J. Embryol. exp. Morph. Vol. 34, 3, pp. 559-574, 1975 559 Printed in Great Britain Mechanical stresses and morphological patterns in amphibian embryos By L. V. BELOUSSOV 1, J. G. DORFMAN 1 AND V. G. CHERDANTZEV 1 From the Department of Embryology, Moscow State University SUMMARY 1. Embryos of Rana temporaria have been dissected and shape alterations of different parts of the embryo, taking place within 1 h of separation, have been studied. Two categories of deformation have been revealed. 2. The first category comprises those deformations which take place immediately after separation. They are insensitive to cooling, cyanide and Cytochalasin B treatment. These deformations, which consist of a shortening of initially elongated cells, are considered to be the passive relaxations of previously established elastic tensile stresses. 3. Deformations of the second category proceed more slowly. They are inhibited by cooling, cyanide and Cytochalasin B treatment, are accompanied by elongation and migration of cells and occasionally lead to rather complex morphodifferentiations of isolated fragments. These processes are considered to be the result of the active work of intracellular contractile systems, either pre-existing or induced de novo. 4. By analysing the arrangement of the passive deformations we have constructed maps of mechanical stresses in embryos from late blastula up to the early tail-bud stage. At several embryonic stages drastic transformations of the stress pattern occur, these transformations being separated by periods during which the pattern of stress distribution remains topologically constant. 5. A correlation between the arrangement of stress lines and the presumptive morphological pattern of the embryo is pointed out. 6. Some possible relations between tensile tissue stresses and active mechanochemical processes are discussed. INTRODUCTION The role of mechanical stresses in the orientation of cell movements in tissue culture was established long ago (Weiss, 1929), but the application of similar principles to morphogenetic processes in entire embryos has been delayed by lack of data on the existence and localization of stresses in intact tissues. This gap has recently been filled by the demonstration of intracellular contractile systems in developing rudiments (Baker & Schroeder, 1967; Wessels et al. 1971;Burnside, 1971). Similar contractile fibrils have been revealed in cells of the distal regions of the actively growing hydroid polyps (Hale, 1960). Periodical contractions of 1 Authors' address: Department of Embryology, Moscow State University, Moscow 117234, USSR.
560 L. V. BELOUSSOV AND OTHERS these fibrils gave rise to pressure stresses in cell layers. Growth, morphogenetic shape alterations and some kinds of rudiment interactions in these species were considered to be directly determined by these stresses (Beloussov & Dorfman, 1974). These data emphasize the role which mechanical stresses may play in regular geometrical alterations occurring during development. Therefore, it seems desirable to study them in other developing systems. The simplest method of revealing mechanical stresses in embryos consists of detecting the tissue deformations just after dissection. These deformations may be considered as direct manifestations of pre-existing stresses. A number of scattered data on the existence of similar deformations have already been obtained. These are: contraction of dissected blastomere furrows in Loligo eggs (Arnold, 1971); contraction of dissected periblast in Fundulus eggs (Trinkaus, 1969); shape alterations of isolated vegetal parts of sea-urchin blastula (Moore & Burt, 1939); and, what pertains closely to this study, the intensive rolling of the layers of dissected eye vesicle in amphibian embryos (Lopashov, 1963). A similar dissection method, combined with the action of some physiological agents and followed by a histological study of intact and separated tissues, has been employed in this work. As a result, a fairly regular pattern of mechanical stresses has been revealed in amphibian embryos from late blastula up to early tail-bud stages. Maps of mechanical stresses have been constructed for successive developmental stages, representing the lines of tensions and the points of branching of these lines. Besides the above mentioned 'immediate' deformations, whose passive relaxatory characters have been confirmed by their non-sensitivity to some inhibitory agents, another class of slower (although still complete within 1 h) and much more sensitive deformations has also been observed in separated tissues. The latter processes were obviously active, leading in several minutes to the creation of new tensile stresses and to rather complex morphodifferentiations of a fragment. It is proposed that they may be considered as simplified models of normal morphogenesis. MATERIALS AND METHODS The majority of the experiments have been made on Rana temporaria embryos from late blastula up to early tail-bud stages. Several operations were made also on Rana esculenta and Xenopus laevis embryos of the same stages. The experiments consisted of complete isolation, three-side separation or dissection of different parts of embryonic tissues. Under physiologically-normal conditions embryos and separated fragments were kept at 18-20 C in normal Holtfreter solution twice diluted (several operations were performed in full-strength Holtfreter solution and gave the same results). The influences of factors inhibiting metabolism (cooling, moderate doses of KCN) and of Cytochalasin B upon rapid deformations have also been studied. In cooling experiments fragments
Mechanical stresses in amphibian embryos 561 and donor embryos were kept in the same solution at 2-5 C for not less than 10 min before operation, during the operation proper and throughout the whole period of observation. KCN was used in a similar way in concentration 10 mg/ ml (10~ 4 M). In Cytochalasin experiments donor embryos and fragments were placed in 0-2 mg/ml (0-4X10~ 6 M) solutions of Cytochalasin B in dimethylsulphoxide (DMSO). The DMSO concentration was 0-4 % aqueous solution. Control embryos were placed in DMSO solutions of the same concentration. Deformations of tissue fragments have been studied both visually and by means of time-lapse filming (35 mm film, exposure interval 1 sec). The rolling-up of the isolated fragments with their external surface outside was designated as positive rolling and that with external surface inside as negative rolling (or bending, folding). A surface facing the gastrocoel cavity was considered as an external one. For histological purposes embryos or tissue fragments were fixed in Bouin's fluid, totally stained with borax carmine, dehydrated in butyl-alcohols and embedded in paraffin-wax. Special attention was paid to the possible deformations of separated fragments during fixation and subsequent treatment. In the great majority of cases no significant post-vital deformations were observed. The following well-known features of the histology of amphibian embryos are relevant to this study. The ectoderm is bilayered, consisting of external epiectodermal and internal hypoectodermal layers (terminology, from Detlaff, 1938). The hypoectoderm of the neural plate is several cell layers thick, whereas in other regions it is a single cell layer. As a rule, the adjacent rows of epi- and hypodermal cells are staggered, so that, in a section, each hypodermal cell is opposite the junction between two epi-cells. This regular arrangement is disturbed in several narrow zones, most of which correspond to slight ectodermal folds to be described in detail later. RESULTS Immediate deformations External appearance of the immediate deformations (1) Late blastula stage (Fig. 1: 1 A). An extirpated and dissected fragment of the marginal zone immediately unfolded to an angle of 90 approximately. An isolated superficial layer of the vegetal hemisphere bent slightly in the negative direction (Fig. 1: 1B). No significant deformations took place in other regions. (2) Middle gastrula stage (Fig. 1: 2 A). When any sector of the marginal zone (lip of blastopore) was extirpated and separated in caudal direction (towards the lip) the fragment walls immediately opened out to about 45-90 (Fig. 1: 2 A-B, cr-caud). When the fragment was separated in a reverse, cranial direction (away from the lip) the divergence angle did not exceed 10-20 (Fig. 1: 2 A-B, caud-cr). When separated from the external layer, the already invaginated material slightly expanded.
562 L. V. BELOUSSOV AND OTHERS B C D I I CSS Fig. 1. For legend see opposite page. The surface cell layer of an entire gastrula, when separated in the immediate vicinity of the marginal zone, immediately bent as shown in Fig. 1: 2B, epi. In other areas no significant deformations were observed. (3) Late gastrula stage (blastopore almost closed) (Fig. 1: 3 A). The results of caudal separation of the dorsal lip zone were the same as in the preceding stage, but now the angle of the opening of the fragment walls was also the same during
Mechanical stresses in amphibian embryos 563 D postbrf. prebrf. (5) (5) Fig. 1. Rapid deformations after dissections of amphibian embryos, (A) 1-12, schemes of separations; (B) immediate deformations; (C) deformations taking place 1-5 min; (D) 5-20 min; (E) 20-60 min after separation. Black wedges indicate directions of cutting; numbers in parentheses denote how many operations of each type were made. Dotted outlines in column A, lines 1-3, 5-8, indicate the areas extirpated; 6-8, latent deformations of the same directions as the preceding immediate deformations, caud.-cr., Caudo-cranial; cr.-caud., cranio-caudal; chin., chordamesoderm; ent., entoderm; epi., epiectoderm; hyp., hypoectoderm; npl., neural plate; postbrf., postbranchial fold; prebrf., prebranchial fold; snf., subneural fold. For other designations see text. 36 EMB 34
564 L. V. BELOUSSOV AND OTHERS cranial separation, revealing a new tensile stress anterior to the lip. Dorsal epiectoderm, when separated in the medial direction, now revealed a slight but distinct negative bending (Fig. 1:3B, epi). A similar separation of the chordamesoderm led to its immediate negative rolling (Fig. 1:3B, chm). The same was true of the ventral gastrocoel wall (Fig. 1:3B, ent). (4) Early neurula stage. By this time some earlier tendencies have been reinforced and several new areas of rapid deformation have arisen. (4a) Anterior (hind-brain) area (Fig. 1:4A). Medial separation of neural epiectoderm led to a sharp negative bending, localized somewhat laterally to midline (Fig. 1: 4B, epi). Neural hypoectoderm bent during a similar operation to a smaller extent. At this stage the neural plate becomes limited laterally and anteriorly by a shallow ectodermal fold which can be designated as the subneural fold (Fig. 1:4A, snf). This fold became much more pronounced immediately after separation of its epiectoderm (Fig. 1:4B, snf). Medial separation of the chordamesoderm also led to its extensive negative bending close to the midline (Fig. 1: 4B, chm). (Ab) Posterior (trunk) area (Fig. 1:5A). A similar separation of neural epiectoderm also led to its negative bending, localized in this case just at the midline (Fig. 1: 5B, epi). Neural hypoectoderm bent in the same direction, but to a less extent (Fig. 1: 5B, hyp). The chordamesoderm behaved as in the anterior area. So far no significant deformations were revealed in the subneural fold region at that level. (4 c) Deformations observed during longitudinal (cranial and caudal) separations at early-middle neurula stages (Fig. 1: 6 A). Cranial dissection of the hind part of the dorsal embryo wall resulted in extensive divergence of neural plate and chordamesoderm. In the head region a similar caudal separation led to an extensive negative bending of the chordamesoderm (Fig. 1: 6B). After extirpation of the whole dorsal embryo wall the initial slight positive bending of the neural plate increased to a certain extent (Fig. 1: 6B, npl). The completely isolated chordamesodermal layer reproduced all the bendings observed during partial separations. (5) Middle-late neurula stage: deformations during transversal separations (Fig. 1: 7A). As in the preceding stages, medial separation of the epiectoderm of the invaginated neural plate led to its extensive negative bending, whereas that of the neurohypoderm resulted in slight negative bending. Separation of subneural fold tissues (including mesoderm) resulted in their negative bending (Fig. 1: 7B, right side). A similar result was observed at early tail-bud stage. (6 a) Early tail-bud stage: dissection of a just closed neural tube (Fig. 1: 8 A). Dissection of the neural tube along its midline led to the immediate divergence of its walls. A similar result was obtained after transverse cutting of a tube wall (Fig. 1:8B). (6 b) Early tail-bud stage: separations in horizontal (frontal) directions
Mechanical stresses in amphibian embryos 565 (Fig. 1: 9 A). At this stage several new transversal ectodermal folds appear in addition to the subneural folds, namely postbranchial, separating branchial and trunk regions, prebranchial, separating branchial and head region, oral fold (mouth rudiment) and two less-pronounced folds, separating the rudiments of the branchial protuberances. When separated as shown at Fig. 1 (6 A), all these folds immediately became much more pronounced (Fig. 1: 9B). This was also true of the corresponding entodermal folds (Fig. 1: 9B, ent). (7) Immediate deformations of non-folded areas of ventral and lateral ectoderm and mesoderm (Fig. 1: 10-12A). The behaviour of these tissues did not alter significantly during the whole developmental period under study. Ectodermal fragments, excepting those taken anteriorly to the prebranchial fold (from the presumptive oral field), contracted approximately to half their size and often slightly rolled in a negative direction, thus forming a shallow cup (Fig. 1: 10B). Considerable tension was shown to be present in the presumptive oral field ectoderm, since its epilayer did not contract much after dissection whereas the hypolayer did. Anterior to the prebranchial fold from the early tail-bud stage onwards, and anterior to postbranchial fold from the middle tail-bud stage onwards, immediate positive rolling of the epiectoderm was observed (Fig. 1: 9B, lower embryo wall). In purely mesodermal and combined ectomesodermal fragments no significant immediate contractions or other shape alterations were observed (Fig. 1: 11-12B). Cooling by up to 2 C did not influence the deformations listed in paragraphs 3-7 above in any way. Negative bendings of neuroectoderm occurred also in KCN solutions of the concentrations employed. The opening angle of the dissected gastrula marginal areas slightly decreased after cooling. It is worth mentioning that in Cytochalasin B solutions, deformations 4-7 above were completely retained and were even more pronounced than normal. This result contrasts sharply with those described below where slower deformations are inhibited completely by the same agent. During immediate deformations the previously elongated and/or oblique cells contracted into rectangular and sometimes even spherical shapes. The sharpening of pre-existing negative folds and similar bendings in other regions seem to be a direct result of the mechanical pressure of these contracted cells upon the adjacent ones. Slower ('latent'') deformations (1-60 min after separation) The immediate deformations described above were stable only at low temperatures and under the influence of Cytochalasin B. Under normal conditions the shape of the fragments seen 0-5-1 min after separation changed again after a further interval. These rather complicated morphological events of about 1 h duration are what we call latent deformations. They in turn fall into two categories. 36-2
566 L. V. BELOUSSOV AND OTHERS (A) Latent deformations in the same direction as the preceding immediate deformations (Fig. 1, lines 2, 6, 7, 8C-E, framed pictures) (1) Prolongation of divergence of dissected marginal area walls. Under normal physiological conditions the immediate opening of the dissected blastopore lip was followed by a slower divergence of its walls, leading in 20-30 min to the complete flattening of the fragment (Fig. 1: 2B-D). At the former tip of the lip a new blastopore arose (Fig. 1: 2E). (2) Foldings and closure of neural rudiments. The entire neural plate, extirpated in the early neurula stage, did not significantly alter its shape in 1 h (Fig. 1: 5B-E, npl). On the other hand, the anterior part of the neural plate, extirpated in the middle neurula stage, began to fold continuously along its midline; and in 1 h it had almost closed (Fig. 1: 7B-E, npl). At the same time, the neural plate bent transversely, thus increasing the sharpness of the immediate bend already there (Fig. 1: 6 compare B and D). The walls of the dissected neural tube, after their immediate divergence, slowly converged in several cases, completely closing again in 15-20 min. In most cases, however, the dissected tube walls continued to diverge (Fig. 1: 8D). All the processes described are obviously similar not only to the preceding immediate deformations but also to the corresponding normal morphogenetic processes. We believe this to be true not only for the closure of the neural rudiments but also for the opening of the blastopore lips, since the latter process may play a significant role in normal gastrulation, promoting the invagination of cell material. It can be deduced therefore that these latent deformations are determined by certain pre-existing mechanisms rather than induced by the operation itself. (B) Latent deformations of opposite direction relative to the immediate deformations (1) Deformations of ectodermal fragments. Under normal conditions 0-5- 2 min after extirpation all the ectodermal fragments, including the neuroectoderm, began to roll intensively in the positive direction, starting from the free edge (Fig. 1,10C-E). The rolling was unequal, with sharp curvatures alternating with practically flat areas. In 20-40 min the rolled fragments were subdivided by narrow furrows, oriented in most cases transversely to the rolling axis (Fig. 1: 10D). After a time some of these furrows disappeared whereas others stabilized (Fig. 1: 10E). (2) Deformations of purely mesodermal and ectomesodermal fragments (Fig. 1: 11-12C-E). One to 2 min after separation the naked surfaces of the fragments began to contract, thus leading to the bending of sufficiently large fragments (Fig. 1: 11C). In 10-20 min the smaller fragments transformed into smooth spheres (Fig. 1: 12D, a), the somewhat larger ones formed a single invagination, and the largest ones were subdivided by several furrows, oriented
Mechanical stresses in amphibian embryos 567 Table 1. Numbers of internally (int.) and externally (ext.) situated cells in mesodermal fragments 3 min and 30 min after fragment isolation (counts on several median sections) Number of fragment Time after isolation (min) Absolute numbers f int. 75 60 33 310 392 182 A ext. 264 145 158 205 339 264 int./ext. ( O/\ \ /o) 1 2 3 4 5 6 3 3 3 30 30 30 28 41 21 146 116 70 radially and localized in the marginal area of the fragment (Fig. 1: 11D, 12D, c, 12 E). During these transformations the cells contacting the naked surfaces elongated perpendicularly to them. Later on some of the cells obviously migrated inside the fragment, an indication of this being the number of cells situated outside and inside at different times after separation (Table 1). Formation of radial furrows began from the immigration of individual epiectodermal cells into the hypoderm. Some of these cells later on were connected to the opposite (naked) surface by elongated cells. In such a way the marginal area of the fragment separated into several parts which rapidly rounded and became almost completely isolated from each other (Fig. 1: 12E). If a fragment included a piece of an intact ectodermal fold (for example, subneural), it separated along this fold much more rapidly than in its absence (Fig. l:7c, fl ). The following simple experiment demonstrated the rise of new mechanical stresses during latent deformations. When dissecting an ectomesodermal fragment parallel to its surface immediately after its extirpation, no additional divergence of dissected edges, i.e. no relaxation movements, were observed (Fig. 1: 12B). However, 10-15 min after separation, a similar dissection led to the immediate extensive rolling of the mesodermal layer (Fig. 1: 12 C, a), as well as of the marginal area of the ectodermal layer (Fig. 1: 12 C, b). Both deformations seem to be determined by the contraction of the previously stretched transverse cell walls. As with other immediate deformations, these were not influenced in any way by cooling. (3) Action of inhibiting agents on latest deformations. All the latent deformations were inhibited by cooling, KCN-treatment and Cytochalasin B. Under these influences the fragments either remained flat or retained the folds established immediately after separation. The action of cooling and Cytochalasin B was completely reversible, whereas that of KCN was irreversible. Cytochalasin B led also to a slow unrolling of already rolled ectomesodermal fragments.
568 L. V. BELOUSSOV AND OTHERS Moderate cooling (7-10 C) promoted the irregular furrowing and subdivision of ectodermal and ectomesodermal fragments. DMSO in the concentrations employed had no significant influence on the processes under study. DISCUSSION Passive and active deformations It may be seen that all the rapid deformations described may be naturally divided into two categories. The first one comprises those deformations which take place immediately after separation, are at the same time insensitive to the inhibiting influences employed, and consist in shortening of the previously elongated and/or oblique cells. Deformations of the second category on the contrary proceed more slowly, are highly sensitive to inhibiting agents and are accompanied by elongation and even migration of cells. Thus the first category may be considered to be passive elastic relaxations of pre-existing stresses whereas the second are active, energy-requiring processes, obviously connected with the contraction of microfilaments. On the other hand, some of these latter processes are due to pre-existing active mechanisms (morphogenesisimitating latent deformations; see Fig. 1C-E, framed pictures), whereas the others are non-specific and are obviously caused by contractile systems activated by the operation itself (see also Burnside, 1972, 1973). The active deformations may be of interest as simplified models of morphogenetic processes, since they demonstrate rather rapid and complex morphodifferentiation and provide useful information on the mechanism of activation of intracellular contractile systems. In this paper, however, the first category of deformations will be mainly discussed in so far as it may be of use in constructing maps of the mechanical stresses existing over successive periods in the development of amphibians. Maps of mechanical stresses To validate the employment of immediate deformations for the construction of stress maps the cellular basis of these deformations must be discussed in greater detail. The existence of immediate relaxatory movements indicates that the cells of amphibian embryonic layers are considerably deformed by the adjacent cells. The layers as a whole are also deformed by the surrounding tissues. The mechanically stable cell shape achieved after separation is approximately cuboidal when the integrity of the cell layer is not disturbed and approximately spherical when cell connexions are disturbed during the operation, or in the regions without a regular layer structure. Let us consider now several somewhat idealized cases of relaxatory movements and their interpretation (Fig. 2). (1) Extirpation of fragments leads to their considerable shortening without
Mechanical stresses in amphibian embryos 569.QrP V>} * * * * * fh i \ \ \ \ \ vvv\ r Fig. 2. Main types of immediate relaxatory movements in cell layers. Dissections are indicated by dotted lines, stretched surfaces by heavy lines, the surfaces elastically relaxing after dissections by asterisks. In E the largest angle a corresponds to the greatest elastic contraction at left from dissection point. For other designations see text.
570 L. V. BELOUSSOV AND OTHERS C E G Fig. 3. Maps of mechanical stresses for several successive stages of Rana temporaria development and for a typical ectomesodermal fragment. (A) Late blastula; (B) mid-gastrula, sagittal section; (C) same stage, transversal section (along the line, indicated on B); (D) transition from gastrula to neurula, posterior region; (E) anterior region of early neurula; (F) posterior region, same stage (D-F - transverse sections). (G) Early-middle neurula, sagittal section; (H) middle-late neurula, frontal section; (I) similar stage, transverse section. (J) A typical ectomesodermal fragment, 10-15 min after its isolation. Heavy contours represent distinct stress lines; dotted contours, dispersed stress-lines; fine lines, non-tense surfaces, separating embryo layers; pre-fo\d, corresponding to plica rhomboencephalica, pe\<-fo\d, corresponding to plica encephali ventralis. any bending (Fig. 2 A). This indicates that both surfaces were elastically stretched to the same extent. (2) A similar operation leads to negative bending without any shortening of lateral cell walls (Fig. 2B): the elastic stretching of the external surface was greater than that of the internal surface. (3) A similar operation leads to sharp negative bending with considerable shortening of the lateral cell walls, up to complete cell rounding (Fig. 2C): both external and lateral cell walls are stretched. (4) Extirpation of an almost flat fragment leads to its strictly localized negative folding (Fig. 2Dj), whereas dissection of an initially bent rudiment leads to its unfolding (Fig. 2D 2 ); both processes are accompanied by extensive contraction and rounding of initially stretched cells underlying the folded zone. This indicates that, along with the tension of external and internal surfaces, there exist tension line(s) going down from the fold and crossing the cell sheet.
Mechanical stresses in amphibian embryos 571 (5) Three-sided separation leads to a sharp negative bending localized exactly at the border between separated and non-separated areas but without any definite prelocalization in the intact embryo (Fig. 2E). This demonstrates the tension of the external surface of the separated zone and the existence of an elastic contraction zone somewhere near the zone of separation. (6) A similar operation leads to negative bending (Fig. 2F) or rolling (Fig. 2G) of the separated area accompanied by initially oblique cells becoming symmetrical. This indicates the uneven stretching of both external and internal surfaces and the elastic stretching of lateral cell walls. Hence all the above deformations point to the stretching of at least one of the layer surfaces, whereas deformations (3), (4), (6) indicates the existence of tension lines which cross the cell layer(s) (cross-lines). In other words, in the latter cases, the bifurcation of tension lines takes place at the corresponding points. The maps constructed by these methods are presented at Fig. 3A-J. The distinct lines of tensile stresses are indicated by heavy lines, those gradually dispersing throughout the tissues by dotted lines and non-tense surfaces by fine lines. The following general properties of stress patterns are to be emphasized. At any developmental stage the tension lines, including cross-lines, are concentrated near to the restricted number of closed surfaces subdividing the embryo. This tensile pattern does not alter gradually in the course of development. Instead it remains topologically constant for a certain finite period of development, and then drastically transforms. The transformations comprise the appearance of new cross-lines as well as (more rarely) the disappearance of some old ones. The periods of development between the two next topological transformations may be designated similarly as topologically invariant periods. Successive topological transformations for the given period of Rana temporaria development are as follows: (1) The establishment of fairly wide strips of stretched cells between the blastocoel corners and the vegetal surface of the late blastula (Fig. 3 A). Their positions correspond to the marginal zone of the gastrula. This stress pattern is not changed qualitatively during gastrulation, although stretching of the dorsal ectoderm increases (Fig. 3B, C). In the regions removed from the blastopore, the stress pattern is circular and is localized mainly in the ectodermal layer (Fig. 3 C). (2) The next and perhaps the most important topological transformation is that the circular stress breaks either exactly along the dorsal midline (in the posterior region) or parallel to it (in the anterior region). Now the external circular stresses pass along these lines to the archenteron roof and spread ventrally, joining stress lines coming from the gastrocoel angles and then gradually dispersing (Fig. 3D). This transformation may be considered as the demarcation point between gastrulation and neurulation. (3) Shortly after, in the early neurula stage, the stretched cells of the dorsal epiectoderm and mesoderm establish new contacts with the lateral embryo
572 L. V. BELOUSSOV AND OTHERS surface just ventrally to the neural plate. This contact line corresponds to the subneural fold which encircles the embryo laterally and anteriorly. Now several other obliquely situated cross-lines dissect the neural plate (Fig. 3E,F). Thus, instead of extended gradually dispersing tension lines, a number of closed tense contours appear. (4) In the early-mid neurula stage the neural plate together with archenteron roof becomes dissected by a number of new cross-lines (Fig. 3 G; compare with Fig. 1, 6B). Several fairly irregular indistinct cross-lines are localized in the caudal region and one especially pronounced cross-line is found at midbrain level (Fig. 3 G, pre-pev). (5) In late neurula-early tail-bud stage several new transversely oriented cross-lines arise in the anterior body region, joining post-, prebranchial and oral folds with the corresponding folds of the oral ectoderm (Fig. 3H). Later on similar cross-lines dividing the rudiments of the branchial arches appear. On the other hand, the cross-line connecting the subneural folds and neural groove almost disappears during neurulation and is replaced by a line localized slightly ventral and going along the roof of the intestinal cavity (Fig. 31). According to the dissection results described above, the tensions of the presumptive oral field ectoderm are localized mainly in its hypo-layer, the episurface bearing no considerable tension (Fig. 3H). A similar map for a typical ectomesodermal fragment, 10-15 min after its extirpation, is presented at Fig. 3 J. The presumptive significance of stress cross-lines: possible relations between mechanical tensions and the active mechanochemical processes in cells It is easy to see that almost all the cross-lines revealed in intact embryos are of clear morphological significance. Indeed, stress lines established in the late blastula outline the marginal zone of the gastrula; those established during the second topological transformation correspond to the neural groove and separate the archenteron roof into its chordal and mesodermal parts. The third topological transformation leads to complete separation of the neural plate from the more ventral regions, whereas a set of transformations (4) results in the differentiation of the tail-bud (in the posterior body region) and in the appearance of highly specific bendings in the head region. Thus in Fig. 3H pre corresponds to the so-called plica rhomboencephalica between mid- and hindbrain, whereas pev corresponds to the plica encephali ventralis, marking the anterior chorda extremity. The destiny of the post-, prebranchial and oral cross-lines is clear from their designations. Therefore, the cross-lines mark the borders between the embryo rudiments, as a rule, much earlier than they become visibly differentiated. Moreover, in several cases - for example, in the neural plate and in the large mesoderm-including fragments - the directions of the cross-lines coincide with those of the active elongation and migration of cells; however, the latter processes are initiated later than the corresponding tensile patterns
Mechanical stresses in amphibian embryos 573 arose (compare the tense but as yet passive neural plate at Fig. 1 (5) with the actively folded neural plate at Fig. 1 (7), npl). The existence of such a correlation indicates that there may be a causal connexion between the mechanical stresses and subsequent activation of the intercellular mechanochemical machinery. These causal connexions, if proved, could be interpreted in terms of the 'positional information' concept (Wolpert, 1969). Further investigations are of course required to prove this hypothesis. On the other hand, reverse relations are conceivable. Indeed, according to the above data, the cross-lines are usually established along the direction of maximal layer stretching, which in its turn is caused by the activity of cellular contractile systems in the preceding period of development. Thus, during gastrulation the dorsal embryo surface stretches most often in a cranio-caudal direction, which coincides with the direction of the dorsomedial and subneural folds. Later on the active medial rolling of the neural plate stretches more ventral ectodermal areas in a ventro-dorsal (transverse) direction, which corresponds to the direction taken by the post-, prebranchial and oral folds. It is also conceivable that the creation of cross-line pre-pev (Fig. 3G) is promoted by the lateral stretching of the anterior part of the neural rudiment due to the backward folding of the anterior neural fold. Ectodermal and ectomesodermal fragments behave in a similar way: the folds created are situated along the directions of the maximal stretching of fragment surfaces (e.g. radially in isotropically rolled fragments). These considerations make it possible to hope that in the not-so-distant future a closed system of causal relationships between morphogenetic processes will be constructed including mechanical stresses as one of its important components. REFERENCES ARNOLD, J. M. (.197.1). Cleavage furrow formation in a telolecithal egg {Loligo pealii). IF. Direct evidence for a contraction of the cleavage furrow base. /. exp. Zool. 176, 73-85. BAKER, P. C. & SCHROEDER, Th. E. (1967). Cytoplasmic filaments and morphogenetic movements in the amphibian neural tube. Devi Biol. 15, 432-450. BELOUSSOV, L. V. & DORFMAN, J. G. (1974). On the mechanics of growth and morphogenesis in hydroid polypes. Am. Zool. 14, 719-734. BURNSIDE, B. (1971). Microtubules and microfilaments in newt neurulation. Devi Biol. 26, 416-441. BURNSIDE, B. (1972). Experimental induction of microfilament formation and contraction. /. Cell Biol. 55, 33 a. BURNSIDE, B. (1973). Microtubules and microfilaments in amphibian neurulation. Am. Zool. 13, 989-1006. DETLAFF, T. A. (1938). Neurulation in Anura as a complex morphogenetic process (Russian). Trudy lnst. eksp. Morf., Tbilisi 6, 187-200. HALE, L. G. (1960). Contractility and hydroplasmic movements in the hydroid Clytia johnstoni. Q. Jl Microsc. Sci. 101, 339-350. LOPASHOV, G. V. (1963). Developmental Mechanisms of Vertebrate Eye Rudiments. Oxford: Pergamon Press. MOORE, A. R. & BURT, A. S. (1939). On the locus and nature of the forces causing gastrulation in the embryos of Dendraster excentricus. J. exp. Zool. 82, 159-171.
574 L. V. BELOUSSOV AND OTHERS TRINKAUS, J. P. (1969). Cells into Organs. New Jersey: Prentice-Hall. WEISS, P. (1929). Erzwingung elementarer Structurverschiedenheiten am in vitro wachsenden Gewebe. Wilhelm Roux Arch. EntwMech. Org. 116, 438-554. WESSELS, N. K., SPOONER, B. S., ASH, J. F., BRADLEY, M. O., LUDUENA, M. A., TAYLOR, E. L., WRENN, J. T. & YAMADA, K. M. (1971). Microfilaments in cellular and developmental processes. Science, N.Y. 171, 135-143. WOLPERT, L. (1969). Positional information and the spatial pattern of cellular differentiation. J. theor. Biol. 25, 1-47. {Received 4 November 1974, revised 19 May 1975)