar to that at implantation. This occurs when the graft has been pushed to one side, outside the host axis; when it lies under the posterior end of the primitive streak; and in regions of the graft which are in one way or another (e.g. by host endoderm, or by rolling up) shielded from the host’s evocating tissue. Grafts which lie outside the host axis at the head level may become vacuolated in a way which resembles epidermis (specimens DD,1 EG, EH). No marked change of shape occurs in those grafts which have not been induced to neural tissue.
(b) Effect of the host on the shape and position of the graft
The original shape of the graft was sometimes square, but usually oblong; and the oblong grafts were generally placed with their long axis at right angles to that of the host. During development the graft in the majority of specimens becomes narrower transversely and lengthened along the axis of the host. As an illustration of some of these changes of shape, the following are typical examples: AD had a graft at first times as long as it was broad; in sections it is eight times as long as broad. In BX it has changed from just longer than it was broad to six times as long. In CF and DF it has changed from one-quarter as long as it was broad to twice as long. Measurements of the grafts of a few specimens are given in the following table, which is a list of maximum dimensions of grafts measured before and after cultivation (the dimensions are in µ). No allowance has been made for shrinkage at fixation.
In these specimens the grafts at operation were rectangles. After fixation, however, their breadth is far from constant throughout their length. In DS, DV and ED it is usually not much less than the maximum (640, 300 and 420 µrespectively); but in DY and EA all except the anterior part of the graft is a solid rod or a small tube of 50-60µ diameter.
The elongation of the graft is almost entirely in a backward direction: the anterior edge of the graft is little or not at all further forward in the host, but the posterior edge is far behind its original position. The main body of the graft is usually distinguishable as a large mass of tissue approximately in its original position ; behind it a long strand of graft tissue of varying width projects backwards beneath the host axis. The elongation was observed taking place in the living specimen in EE and EL, where the grafts remained clearly visible during development. It has occurred in about three-quarters of the cases where fairly good development of the host has taken place with the graft in its axis; and in most of these specimens the elongation has been considerable.
Elongation frequently takes place in such a way that the posteriorly projecting strand of the graft lies exactly in the centre of the host axis (Pl. I, fig. 2; and Text-figs, 1, 3, 8), in spite of the usual slight asymmetry of the anterior main mass of the graft, which lies where it was placed. This produces in section a perfect symmetry of graft with host; the phenomenon may be referred to as “centring”.1 Centring, however, by no means always occurs; there are often slight deviations in an otherwise central graft at the posterior end of the elongation, generally behind the level where the host somites differentiate (e.g. AD, BH; Text-fig. 4); and in many the posteriorly projecting strand throughout lies to one side of the axis, though never lateral to the edge of the host neural plate, nor lateral to the axial mesoderm ; but where no centring has occurred, the graft had usually been placed to one side of the host axis. The occurrence of centring can probably be generalized by saying that the elongation takes place backwards from the most posterior part of the graft within the lateral limits of the host neural plate; or, if the posterior edge of the graft within these limits is at right angles to the host axis, then from the part of the graft exactly under the mid line of the host.
The elongation only occurs in grafts which have been neurally induced, but it is not simply a consequence of neural induction, since some grafts are found which are fully neural, but, lying outside the host axis (e.g. AY, which has been shifted sideways after induction by the host primitive streak), or in the head (AX), are not elongated. In no case is a backward shifting of the graft as a whole observed.
The elongation may be due either to a redistribution of the material of the graft ; or to its rapid growth directed backwards in the axis of the host, combined with necrosis and resorbtion of its lateral edges when these lie outside the axis. The rate of change of shape in the cases where it has been observed was probably too rapid, and the number of mitoses orientated in the required direction is too few, for growth to account for more than a small proportion of the elongation ; and no sign of necrotic tissue is ever observed along the lateral edge of the graft even in those fixed while the change of shape was still in progress. The elongation therefore appears to be due to a redistribution of the material of the graft.
There is considerable evidence to show that when the primitive node of a chick blastoderm appears to move backwards, this represents an actual backward movement of tissue, and not merely a wave of differentiation. This movement has been described by Grâper (1929) and Wetzel (1929), who made direct observations; it is supported by the “tails” of Waddington (1932), obtained after transection of the blastoderm; and it is suggested also in some preliminary marking experiments of my own. It appears to be just at the time when this backward movement is happening in the host, as the primitive node moves backwards, that the elongation of the graft occurs. This is shown by the specimens which were watched (EE and EL), and by the position of the posterior end of the graft in relation to the primitive node in specimens fixed at various stages. The posterior end of the graft seems to accompany the primitive node of the host as it moves backwards. Consequently in many specimens, and especially in those which have been fixed while the elongation was in progress, the posterior end of the graft lies under the primitive node (e.g. BX, EE; Pl. I, fig. 1). Sometimes in the fixed specimen the graft tails off before the posterior end of the trunk region, presumably because the graft material became exhausted before the primitive node had stopped moving (e.g. AD) ; in others, though the graft reaches to the primitive node, it never goes farther posterior. The result of the movement is that the disposition of the graft must throughout development correspond more or less closely to the disposition of the host neural region above it.
It might therefore be supposed that it is the mechanical pull of this host movement which distorts the graft. The graft is perhaps held firmly at each side by its lateral edges, which usually lie in a region where no host backward movement is occurring, and is then pulled backwards in the middle, the lateral extensions of the graft being gradually drawn in to replace the tissue which had been dragged backwards. A rather similar mechanical extension of grafts often occurs in amphibian grafts (e.g. Mangold, 1924; Spemann & Geinitz, 1927). But against an explanation of this kind it may be pointed out that, when they do produce some effect, the most common result of the other host tissue movements (the outgrowth of the headprocess mesoderm, and the sideways movement of the lateral mesoderm) is to shift the graft bodily; but no case occurred where the graft was moved backwards as a whole. Further, a series of very narrow grafts was made, each hardly wider than the host primitive streak, and almost certainly lying entirely within the lateral borders of the region where backward elongation of the graft and backward movement in the host occurs; but these grafts also showed no bodily backward movement, and elongation usually took place (DY, EA). These considerations suggest, as an alternative to the hypothesis that the change of shape is due to a mechanical drag, the hypothesis that an induction of movement takes place, that is to say that the graft reacts to some directive stimulus from the host by an active movement of its own. The host presumptive neural tissue, in its backward movement, may react to the same stimulus. Such an induction would obviously be comparable with the “morphodynamic” induction of the primitive streak by the endoderm, discovered by Waddington (1932, 1933); and it may perhaps also be compared with the induction found by Raven (1935) in grafts of presumptive epidermis of Amphibia, of the power of rolling in, like a blastophore, after implantation for a short time in the dorsal lip of other gastrulae.
It is to be noted that the graft endoderm, in those specimens in which endoderm is present, does not appear to show any backward elongation (DI, DW); nor, according to my marking experiments, does the endoderm of the whole blastoderm take part in the backward movement.
The outgrowth of the head-process mesoderm produces little effect on the shape or position of the graft unless the graft is small and has been placed just in front of the primitive streak, when it is usually pushed forward, eventually lying in front of the head (BY, CB). Where the graft is large, or at any rate wide, or lies behind the primitive node, no forward movement is generally observed. It is doubtful whether a forward elongation occurs; eight specimens, however, have a narrow anteriorly projecting strand of tissue.
The sideways movement of the host mesoderm occasionally shifts an ex-centrically placed graft, or a graft which has become excentric through movement of the host primitive streak (AY), bodily sideways, but usually it produces no effect whatever. It never causes a sideways elongation; in fact the opposite generally occurs.
(c) Induction of neural tissue
This is the characteristic tissue formed by the grafts. Nearly all parts of grafts which lie in the host embryonic axis, in a region where the host itself has formed neural tissue, have been induced to neural tissue (Pl. I, figs. 2, 3, 4). This production of neural tissue cannot be self-differentiation, firstly, because in most cases the graft was probably not presumptive neural tissue, secondly, because the graft neural structures correspond exactly with the host inducing structures, and thirdly, because control implantations in the area pellucida away from the host axis do not produce neural tissue. A thickening of the graft tissue, lacking however the full characteristics of neural tissue, probably represents a stage towards neuralization, and may be called “neuroid”; it occurs frequently in various regions of the graft, particularly posteriorly, where the host medullary plate is itself only neuroid, and in specimens where the host is poorly developed (CP). There is no doubt that the extra neural tissue found in the specimens is graft-derived. Although the host sometimes forms diverticula of its neural plate, these are always short, and clearly connected at their anterior ends with the main neural plate. The neural tissue which is derived from the graft is always much more extensive, and only rarely connected with the host neural plate (and then often at its posterior end). In practice confusion does not arise, since there are no conditions sufficiently intermediate to be doubtful.
In approximately one-half (twenty out of forty-one) of the cases where the graft has formed neural tissue, its neural tissue is (in part at any rate) distinctly thicker than that of the host (Text-figs. 1, 9). The host neural tissue, in comparison with other blastoderms grown in vitro, and with in vivo chicks, is of normal histology. The superiority of the graft neural tissue cannot be explained by the fact that the graft has been in contact with the organizing tissues of the host longer than has the host neural tissue. In the antero-posterior axis, the stage of neuralization of the graft (except in the thickness of the epithelium) corresponds with the stage of neuralization of the host at the same level. This is so whatever the comparative ages of donor and host. But the graft lies under the primitive streak as a competent epithelium for some time before the host neural material appears above as an epithelium. During this time no inducing effect is produced on the graft, as is to be expected, since the primitive streak material overlying it will become side-plate mesoderm. The evocator only appears to be liberated at the time when the axial mesoderm comes to lie over the graft, a time which almost coincides with the appearance of host presumptive neural tissue above the axial mesoderm. So the graft, in spite of its occasional greater age and therefore it might be supposed more advanced competence, is only evocated contemporaneously with the host medullary plate. The greater thickness of the graft tissue may be connected with the fact that the host neural tissue is part of the whole embryonic individuation field, and is thereby restricted to a normal differentiation (compare Waddington (1936) for the effect of the individuation field on evocation in Triton), while the graft is a comparatively isolated piece of tissue, at liberty to differentiate in a neural direction without restraint.
The graft neural tissue is rarely greater than the host’s in area. The longitudinal extent of the graft is largely controlled by the change in shape it may undergo; as has been mentioned, its degree of neural differentiation corresponds to that of the host for a given level, and its neural tissue is never greater in longitudinal extent. Grafts which stretch far to each side have usually caused a lateral extension of the host axial mesoderm, and of the host neural plate, so that though widely induced they are not more so than the host (e.g. DC, BS, DJ). Most grafts, owing to the elongation and narrowing they undergo, have a lateral extent no greater than, and usually less than, the host’s neural plate, so that from mere lack of material they cannot be greater in neural extent. A few large grafts, by rolling at the edges or by forming flattened tubes, accommodate an extent of neural tissue greater than the host’s without extending beyond the host axis (e.g. AR, AX, BO; Text-fig. 2).
With rare exceptions, either the graft neural material has formed a neural tube while that of the host has not; or else the graft has a more pronounced neural groove than the host (Pl. I, fig. 2; Text-figs. 1, 5, 7, 9). This is partly due to the frequent inhibition of the rolling up of the host neural plate common in chicks grown in vitro (and occurring also in chicks in vivo, see Grünwald (1935)), which is intensified (presumably for mechanical reasons) when a graft is present beneath the neural plate. The graft does not suffer from this inhibition. Two factors seem to be involved in the rolling of the graft. Firstly, the graft, when freed of its endoderm, immediately tends to roll up; but this is usually soon inhibited by the conditions in the interior of the host blastoderm. Secondly, neuralization probably produces a tendency to roll (as shown in Amphibia by, for instance, Boerema (1929)). These two processes cause rolling in opposite directions; but much the most frequently met with is that in the neural direction. A factor preventing rolling is the strong tendency for the graft to sink into the endoderm, which becomes attached to its edges, so that it forms structurally a part of the endoderm sheet (Pl. I, fig. 3; Text-figs, 1, 3). The association of graft neural tissue with the endoderm occurs also in grafts of primitive streak made into the area pellucida (Waddington, 1932). It is no doubt connected with the tendency for endoderm and neural plate (or epidermis) to fuse perfectly with each other when a split occurs in the embryo (as in K, CW of my specimens ; and see Waddington (1932), p. 188, and Waddington & Cohen (1936)). The fusion between graft and endoderm only occurs with dv. grafts,1 whose orientation (i.e. the position of the mitoses) is the same as that of the host endoderm; dd. grafts, whose normal curvature takes their lateral edges away from the endoderm, frequently become connected with the host epidermis, though of course they never become structurally part of it when they are under the primitive streak (CF, DI; Pl. I, fig. 4); when outside the primitive streak, and therefore retaining their epidermal character, they may, however, be incorporated into the host epidermis (CK, CO). Graft neural tubes are never embedded in the endoderm in this way; and it seems probable that the attachment of the edges of graft neural plates to the endoderm prevents, to some extent, their rolling up.
The graft-derived neural structures, when not too large, are usually symmetrical in themselves and of normal morphology in transverse section (Pl. I, fig. 3; Text-figs, 1, 3, 7, 8). This is so whether the graft is a plate, when it may show the characteristic nick in the middle corresponding to the host notochord (e.g. AD, Text-fig. 3), or is a tube, when it may have the walls adjacent to the somites thickened in the usual way (e.g. BX, Pl. I, fig. 2). There seems, however, to be no correspondence with the host in morphology down the longitudinal axis. The main mass of the graft neural tissue is often in the head region of the host, only a comparatively narrow strand projecting posteriorly beneath the trunk of the host. This might be taken to be an expression of the action of the host individuation field in producing antero-posterior regional differentiation, since the normal embryo has also more neural tissue in the head than in the trunk for a given cross-section.
However, the extra graft tissue is undoubtedly due to the fact that the graft was originally put in this region, and specimens do occur where the posterior elongation has narrowed the whole graft to a uniform width (e.g. Y).
But though the graft tends to form an autonomous organ (neural plate or tube), this organ is not entirely independent of the host. In eighteen out of thirty-one specimens where a fairly narrow neural plate or tube is formed, an exact “centring” has taken place, so that the graft forms a perfectly symmetrical whole with the host (Pl. I, fig. 2; Text-figs, 1, 3, 8). Frequently the host notochord (AD, and Text-fig. 1) or somites (BX, Pl. I, fig. 2) are more associated with the graft than with the host, so that the embryo appears at first sight in section to be upside down, with the host plate the supernumerary one.
A certain number of cases of incorporation of graft neural tissue, where it has fused up with the host’s neural tissue, are found (EE, Text-figs. 5, 6). This fusion with the host seems to occur only at the anterior or posterior extremities of the graft. Similar cases, where neural tissue from grafts of the primitive streak has fused more or less completely with the host neural plate, are also found in Waddington’s experiments (e.g. see Waddington & Schmidt, 1933, p. 528). Such incorporations are, however, rare in the present work, and frequently a graft neural tube is tightly pressed against the under-side of the host neural plate without there being the slightest tendency apparent towards fusion. This is remarkably shown in DY, where the graft is for a time completely embedded in the host neural plate, and yet maintains its own radial arrangement of cells (Text-fig. 7). Perhaps the fact that the orientation of the graft neural tissue was usually opposite (dv.) to that of the host explains the general absence of incorporation ; but there appears to be a general lack of capacity for fusion between sheets of neural tissue at this stage (see for instance Waddington & Cohen, 1936).
The dorso-ventral polarity of the neural tissue formed by the graft can be recognized by the position of the mitoses, which always take place near its dorsal surface. This can of course only be determined where the graft has not formed a complete tube. In nearly all cases where it can be determined, the neural tissue has maintained its original polarity, whether dv. (Pl. I, fig. 3) or dd. (Pl. I, fig. 4) to the host. This is the case in twenty dv. and in eight dd. specimens. In only three cases does this rule seem to be broken, but in these it is easily explicable by the folding and shifting which the graft has obviously undergone. The presence or absence of endoderm on the graft makes no difference to the fixity of the polarity.
These experiments do not of course test whether the dorso-ventral polarity of the epiblast in the primitive streak stage is determined. It is quite unknown whether, in the position where the graft was placed, there exists any influence which would tend to orientate the graft epithelium either way. There is in fact no a priori reason why, if there is any influence, it should act one way rather than the other. For since the graft is of epiblast, it might be supposed to come under an influence tending to orientate it like the host epiblast, that is dd. ; or it might be supposed to come under the influence which orientated the endoderm, in whose neighbourhood and in close association with which the graft often lies, that is dv. But the graft certainly always (with three possible exceptions) maintains its orientation, either dv. or dd., so if there is any influence tending to orientate the graft in either direction, the polarity of the graft is determined, and if there is no such influence, then the polarity of the graft is at least labilely determined.
In Amphibia, the behaviour of grafts under comparable conditions is not known. The outer-inner orientation of epidermis in Amphibia can be reversed when brought under a strong influence in that direction (by grafting pieces inside out into other embryos) at least up to the stage of the neurula (Luther, 1934). And the endoderm is known to be particularly responsive to external conditions in its orientation (Holtfreter, 1933).
A few grafts were made in which the antero-posterior orientation was noted (seven ap. grafts, three aa. grafts, one a.-side graft). No difference was found between ap. and aa. grafts in the width of the neural plate in various regions of the graft, nor in any other character.
Induction of neural tissue occurs irrespective of the place of origin of the graft. Presumptive epidermis and area opaca ectoderm have already been shown to be competent (Waddington, 1932, 1934). This is confirmed in the present experiments, and even the extreme outer edge of the area opaca proves to be competent (CP); and, more important, presumptive mesoderm can also be induced to neural tissue (BF, DI, Pl. I, fig. 4, Text-fig. 9). The competence of presumptive mesoderm is discussed below (see Discussion).
(d) Induction of mesoderm
If the graft has ever formed mesoderm, then on the whole it has conformed to the host, that is to say, mesenchyme axially in head and primitive streak regions, somites axially in the trunk region, and mesothelium laterally. Such graft-derived mesoderm would, unlike neural tissue, be in the normal position for this tissue in the host. But because of the slight irregularities of mesodermal tissues (extra tissue and extra organs) always present in blastoderms in vitro (and probably, like other irregularities, accentuated by the presence of the graft) it is generally impossible to be certain that extra tissue is graft derived. In no case is there adequate evidence that mesothelium, head mesenchyme, or somites (or notochord also) have been formed by the graft. But the graft has in several specimens lost, to a greater or lesser extent, its epithelial structure, and has become a mass of mesenchyme often associated with broken-up epithelium. This occurred in those grafts which were put under the posterior end of the host primitive streak (DG, DN); in the posterior part of several grafts when this part lies in the primitive knot region of the host (DY, BX) ; and occasionally in grafts which lie outside the axis of the host (DC). It seems, therefore, that in those cases where the graft rests a long time in the host mesenchyme, without the action of the host neural evocator (e.g. a graft in the posterior part of the host primitive streak), it eventually breaks up more or less into mesenchyme. Where, however, the graft has only been in the mesenchyme a fairly short time, as in the specimens fixed from 3 to 11 hours after the operation (e.g. EE), it remains fully epithelial. And where the graft is soon acted on by the evocator, whilst still an epithelium, it becomes probably entirely neural. This graft-derived mesenchyme, since it closely resembles the primitive streak mesenchyme in which it lies, may be considered as homoiogenetically induced mesodermal tissue.
IV. EFFECT OF THE GRAFT ON THE HOST
The graft seems to cause a considerably greater proportion of failures in development than is usual in cultures of chick embryos. This is roughly speaking more frequent the larger the graft used, and, in particular, large grafts from the area opaca, which wrinkle up considerably, suppress development very often. A possible cause of the failures to continue development may be that the graft obstructed the backward movement down the host axis: An intense thinning of the blastoderm immediately behind the graft was frequently observed after about 6-12 hours’ cultivation, suggesting perhaps that tissues were drawing away to make room for the backward streaming of the axial tissues which was not taking place.
One of the most common effects of the graft, when it lies in the head of the host, is to prevent the formation of the head-fold. This occurred in twenty-six out of thirty specimens, which is a much higher proportion than in normal in vitro development. It may well be a purely mechanical effect.
One case (CG) occurred where the forward-growing head-process mesoderm had apparently split on the graft, so that two probably imperfectly regulated half-heads were formed. Each half-head has its own head-fold, a symmetrical closed foregut derived from the foregut diverticulum of its side, and the ventral aorta of its side symmetrically slung in its myocardium beneath the foregut. The separate foreguts appear considerably posterior to the bifurcation of the neural plate, each accompanied by its own head mesenchyme; but anteriorly two neural plates also become distinct. The graft is comparatively small in this region, and dies away soon after the split starts; it was of area opaca ectoderm, which by the crumpling up it undergoes might easily be an obstacle to the head-process mesoderm (Text-fig. 10).
The only other profound effect of the graft is its influence on foregut morphology when it lies in the head. In some specimens the formation of the foregut diverticula is entirely or partially suppressed. In that case the thickening of the endoderm representing the lateral walls of the diverticula nevertheless takes place, but now on a flat sheet of endoderm (CG). More usually the diverticula are formed. But they are frequently very long, both absolutely, and very markedly in comparison with the dorso-ventral depth of the embryo, which, because of the usual absence of the head-fold, is less than normal. They are often long enough to reach the ectoderm, below which they may become flattened (AR) ; and they may show great contortion and extend far sideways (Al), and have unusually thickened walls.
In many specimens the foregut gets no further than the formation of these diverticula (AD, BH, AR). But a closed foregut may be formed even though the host is greatly flattened, without head-fold, and a bulky graft is present in the head This closed foregut may be above or below the graft. If below, it is apparently fairly normal, though it may be so flat that it has no trace of lumen (BF). If above the graft, it may be connected with one or both diverticula at its posterior end, the diverticula curving round the graft to reach the closed foregut above; in this case it must have been formed in a most abnormal way (CU, BX, Text-fig. 11). The graft may, when it lies in the head as a plate, form part of the roof of the closed foregut (DY), as in the same way it may replace the axial strip of endoderm in the trunk region.
V. DISCUSSION: INDUCTION AND INCORPORATION
Mangold (1924) made grafts, of presumptive epidermis or of presumptive neural tissue, from gastrulae and neurulae of Triton, into other Triton gastrulae (mostly into the yolk-plug or lateral blastopore lip), in such a way that they were carried into the archenteron by the gastrulation movements, eventually forming part of the archenteron roof. The graft thus came to lie in or near the axial organs of the host; and it proceeded to differentiate into any (frequently several) of the following tissues: myotome, lateral plate mesoderm, pronephros, notochord, connective tissue. This differentiation was according to the situation of the graft in the host, and it was usually a homoiogenetic complementary induction. In several a tissue of no particular structure was formed by the graft, even when it lay against the myotomes, and occasionally autonomous (but still apparently homoiogenetic) inductions, particularly of myotomes, appeared. Grafts from neurulae differentiated according to their place of origin, and we are not therefore concerned with them. Similar results were obtained in some of the experiments of Spemann & Geinitz (1927), who implanted pieces of presumptive epidermis into the dorsal lip of early Triton gastrulae; the graft was carried in with the host mesoderm, and differentiated (by complementary induction) into myotome, notochord, and (unlike Mangold’s experiments) into neural tissue continuous with the host’s.
With regard to the relative positions of host organizing tissue and of graft, Mangold’s experiments, or at least many of them, closely resembled those described in the present paper: that is to say, the graft ultimately lay between the axial chorda-mesoderm and the endodermal gut-roof. In Mangold’s experiments, however, the graft was probably in most cases more pressed into the host axial organs; and in many of Mangold’s, and also in those of Spemann & Geinitz, the graft lay actually in the host axial mesoderm.
Bytinski-Salz (1929) extended the experiments of Mangold, and his results were much more nearly like those obtained in the present experiments on the chick. Instead of implanting the grafts he merely pushed them into the archenteron. The grafts were presumptive epidermis or neural tissue from gastrulae of various ages. Usually they eventually lay between axial chorda-mesoderm and gut-roof, that is, in a position comparable with the chick experiments. Here they were quite frequently incorporated, as in Mangold’s work, but on the other hand many (at any rate of the presumptive neural tissue grafts) differentiated into autonomous neural tissue. Often the graft was considerably broken up, some pieces being incorporated, others forming neural tissue. It seems highly probable that the neural differentiation was due to a downward evocation from the chorda-mesoderm.
The neural parts of the graft often formed part of the gut-roof, and were kept by the attachment of their edges to the endoderm from rolling up into a neural tube, precisely as in the chick grafts. Presumptive epidermis sometimes differentiated into epidermis, however, even when forming part of the gut-roof, a thing which did not happen in the chick. Bytinski-Salz suggests that the difference between his results and those of Mangold may be put down partly to the difference in the technique of operation—by using smaller pieces and actually implanting them into host tissues (causing thereby some damage) a much closer association of graft and host was obtained than by merely inserting rather large grafts into the archenteron–and partly it may be put down to the use of different species as donor.
If in the chick experiments results similar to those of Mangold had been obtained, that is, complete incorporation, the graft would not be detectable, since such incorporation involves only comparatively slight abnormalities of the host, such as occur frequently in blastoderms in vitro. The apparent disappearance of the graft, which has occurred in some specimens, is not sufficient evidence of incorporation. But partial incorporation, in rather the same way as that found by Bytinski-Salz, can be recognized when the incorporated part is continuous with an autonomous part. This has already been noticed as occurring to graft neural tissue (EE, Text-fig. 6), and it also occurs in grafts of presumptive epidermis placed well anterior to the axis of the host. The mesenchyme derivatives of the grafts in the posterior already described can also be considered as partially incorporated, since they merely add to the general mass of host primitive streak mesenchyme, and do not form an autonomous, demarcated graft region. The breaking up of an epithelium into mesenchyme is what normally happens during invagination of the mesoderm material at the primitive streak ; and it is probably also an essential first step before the epithelial graft can be induced to, and consequently incorporated into, mesodermal structures (and this, because of the position of the graft in these experiments, is usually the only kind of incorporation possible). Incorporation of graft-derived mesenchyme into somites or notochord, in the same way as the host mesenchyme forms the axial organs, might therefore follow (if the graft tissue remained in the axis) as these organs later on develop in the host, but most of the mesenchymal part of the graft would perhaps be incorporated into the side-plate mesoderm of the host. That extra mesenchyme in the axis does tend to be incorporated is shown by experiments where grafts of pieces of primitive streak are placed under the host primitive streak (Abercrombie & Waddington, 1937), when they show a much greater tendency towards incorporation than do ectoderm grafts, particularly if taken from the posterior part of the primitive streak (and therefore more likely to differentiate exclusively into mesenchyme). It is therefore probable that if a graft of presumptive epidermis could be made to invaginate at the primitive streak, it would become completely incorporated, and in the experiments of Waddington & Taylor (1937), when a graft of epiblast is placed in a hole cut in the primitive streak, the graft is in fact sometimes incorporated, after presumably invaginating with the mesoderm. Hence in the chick it is probable that incorporation into the mesoderm depends not so much on how closely graft and host are associated (as Bytinski-Salz suggests for Amphibia) but on whether the graft epithelium becomes dissociated into mesenchyme or not. And, as already mentioned under “Induction of Mesoderm” (III (d)), this seems to depend on how long the graft lies in the host mesenchyme without being evocated.
Though the bulk of the graft is nearly always accounted for by visible graft tissue, it is clearly impossible to rule out incorporation of the type found by Mangold and Bytinski-Salz. But this cannot be further profitably discussed until some method for distinguishing graft and host tissues is known.
What has usually happened in these experiments is of course a heterogenetic autonomous induction of neural tissue from the graft epithelium by the host axial chorda-mesoderm. The induction has taken place downwards from the overlying host chorda-mesoderm. Such induction downwards relative to the inducing tissues occurred, of course, in the fundamental organizer experiments in the chick, when grafts of primitive streak, placed dv. in the area pellucida of the host, induced a neural plate above themselves (i.e. morphologically ventral to themselves, see Waddington & Schmidt, 1933). In the experiments of Bytinski-Salz a similar downward-acting induction is supposed to have occurred.
It appears that all epithelial parts of the chick epiblast are competent to form neural tissue, when directly acted on by the evocator, whatever their presumptive fate. The fact that presumptive mesoderm behaves just like presumptive epidermis, in that it can be induced to form neural tissue if its epithelial structure is maintained, should be emphasized. In the Amphibia also there are examples showing that if presumptive mesoderm is left on the surface of the gastrula it can be induced to form neural tissue. Thus Bruns (1931), by disarranging the presumptive tissue areas through the removal of small portions, showed that presumptive mesoderm could come to lie in the presumptive neural region, and there be induced to neural tissue. Lopaschov (1935) grafted organizer material into the neural plate of a neurula, and showed that it could there become neural tissue. And finally Tôndury (1936), by making inversion transplantations, obtained ortsgemäss differentiation of presumptive chorda when it lay in other regions, especially in the neural region. It can therefore be suggested that before its invagination at the primitive streak in chicks, as at the dorsal lip of Amphibia, there has been no final determination of the presumptive mesoderm (though there may have been a labile determination).
The chick differs from the Amphibia in the strong tendency of the graft to remain an autonomous epithelium; and there is probably also a difference in the subsequent incorporation of the graft-derived structures into the host organs, which seems to occur more readily in Amphibia than in chicks. This is reflected in the ordinary organizer grafts. In Amphibia, provided host and donor are of nearly allied species and about the same age, implantation of a dorsal lip produces a single fairly normal axis, host and graft tissues being intermingled by complementary induction. In primitive streak grafts in the chick, however, it is more usual for graft and induced structures to maintain their autonomy.
ACKNOWLEDGEMENTS
I should like to express my gratitude to Mr C. H. Waddington for suggesting this work, for his help while it was in progress, and for reading and criticizing this paper. I am very grateful to Dr Honor Fell for her kindness in allowing me to work in the Strangeways Research Laboratory, Cambridge; to Prof. E. S. Goodrich for reading this paper; and to Mr J. Armstrong for assistance with the photographs.
REFERENCES
EXPLANATION OF PLATE I
That is, an induction where the induced tissues are built into a normal whole with the inducing tissues. This is not so in an autonomous induction (Mangold, 1932).
Using incorporation to mean the building of the graft tissues, whether induced by the host or not, into a normal whole with the host tissues.
The various specimens are throughout referred to simply by their protocol designations, consisting of two capital letters.
I am indebted to Dr Julian Huxley for this term.
Dv. (dorso-ventral) grafts have their outer-inner axis the reverse of that of the host, that is, they are upside down; dd. (dorso-dorsal) grafts have this axis the same as that of the host. For the antero-posterior axis the corresponding conditions are expressed by the abbreviations ap. and aa.respectively.