Experiments are described in which sectors including dorsolateral mesoderm from early-neurula-stage amphibian embryos are grafted to the mid-ventral region of gastrula-stage hosts. The grafted tissue pursues an autonomous developmental sequence, though integrated into the host mesodermal mantle, so that such embryos develop a ventral strip of ectopic somite tissue, occasionally with a pronephric formation at one side. When the proportions in which mesodermal tissue has been assigned to the four basic territories of the host’s mediolateral pattern are assayed, a significant deficit in somite is characteristically found, though the phenomenon is variable in magnitude. It seems that the size of the host’s prone-phric territory may be diminished in a similar way, if an earlier differentiating ectopic pronephros is already joined to the system. These phenomena are discussed in relation to theories of biological pattern formation.

Pattern formation during early development of vertebrates involves the assignment of particular determined states to successive groups of cells within a sheet of tissue. Cells’ fates in this process are determined by their positions, in relation to the amount of tissue available in each individual embryo and in relation to a restricted ‘organizer’ region (Spemann, 1938 review; Cooke, 1972b, 1979a, 1981). Lineage is not involved in the early designation of cell fate as it is in some other types of embryo (Laufer, Bazzicalupo & Wood, 1980). It has been proposed that such development is controlled by a simply shaped gradient in the level of some positional signal which is set up across the tissue, with local levels of the signal being ‘interpreted’ directly by cells in effecting developmental decisions (Wolpert, 1971). On this view, no specific interactions are believed to occur between emerging pattern parts to control their relative proportions. This latter control would occur automatically by the independent positioning of frontiers between the successive parts, according to threshold levels of the positional signal in the normal gradient profile across the original cell sheet.

In amphibian embryos, the medial-to-lateral pattern that characterizes the vertebrate mesoderm becomes established over a period when this layer is es-sentially a cylindrical jacket, one or a very few cells thick, migrating anteriorly between neurectoderm and yolky endodermal mass. It consists of mid-dorsal notochord and then paired somite, pronephric and lateral plate/blood island territories respectively, and these territories become fixed in a mediolateral sequence (see Fig. 1) across a period extending from early in gastrulation (notochord), probably into later neurula stages. Frontiers within the mesoder-mal population are positioned with considerable accuracy in normal develop-ment, in the sense that quite constant proportions of the cells are used to found each territory (Cooke, 1981, 1982). In this paper I present evidence that the proportion of cells devoted to the somite territory is specifically reduced in the host’s pattern when gastrulae develop after ectopic (i.e. mid-ventral) implanta-tion of additional determined somite tissue from neurulae. The effect is con-centrated in the region of the long axis lying opposite the differentiated ectopic pattern part, but size of the host pronephric territory can be similarly reduced by prior, ventral differentiation of pronephros in grafted material anywhere along the axis.

Fig. 1.

Determination of mediolateral pattern in amphibian embryos. Schematic cross sections of (A) midgastrular and (B) early larval stages of development are shown. The mesodermal cell layer and the four territories it forms in the developed pattern are stippled with decreasing density from mid-dorsal (arrowhead) to lateral or ventral. Stippling density indicates differentiation tendencies at the early stage where the mesoderm is essentially a monolayered cylinder; see the later territories notochord (NC), somite (Som), pronephros (PN) and lateral plate (L Pl), after the dorsal convergence movements and induction of the overlying nervous system (CNS). Except for notochord in the dorsal midline, however, determination of the precise positions of frontiers between the pattern parts in the cell sheet does not occur until later in gastrulation and neurulation. Scale bar represents 1 mm approx.

Fig. 1.

Determination of mediolateral pattern in amphibian embryos. Schematic cross sections of (A) midgastrular and (B) early larval stages of development are shown. The mesodermal cell layer and the four territories it forms in the developed pattern are stippled with decreasing density from mid-dorsal (arrowhead) to lateral or ventral. Stippling density indicates differentiation tendencies at the early stage where the mesoderm is essentially a monolayered cylinder; see the later territories notochord (NC), somite (Som), pronephros (PN) and lateral plate (L Pl), after the dorsal convergence movements and induction of the overlying nervous system (CNS). Except for notochord in the dorsal midline, however, determination of the precise positions of frontiers between the pattern parts in the cell sheet does not occur until later in gastrulation and neurulation. Scale bar represents 1 mm approx.

Such phenomena are not to be expected if a purely positional information mechanism controls the pattern, and I propose a model that departs from that idea for this particular system. This model shares with earlier ones of Rose (1957, 1970) and Braverman (1961) the proposal that substances or signals, produced specifically within each emerging pattern part, flood the system successively to form a record of its history and ensure that remaining uncommitted tissue elsewhere is diverted to give other parts. The idea was proposed in outline elsewhere because of recent data (Cooke, 1981,1982) on the pattern proportions in mesoderms developing at artificially small size, and in those shared between two organizing regions after Spemann dorsal lip grafts. These data are very difficult to match with the expected performances of plausible physicochemical systems such as might set up and regulate gradients, on the positional information hypothesis that the absolute levels of the gradient profiles attained would be directly ‘interpreted’ to give the observed pattern configurations. A pattern-control system of the type here proposed, by contrast, would explain both these and the present quantitative results. It is also consistent with a certain statistical feature of the pattern proportions within the whole population of embryos so far measured in my laboratory collection, which is presented for the first time.

The work utilized embryos of Xenopus laevis the S. African clawed frog, and Ambystoma mexicanum the axolotl. Very different cell numbers are involved in their primary mesoderms, their rates of development differ more than twofold, and appreciable differences in morphogenetic movements of gastrulation may place these amphibians almost in separate vertebrate classes. Their embryos have nevertheless been found to be susceptible to closely comparable surgical operations and to develop their primary patterns with indistinguishable propor-tions (Cooke, 1981,1982). Fertile eggs of Xenopus were obtained from spawn-ings induced by human chorionic gonadotrophin (Pregnyl-Organon Ltd.) injections, while those of Ambystoma were from both spontaneous and induced ovulations in the laboratory. Routine procedures for amphibian care, dejellying and demembranation of embryos, preparation for operations and subsequent development have been described (Cooke, 1972a, 1979a). Operations were per-formed in 66% Niu-Twitty saline brought to pH 7-3 with diluted HC1, and antibiotic (Gentamycin 25 μg/ml) was used in all solutions after demembrana-tion. Control and experimental embryos of both species were placed at a reduced temperature of 14 °C (normal ambient 20-21 °C) for a 24 h period beginning at the close of gastrulation some 2-6 h after operations. The latter procedure, now routine in the author’s laboratory, has been found to optimize graft-host tissue integration and interaction in a variety of recombination operations on young embryos. Operations were performed with electrolytically burnished tungsten needles and a hair loop, with embryos held in depressions in dental wax under the operating solution. Ionic strength was reduced to 15 % Niu-Twitty saline once continuity of the neurectodermal layer had been restored to avoid exo-gastrulation. Matched sets of experimental and control (sham-operated) syn-chronous sibling embryos (see Results (a) section) were kept in the same dishes and fixed and processed together.

Fixation for histological analysis was at the tailbud or early larval stage 30 in Xenopus (Nieuwkoop & Faber, 1957) and 33/4 in Ambystoma (Schrekenberg & Jacobson, 1975), attained 48 h and 6 days, respectively, after operations on gastrula stages. Fixation, embedding, precisely transverse serial sectioning (7 μm), staining with Feulgen/Light Green/Orange G, and the assay of size and proportions of mesodermal pattern by counting of nuclei in the four pattern parts in sample sections were all as in previous work. In the present study the rate of sampling varied, between different sibling sets, from every fifth section in the series to every twelfth (in the longer Ambystoma larva). Sampling always exten-ded between the anterior limits of the pronephroi, and the posterior limits of ectopic differentiations in experimentals (or equivalent position in controls) which lay just anterior to the proctodaeum. The estimate of a pattern’s propor-tions and its extent (mean total cells per transverse section) thus characteristic-ally involved some 25 sections and 5-8000 nuclei.

The histological schedule employed allows ready assignment of cells to the four structures that comprise the pattern. Notochord is highly characteristic with its vacuolated differentiation, circular profile and sheath; somite shows a distinc-tive cell alignment with bluish nuclei and large nucleoli; pronephros shows a tubular configuration with cuboidal epithelium. Lateral plate and blood-forming tissue is evident from its primitive state of differentiation and position in the lateral or ventral part of the mesodermal mantle (see Fig. 2, and the schematic representation of structures in transverse sections of Fig. 3). In previous work (Cooke, 1979a) estimating total cell numbers in mesoderms at earlier stages of morphogenesis, the present sample-counting technique was coupled with use of the Abercrombie factor for correction of errors in estimating absolute nuclear population from sections. This was not relevant to the present work, since no absolute estimate of cell numbers incorporated into each pattern part is made or indeed required in order to answer the questions being posed. The requirement is for a parameter, the percentage of the total nuclei observed to fall into each part in a standard section series, which varies sensitively with actual proportions in which tissue has been assigned to structures at initial pattern formation. At tailbud stages, configurations and sizes of cells differ between the pattern parts within embryos of each set, but care was taken only to compare, as sets, control and experimental siblings of equal axial length, as work in the other, horizontal plane of section has shown that mean A-P distance between nuclei in tissues is constant in such cases. The cell cycle in mesoderm between stages of pattern determination and the tailbud larva is slow relative to development, effectively zero in notochord and somite after determination, and not affected by operations altering mesodermal size and polarity relations (Cooke, 1979a, b). Repeat sam-pling (using different sections) of particular embryos in the present study revealed the error involved, in percentages to each pattern part and mean cells per section, to be smaller than the real differences encountered between suc-cessive sibling embryos analysed.

Fig. 2.

Transverse sections of experimental Xenopus embryos with midventral ectopic structures. Sections are from posterior trunk levels of stages 30-32. Symbols for territories of the host pattern as for Fig. 1. Ectopic somite (Ec Som). A small concentration of tissue with the appearance of lateral plate, but probably graft-derived, often flanks the ectopic somite. Its inclusion or exclusion from the total cell count for the ‘host’ pattern makes too little difference to affect the results in this species (see text). In the examples, there has been no development of ectopic CNS by graft ectoderm. Scale bar represents 1 mm approx.

Fig. 2.

Transverse sections of experimental Xenopus embryos with midventral ectopic structures. Sections are from posterior trunk levels of stages 30-32. Symbols for territories of the host pattern as for Fig. 1. Ectopic somite (Ec Som). A small concentration of tissue with the appearance of lateral plate, but probably graft-derived, often flanks the ectopic somite. Its inclusion or exclusion from the total cell count for the ‘host’ pattern makes too little difference to affect the results in this species (see text). In the examples, there has been no development of ectopic CNS by graft ectoderm. Scale bar represents 1 mm approx.

Fig. 3.

The heterochronic grafting operation. (A) The operation whereby a dor-solateral sector of neurula, including neural fold (NF), presumptive somite mesoderm (M) and the archenteron roof (AR) is grafted to a ventral site made by cutting a slot in the ectoderm back to the marginal zone in a mid-gastrula host. Light arrow marks the dorsal midline of the host, which is seen from the yolk-plug aspect. Heavy arrows show how cell layers of graft and host, particularly in the blastoporal zones of mesodermal recruitment (BP) where these layers join, are carefully matched up at the operation to ensure development of an integrated cylindrical sheet of mesoderm. (B) Cross sec-tional appearance a few hours after operations such as in (A). The midventral graft has integrated but is essentially pursuing its original, more advanced schedule of develop-ment as seen by the gathering of the mesoderm as in somite formation and, often, a co-ordinated neural area in the ectoderm. The host is at early neural fold (Xenopus st. 13 12 equivalent) stage. (C and D) Camera lucida outline cross sections of experimental Xenopus and Ambystoma tailbud larvae at the time of pattern assay. The cellular tex-tures of the four pattern parts, in which nuclei are sampled, are indicated schematic-ally (cell numbers not to scale). Symbols as in Figs 1 and 2. Note the asymmetrical development of ectopic structures (i.e. according to presumptive fate for the dor-solateral position of graft origin). The black arrowheads indicate limits of the mesoderm which amounts to a normal cross-sectional cell population for the sibling set, counting from the dorsal midline. It is seen that in Ambystoma there is characteris-tically an excess of lateral plate/blood tissue, presumably contributed by the graft. Inset shows the (more anterior) region where the major proportion of the pronephric tissue lies, illustrated for an Ambystoma larva. Scale bar = 1 mm approx.

Fig. 3.

The heterochronic grafting operation. (A) The operation whereby a dor-solateral sector of neurula, including neural fold (NF), presumptive somite mesoderm (M) and the archenteron roof (AR) is grafted to a ventral site made by cutting a slot in the ectoderm back to the marginal zone in a mid-gastrula host. Light arrow marks the dorsal midline of the host, which is seen from the yolk-plug aspect. Heavy arrows show how cell layers of graft and host, particularly in the blastoporal zones of mesodermal recruitment (BP) where these layers join, are carefully matched up at the operation to ensure development of an integrated cylindrical sheet of mesoderm. (B) Cross sec-tional appearance a few hours after operations such as in (A). The midventral graft has integrated but is essentially pursuing its original, more advanced schedule of develop-ment as seen by the gathering of the mesoderm as in somite formation and, often, a co-ordinated neural area in the ectoderm. The host is at early neural fold (Xenopus st. 13 12 equivalent) stage. (C and D) Camera lucida outline cross sections of experimental Xenopus and Ambystoma tailbud larvae at the time of pattern assay. The cellular tex-tures of the four pattern parts, in which nuclei are sampled, are indicated schematic-ally (cell numbers not to scale). Symbols as in Figs 1 and 2. Note the asymmetrical development of ectopic structures (i.e. according to presumptive fate for the dor-solateral position of graft origin). The black arrowheads indicate limits of the mesoderm which amounts to a normal cross-sectional cell population for the sibling set, counting from the dorsal midline. It is seen that in Ambystoma there is characteris-tically an excess of lateral plate/blood tissue, presumably contributed by the graft. Inset shows the (more anterior) region where the major proportion of the pronephric tissue lies, illustrated for an Ambystoma larva. Scale bar = 1 mm approx.

(a) The heterochronic grafting operation

Operations were designed to test the effect, upon the final proportions of patterns still to be determined in young gastrulae, of extra tissue grafted to one ‘edge’ of the system from determined pattern parts of more advanced mesoderms. The definition for the ‘determined’ state of embryonic material must always be operational, i.e. related to the precise treatment given to the cells before they differentiate (grafting vs. explantation in vitro, disaggregation or maceration vs. preservation of tissue structure, etc.). In the present work regions of mesoderm are said to be determined when they differentiate according to original fate, after heterotopic grafting as intact tissue from neurulae into gastrulae (heterochronic grafting). In both species used, dorsolateral pieces composed of invaginated presumptive somite from neurulae, but excluding the visible notochord, gave rise to longitudinal strips of ectopically differentiated somite after grafting to the midventral marginal zone of gastrulae. The operation and its typical results are shown in Fig. 3A-D. Such autonomous behaviour was reliably seen from donor pieces of Xenopus stages (or Ambystoma equivalent stages), grafted into host gastrulae of stages with an average host/graft difference of two stages. It is of interest that in other work (Forman & Slack, 1980), comparable material grafted to the belly region appeared not to express determination as somite, but the work in question involved smaller grafts, and hosts of comparable age to the donors.

Certain cases in the present series of operations showed unambiguous pronephric formations at one side of the midventral ectopic somite, in accord with the location of the pronephric territory just lateral to that of somite in the donor mesodermal mantle. Younger grafts, down to the host age, may be assimilated to give a relatively undisturbed host pattern (no ectopic structures), while any age disparity between tissues of more than two stages makes difficult the matching up and fusing of graft and host marginal zones of mesodermal recruitment (Keller, 1976; Cooke, 1979b). Such matching is required for success-ful integration of a strip of donor cells into the host mesodermal cylinder.

In Ambystoma, an enhancement of total mesodermal cell number in the host pattern (seen entirely as addition to the lateral and ventral blood island territory), as well as the presence of extra tissue as ectopic somite, was usual in successful cases. In Xenopus the total host pattern cell numbers were not significantly less than those of the controls. The slot cut in host gastrulae involved removal of many fewer mesodermal cells than were then added as part of the graft with its thickened (presomite) mesoderm, and indeed embryos cut as were the present hosts, but then allowed to heal and develop without graft addition, show normal cross-sectional cell numbers and pattern proportions (Cooke, in preparation). The observation that the ectopic differentiations were additional to a normal or even an unusually large cross sectional cell number in host patterns (Fig. 3C, D) was therefore not unexpected.

Sham operations, performed on siblings of the experimental hosts, consisted in removal of the midventral strip of marginal zone and its replacement with a strip of similar origin from another gastrula (-homotopic, synchronic grafting). This resulted in development of patterns indistinguishable from those of normal embryos. Precisely matched synchronous sets of experimental and their sham-operated control siblings were fixed and sectioned as tailbud larvae (see Materials and Methods), and the proportions (% nuclei per pattern part) and overall extents (mean nuclear number per T.S.) of their mesodermal body pat-terns were recorded as already described. A few larvae showing ectopically developing notochord at their graft positions (due to graft miscutting) were excluded from the analysis since the aim was to study the effects, upon host pattern, of tissues other than those developed from the organizer (Cooke, 1982). In the remainder, presence or absence of ectopic differentiated somite and pronephros was noted at each craniocaudal axial level throughout the region examined. In Xenopus, subtotal results for anterior, middle and posterior thirds of the total region (see Materials and Methods) were computed separately, while in Ambystoma the region was treated as anterior and posterior halves (see Fig. 4). In both species, columns of midventral ectopic structures tended to occupy about the posterior half of the axis between anterior host pronephros and host proctodaeum. Ectopic somite nuclear numbers might rise to 30-50 % of typical values for the host pattern at the levels of maximum ectopic developments.

Fig. 4.

External appearances of larvae. Control (sham operated) and experimental Xenopus (A, B) and Ambystoma (C, D) larvae are shown from the left lateral-ventral aspect. Ear vesicle (EV), pronephros and somite outlines (PN, SOM) and proctodaeum (BP; the original blastopore) are indicated on (A). Heavy dashed lines indicate cross sectional levels bounding the subregions of the total region assayed for pattern proportion in the two species (see Fig. 5and Table 3). Ectopic structures are revealed as a ventral ridge (with or without a neural formation) run-ning one third to one half way along the assayed region from the zone of original mesodermal recruitment at the proctodaeum, and sometimes ending anteriorly at an ear vesicle. Scale bar = 1 mm approx.

Fig. 4.

External appearances of larvae. Control (sham operated) and experimental Xenopus (A, B) and Ambystoma (C, D) larvae are shown from the left lateral-ventral aspect. Ear vesicle (EV), pronephros and somite outlines (PN, SOM) and proctodaeum (BP; the original blastopore) are indicated on (A). Heavy dashed lines indicate cross sectional levels bounding the subregions of the total region assayed for pattern proportion in the two species (see Fig. 5and Table 3). Ectopic structures are revealed as a ventral ridge (with or without a neural formation) run-ning one third to one half way along the assayed region from the zone of original mesodermal recruitment at the proctodaeum, and sometimes ending anteriorly at an ear vesicle. Scale bar = 1 mm approx.

(b) Cell numbers in pattern territories, and the proportions between them

A total of 20 experimental Xenopus embryos were analysed with 22 matched sibling controls, in five sets. In this species the patterns of the experimental host embryos, as a population, were indistinguishable in size (mean nuclear number per T.S.) from those of their sham-operated siblings. The ectopic differentiations could thus be treated as simple additions to one ‘border’ (the ventral border) of the region wherein pattern was to form, and the results expressed either as pattern proportions or as absolute nuclear numbers seen in pattern parts within the normal medial-to-lateral sequence of four territories in the host. Table 1 displays these results for all Xenopus matched sets so far studied. Fig. 5 shows, in histogram form with standard errors, the absolute nuclear numbers per section per embryo for notochord, somite and lateral plate/blood island in anterior, middle and posterior thirds of the total body region assayed. Only pooled results for three matched sets whose 16 control members showed statistically homogeneous proportions and cell numbers in T. S. have been used for the latter figure, and pronephros has been omitted because its very unequal distribution between the thirds makes its inclusion confusing and statistically invalid.

Table 1.

Proportions of nuclei encountered in each territory of the host pattern, in control (sham-operated) Xenopus embryos and in their siblings with midventral ectopic somite. Results over the total body region assayed (see text)

Proportions of nuclei encountered in each territory of the host pattern, in control (sham-operated) Xenopus embryos and in their siblings with midventral ectopic somite. Results over the total body region assayed (see text)
Proportions of nuclei encountered in each territory of the host pattern, in control (sham-operated) Xenopus embryos and in their siblings with midventral ectopic somite. Results over the total body region assayed (see text)
Table 3.

Mean total cell numbers per section in the host pattern, and numbers in somite and pronephros, in control and experimental Ambystoma embryos. Results for anterior and posterior halves of the body region assayed

Mean total cell numbers per section in the host pattern, and numbers in somite and pronephros, in control and experimental Ambystoma embryos. Results for anterior and posterior halves of the body region assayed
Mean total cell numbers per section in the host pattern, and numbers in somite and pronephros, in control and experimental Ambystoma embryos. Results for anterior and posterior halves of the body region assayed
Fig. 5.

Mean cell numbers per pattern part per cross section, in control and experi-mental Xenopus embryos. Each group of three columns gives the mean and standard error for cell number in a pattern part in anterior, middle, and posterior subregions of the axis assayed, going from left to right. Data from three sibling sets where 16 control members showed statistically homogeneous cell numbers in somite. * = different from the equivalent control value, P<·01. ** = different from the equivalent control value, P< 0·01.

Fig. 5.

Mean cell numbers per pattern part per cross section, in control and experi-mental Xenopus embryos. Each group of three columns gives the mean and standard error for cell number in a pattern part in anterior, middle, and posterior subregions of the axis assayed, going from left to right. Data from three sibling sets where 16 control members showed statistically homogeneous cell numbers in somite. * = different from the equivalent control value, P<·01. ** = different from the equivalent control value, P< 0·01.

Though variable, the diminution in proportions and absolute cell numbers devoted to somite, in host patterns flanked with ectopic somite, is highly statistic-ally significant (χ2 and Students ‘T’ tests on null hypothesis that controls and experimentáis are random samples from a single population with the overall mean value, within each set). The normal mesodermal cell numbers are incor-porated as a reciprocal increase in the relative extent of the lateral plate/blood territory. The phenomenon varies at the individual level from patterns with somite proportions within the normal range (-though never as high as the average control of their set) to those whose deficiency in somite cell numbers is obvious even before counting, in the regions occupied also by ectopic structures. It can be seen from Fig. 5 that the somite deficit in Xenopus patterns is not apparent anteriorly, but is concentrated relatively posteriorly opposite the great bulk of the ectopic somite. Lack of suppression of somite proportion in in-dividual cases was associated with small and/or primitively developed ectopic somite mass, whereas cross sections of normal total cell count opposite well-developed ectopic somite might contain only 50-60 % as many somite cells as their controls.

Variance for pronephric proportion is greater, in embryos generally, than that for somite (see also Cooke, 1981). Even so, a marginally significant deficit in pronephros is seen in the experimental Xenopus population as a whole. This modest and variable diminution may be related to the appearance, in many Xenopus ectopic formations, of small intermittent cell groups of pronephros-like morphology at one junction between somite and the host pattern’s lateral/ ventral edge. In five cases however (Table 1), host pronephric size was diminished altogether below the normal range, and this was associated with the only well-defined and anatomically coherent ectopic pronephric developments found in the series. Host notochord size appears unaffected by the operation.

In the Ambystoma material the situation is more complex, since the integrated host patterns in experimental embryos tend strongly to contain more cells in toto than do the control patterns. It is particularly in this species that the donor’s dorsolateral mesodermal strip contains more cells than are removed from the host’s mesoderm to accommodate it (see Results (a)). In the absence of a cell marker, and in view of earlier work that makes intercalary growth an unlikely response to these operations (Cooke, 1979a), it is reasonable to assume that the extra tissue flanking the ectopically differentiated structures is of donor origin and has been diverted from its original fate as somite or pronephros to become lateral plate/blood island integrated into the host pattern. We are left with the decision as to whether to assess the effects of ectopic differentiations in terms of the proportions (%) devoted to structures within the total cross-sectional popula-tion of the host pattern, or in terms of the actual nuclear numbers in these structures.

Proportions are known to be regulated towards constancy against overall size variation in normally patterned embryos (Cooke, 1981). Thus % somite might seem an appropriate statistic since we can deduce an expected % (on the hypothesis of no effect), as being that seen in the normal-sized control patterns in each matched set. This has been done for all Ambystoma sets so far studied, in Table 2. The results appear more dramatic than those for Xenopus in that no experimental embryos achieve somite proportions quite as high as any of their control siblings, while in many the somite territory is under-represented quite below the normal range of variation.

Table 2.

Proportions of nuclei encountered in each territory of the host pattern, in control (sham-operated) Ambystoma embryos and in their siblings with mid-ventral ectopic somite. Results over the total body region (see text)

Proportions of nuclei encountered in each territory of the host pattern, in control (sham-operated) Ambystoma embryos and in their siblings with mid-ventral ectopic somite. Results over the total body region (see text)
Proportions of nuclei encountered in each territory of the host pattern, in control (sham-operated) Ambystoma embryos and in their siblings with mid-ventral ectopic somite. Results over the total body region (see text)

A more conservative and reliable indication of specific inhibitory effects of ectopic structures, however, is the observation of actual reductions of cell num-bers in the appropriate parts of host patterns. Mean nuclear numbers per cross section in the pattern parts are given, for anterior and posterior halves of the region scored, in Table 3. Absolute reduction in somite is seen in several em-bryos but there are also embryos where somite cell numbers are indistinguishable from normal, though embedded in a pattern whose overall size has been enhan-ced as described. It is impossible to be sure whether such embryos are evidence for inhibitory effects or not, since we do not really know what their ‘expected’ somite cell numbers would be. As with the Xenopus series of material, however, the most striking cases of somite under-representation are correlated with the greatest masses of ectopically differentiating somite, while marginal cases are associated with the appearance that much of the implant was unable to maintain its developmental tendency and was instead assimilated into the host pattern as lateral plate/blood forming tissue.

In both species, the ectopic midventral somite formations of experimental embryos are only sometimes accompanied by neural differentiation of the over-lying ectoderm and its formation into a tube. Several examples of pronounced host somite suppression were associated with absence of any ectopic neural formations, but such suppression was always accompanied by a well-formed ectopic somite mass. Fig. 2 shows examples of such experimental embryos for Xenopus.

The observation of a well-formed ectopic pronephros in Ambystoma is associated with a reduction in size of the host’s in situ pronephros below the normal range of variation in two cases, but not in another two. Another feature of this more restricted set of results, as compared with the series from Xenopus, is that the diminution in somite cell numbers can be present throughout the axial region examined, including anteriormost (earliest differentiated) regions, as well as opposite the somite from implants.

From our current understanding of the normal fate map of Xenopus develop-ment (Keller, 1976) it appears that tail somites could derive from precursor cells situated in the midventral marginal zone of the young gastrula. On this view, the contention that the present phenomena are true effects on pattern formation is subject to the conceivable criticism that the grafts have mechanically prevented the migration of specific cell populations into the dorsal axis in the more posterior regions assayed, and trapped these cells instead to contribute to ectopic ventral somite. In matched sets of Xenopus and of Axolotl, where several experimental members showed marked somite diminution, the total cell populations were compared in the tailbud mesoderms, posterior to the proctodaeum and thus to the region occupied by ectopic structures. This population is in fact very largely of somite and presomite cells, and was found to be indistinguishable in experimentáis and controls. Tail regions of experimental embryos show no ten-dency to be deficient to outer inspection. It is thus highly unlikely that the effect of midventral fragments is caused by any specific blockade to gastrulation move-ments, or to any selective aggregation of host presumptive somite cells into the graft masses.

The two lateral parts of the pattern being assayed, namely pronephros and lateral plate, continue with cell division at (possibly specific) cycle times of some 10-15 h, after their initial histogenesis. Their slow growth between foundation and assay of the pattern could therefore complicate the interpretation. But notochord is without cell division and, up to the tailbud stages here assayed, the myotomal bodies of somites which alone were included in the counts show a very low mitotic rate in urodeles while in Xenopus a mitotic figure has never been observed. Thus for somite, the principal pattern part being studied in the presence of grafted homologous tissue elsewhere in the embryo, the nuclear numbers scored reflect directly the size of the cell population originally set aside in pattern formation. Furthermore in the Xenopus series the overall cell popula-tion of hosts’ patterns is of normal size at the time of assay, with the somite deficit balanced by an excess of lateral plate. The strong presumption is thus that there has been an alteration in the initial proportions in which the mesodermal cell population has been allocated to found the pattern parts.

It is hard to avoid concluding from these results that tissue which is relatively advanced, in its determination as particular territories of the embryonic pattern, emits signals that can alter the balance of determinations elsewhere in the em-bryo. Specifically, it appears that such signals from implanted tissue can cause diversion of younger cells from the homologous pathways of determination into those proper to more lateral regions of the embryo, since the positions of certain frontiers between pattern parts in the original mesodermal mantle are shifted medially. Pattern determination probably proceeds from medial to lateral across some considerable time, whose precise limits in relation to the progress of gastru-lation are unknown, while the specific inhibitory signals whose existence is in-dicated by these results may each have a short time course in relation to develop-ment. Thus the inconstancy and variability of the observed effect on somite is not surprising, in view of the variability in precise graft and host ages and age dif-ferences, among the operations performed to date. Although few in number, the examples of apparent diminution of pronephric size by prior ectopic pronephros differentiation are striking, especially as the ectopic territory lies well posterior to the site of the host organ as well as in a ventral position. We do not understand enough about differences between gastrulation and the timing of determination, in Xenopus and Ambystoma, to speculate on why in the latter case even the earliest differentiating somite can be diminished by ectopic differentiations which end up posteriorly, whereas in the former, somite inhibitory effects seem confined to particular axial levels.

The theory that positional information underlies pattern control, though abs-tract, is precise and makes certain potentially testable postulates. Some of these concern the behaviour of pattern proportions at abnormally small field sizes and under conditions of artificial bipolarity, in relation to the expected performances of plausible physicochemical systems that might underlie the signal profiles which are interpreted to give the patterns. Work of this type, whose results were problematical for almost all gradient models, has been presented elsewhere (Cooke, 1981, 1982). A further, strong postulate of the positional information idea, as stated in the introduction, is that no specific interactions between emerg-ing pattern parts are to be expected in the control of their proportions. Propor-tioning is achieved instead from interpretation of local signal levels along the simple but regulated gradient profile between boundary values, so that frontiers between cell groups progressing to different determinations are independently positioned. The ectopic addition of advanced tissue belonging to some element of the normal pattern should be without effect on the positions of such frontiers elsewhere in the embryo, and thus should not affect the relative sizes of elements in the host pattern.

The material of the grafts used in the present work does not represent the organizing boundary for the pattern, since no notochord differentiated, and the visible territory for this structure was in fact left in the donors. The boundary is in fact represented by the dorsal blastoporsal lip, essentially the presumptive notochord territory, transplanted in previous work (Spemann & Mangold, see Spemann, 1938; Cooke, 1972a,b, 1979). The developmental stage of the present grafts is such that they differentiate relatively autonomously and do not involve the surrounding host material in a new edition of pattern centred on themselves. They do affect the proportions of homologous pattern parts elsewhere, however, in contravention of the expectations from the positional information hypothesis. Together with the previous work, these results make it unlikely that positional information, sensu stricto, underlies control of the mediolateral dimension of the vertebrate body plan. There follows the outline of a model mechanism which would accommodate the data. It is related to previous ideas of Rose (1957,1970; see also Braverman, 1961) and, recently, of Meinhardt & Gierer (1980), though differing in particulars from these. It is also perhaps a reversion towards the original conception by Child (1941) of physiological gradients and dominance hierarchies controlling morphogenesis.

The region of the embryo that will become the dorsal blastoporal lip and the anterior medial mesoderm and endoderm is certainly an organizing boundary, which controls the orientation with which the mesodermal body plan develops.

It is the only region to give any evidence of autonomous boundary properties after grafting operations in these embryos, being the organizer of Spemann (review, Spemann, 1938). In the present model the organizer is assumed to control only the overall field polarity, expressed as graded rates of physiological progress towards determination by mesodermal cells, smoothly ranked from medial or dorsal (i.e. in contact with the organizer) to presumptive lateral or ventral (i.e. far from the organizer) extremes of the tissue. Such control from a local region could be via a morphogen signal gradient declining smoothly with distance from the boundary source, and setting local rates of cell development by its level. Alternatively it could be by a spreading wave-like process of activa-tion that was necessary to begin some schedule of cellular maturation. Such an ‘activation wave’ would produce a subsequent wavefront with respect to cells’ arrival at particular states of maturity, passing coherently across the mesoderm. In the model now to be proposed, neither sort of control process would have any further exact informational function, such as would have been required of a regulating gradient used as true positional information. Either a primitive and plausible diffusion-controlled mechanism or one involving spreading autocata-lytic activation would be adequate to ensure the direction and continuity of such a wavefront of development, and thus the reliable temporal order in which onset of determination occurs within the tissue space. Normal spatial ordering and proportional extent of determined territories are then assumed to result from (a) the logic governing access to various determined states by cells and (b) the history of the system as development proceeds. All cells that can develop fast enough, mature towards one particular state, i.e. notochord determination. This continues until a diffusible signal, produced specifically by cells that have entered that particular committed state, builds up to a systemic level that prohibits subsequently developing cells (i.e. those ‘slower’ because more lateral in position) from developing further along the same pathway. The logic is such that these cells are diverted towards a next available state, i.e. somite determina-tion. But somite as a committed cell state involves production of a further diffus-ible systemic signal, which finally diverts even less-advanced cells towards a third available state (pronephros), and so on. Each specific signal is sufficiently ‘diffus-ible’ within the mesoderm for its level to act effectively as a sensor for the proportion of the individual embryo that has become devoted to the cell type that produces it. The mediolateral wavefront of determination is thus serially diverted in character during its passage across the tissue when particular proportions have been reached for each territory.

The scaling down of pattern in small mesoderms and in those being simul-taneously invaded by wavefronts from two dorsal organizing boundaries is an automatic consequence of the mechanism, because of diversion of the character of the progressing determination whenever a particular overall proportion has been achieved, in the whole mesoderm, of cells producing each successive in-hibitor. The negative feedback on sizes of specific territories from prior ectopic differentiations, reported in this paper, would be due to a priming or pre-empting effect when the system encounters appreciable concentrations of a specific signal already supplied from elsewhere. The creation of the frontier between pattern parts, corresponding to diversion of determination from this particular character to the one lateral to it, will occur precociously and thus be shifted medially to diminish the relative size of the pattern part.

The inset in Fig. 6 shows a typical gradient profile (a) such as could be produced by reaction diffusion or source-diffusion controlled at one boundary in a tissue, and its family of variants (dotted curves b and c) such as might result from chance fluctuations or imperfections in its regulatory mechanisms. Signal levels are shown that might correspond in a positional information model to interpretation thresholds for two ‘internal’ territories, such as the somite and pronephros studied here. It can be seen how such a model predicts that a significant com-ponent of the irreducible variance in pattern proportions, among individual em-bryos, should take the form of positive correlations (or conceivably, for curves c, inverse correlations) between the proportions devoted to these internal ter-ritories, with lateral plate always occupying the ‘left-over’ tissue. The scatter diagram of Fig. 6 represents the deviations from mean proportion for somite plotted against deviations from mean proportion for pronephros, in the patterns of all normal-sized Xenopus and Ambystoma mesoderms ever assayed in this laboratory. Deviations are estimated on the means within all control embryos of each species and within all embryos where pattern had been influenced by ectopic differentiation. A complete lack of correlation is seen between the intrinsic varia-tions in proportion of each element, suggesting that in each embryo proportion assigned to somite is independent of, and has no influence upon, proportion assigned to pronephros. Such an observation is consistent with the idea that, as in the present serial diversion model, each pattern element is controlled by an independent size-assessment event in the mesoderm (via a diffusible inhibitor and threshold arrangement) that is subject to its own performance variation.

Fig. 6.

The independence of variations in proportion between somite and pron-ephros. Within each subset of matched synchronous siblings (i.e. all controls or all those with patterns experimentally influenced by heterochronic grafting), the means of % nuclei scored in somite and in pronephros were computed, and the individuals inserted in the scatter plot according to the relative deviations from these means (expressed as decimal fractions of the mean % value) seen in their own patterns. Black spots; control (sham-operated or unoperated) embryos. Open circles; host patterns of embryos which also have ectopic structures. The resulting scatter shows no significant correlation, positive or negative, between embryos deviations from their sample mean for these two proportions. Ordinate; relative deviation in pronephros proportion. Abscissa; relative deviation in somite proportion.

Insert represents (a) a particular ‘ideal’ or ‘normal’ gradient profile for a signal, controlled from one end of a system, which might be interpreted to give pattern, and (b), (c) different families of variant profiles (shown in extreme form for clarity) that might result from imperfect replicability of the gradient mechanism among in-dividuals. Curves of the family (b) differ from the ‘ideal’ in steepness, while those of family (c) differ in degree of inflection, or non-linearity. The marks on the ordinate represent signal levels which could be taken to establish frontiers of two successive ‘internal’ pattern parts such as somite and pronephros in the present case. Reference to these marks and to the families of curves shows that on a positional gradient hypothesis for overall pattern control, a component of correlation (positive or con-ceivably inverse) would be expected, between deviations from the mean in propor-tions devoted to these pattern elements in individual embryos.

Fig. 6.

The independence of variations in proportion between somite and pron-ephros. Within each subset of matched synchronous siblings (i.e. all controls or all those with patterns experimentally influenced by heterochronic grafting), the means of % nuclei scored in somite and in pronephros were computed, and the individuals inserted in the scatter plot according to the relative deviations from these means (expressed as decimal fractions of the mean % value) seen in their own patterns. Black spots; control (sham-operated or unoperated) embryos. Open circles; host patterns of embryos which also have ectopic structures. The resulting scatter shows no significant correlation, positive or negative, between embryos deviations from their sample mean for these two proportions. Ordinate; relative deviation in pronephros proportion. Abscissa; relative deviation in somite proportion.

Insert represents (a) a particular ‘ideal’ or ‘normal’ gradient profile for a signal, controlled from one end of a system, which might be interpreted to give pattern, and (b), (c) different families of variant profiles (shown in extreme form for clarity) that might result from imperfect replicability of the gradient mechanism among in-dividuals. Curves of the family (b) differ from the ‘ideal’ in steepness, while those of family (c) differ in degree of inflection, or non-linearity. The marks on the ordinate represent signal levels which could be taken to establish frontiers of two successive ‘internal’ pattern parts such as somite and pronephros in the present case. Reference to these marks and to the families of curves shows that on a positional gradient hypothesis for overall pattern control, a component of correlation (positive or con-ceivably inverse) would be expected, between deviations from the mean in propor-tions devoted to these pattern elements in individual embryos.

There is a tempting analogy between the concept of serial diversion signals controlling initial pattern proportions, and that of ‘chalones’ (Bullough, 1967), the systemic feedback signals known to regulate the functional size of various differentiated tissue systems in the later body. But at the time of pattern control the embryo has no vascular system. Furthermore, what have been referred to here as territories of a primitive pattern are not states of differentiation in the histologist’s or physiologist’s sense, but states of potency restriction as to the parts of the body to which cells’ descendants can contribute. The postulated signals are thus more allied to traditional morphogens than to hormones or chalones. One demanding feature of the present model is the rapidity with which specific signals appear to be able to spread through tissue, in relation to past calculations about diffusion-controlled changes (see Crick, 1970). Empirical study of the effective diffusion rates of marked molecules of known charge and size through embryonic cell sheets is urgently required.

I thank Jonathan Slack and Malcolm Maden for batches of Ambystoma embryos, and am indebted to John Webber, Malcolm Maden and Fiona Harvey for critical discussions of the work, June Colville for skilled histology and Fay Morris for typing the manuscript.

Bullough
,
W. S.
(
1967
).
The Evolution of Differentiation
.
London
:
Academic Press
.
Bra Verman
,
M. H.
(
1961
).
Regional specificity of inhibition within chick brain
.
J. Morphol
.
108
,
163
285
.
Child
,
C. M.
(
1941
).
Patterns and Problems of Development
.
Chicago
:
University of Chicago Press
.
Cooke
,
J.
(
1972a
).
Properties of the primary organisation field in the embryo of Xenopus laevis. I. Autonomy of cell behaviour at the site of initial organiser formation
.
J. Embryol. exp. Morph
.
28
,
13
26
.
Cooke
,
J.
(
1972b
).
Properties of the primary organisation field in the embryo of Xenopus laevis. II. Positional information for axial organisation in embryos with two head organisers
.
J. Embryol. exp. Morph
.
28
,
27
46
.
Cooke
,
J.
(
1975
).
Control of somite number during development of a vertebrate, Xenopus laevis
.
Nature
254
,
196
199
.
Cooke
,
J.
(
1979a
).
Cell number in relation to primary pattern formation in the embryo of Xenopus laevis. I. The cell cycle during new pattern formation in response to implanted organisers
.
J. Embryol. exp. Morph
.
51
,
165
182
.
Cooke
,
J.
(
1979b
).
Cell number in relation to primary pattern formation in the embryo of Xenopus laevis. II. Sequential cell recruitment, and control of the cell cycle during mesoderm formation
.
J. Embryol. exp. Morph
.
53
,
269
289
.
Cooke
,
J.
(
1981
).
Scale of body pattern adjusts to available cell number in amphibian embryos
.
Nature
210
,
775
778
.
Cooke
,
J.
(
1982
).
The relation between scale and the completeness of pattern in vertebrate embryogenesis; models and experiments
.
Amer. Zool
.
22
,
91
104
.
Crick
,
F. H. C.
(
1970
).
Diffusion in embryogenesis
.
Nature
225
,
420
422
. .
Forman
,
L.
&
Slack
,
J. M. W.
(
1980
).
Determination and cellular commitment in the embryonic amphibian mesoderm
.
Nature
286
,
482
484
.
Keller
,
R. E.
(
1976
).
Dye mapping of the gastrula and neurula of Xenopus laevis. II. Prospective areas and morphogenetic movements of the deep layer
.
Devi Biol
.
51
,
118
137
.
Laufer
,
J.
,
Bazzicalupo
,
P.
&
Wood
,
W. B.
(
1980
).
Segregation of developmental potential in early embryos of Caenorhabditis elegans
.
Cell
18
,
569
577
.
Meinhardt
,
H.
&
Gierer
,
A.
(
1980
).
Generation and regeneration of sequence of structures during morphogenesis
.
J. theoret. Biol
.
85
,
429
450
.
Nieuwkoop
,
P. D.
&
Faber
,
J.
(
1957
).
Normal Table o/Xenopus laevis
.
(Daudiri)
.
Amsterdam
:
Elsevier, North Holland
.
Rose
,
S. M.
(
1957
).
Cellular interaction during differentiation
.
Biol. Rev
.
32
,
351
382
.
Rose
,
S. M.
(
1970
).
Differentiation during regeneration caused by migration of repressors in bioelectric fields
.
Amer. Zool
.
10
,
91
99
.
Schrekenberg
,
G. M.
&
Jacobson
,
A. G.
(
1975
).
Normal stages of development of the Axolotl, Ambystoma mexicanum
.
Devi Biol
.
42
,
391
400
.
Spemann
,
H.
(
1938
).
Embryonic Development and Induction. Reprinted 1967
,
Hafner, N.Y
.:
Yale University Press
.
Wolpert
,
L.
(
1971
).
Positional information and pattern formation
.
Current Topics in Devi Biol
.
6
,
183
224
.