In this paper, we study the mechanism by which a Xenopus cell line-derived mesoderm-inducing factor (MIF) might establish the spatial-pattern of cellular differentiation in the mesoderm. The effects of the factor on competent animal pole tissue are consistent with it being identical to the natural mesoderm-inducing factor. The signal can only act on those membrane domains of the animal pole that face the blastocoel, but it can be stably recorded there, such that axial mesoderm is formed, after 15 min exposure or less. This exposure can end some hours, or several cell cycles, before the onset of RNA synthesis yet nevertheless be fully effective, although competence to respond also extends well after the onset of transcription.

Exposure of the entire blastocoel lining of intact embryos to MIF causes a synchronous and sudden transformation of the behaviour and adhesive properties of all inner animal cap cells. This transformation mimics and is contemporaneous with the involution behaviour of normal mesoderm in the early gastrula marginal zone. Although high concentrations of MIF totally disorganize gastrulation, lower concentrations permit gastrulation to proceed. However, the pattern of mesoderm in these embryos is disrupted and ectopic mesoderm is formed around the blastocoel remnant.

When MIF is injected directly into blastomeres, rather than into the blastocoel, it has no effect. This suggests that the molecule is secreted from source cells and affects target cells through an extracellular receptor.

Finally, we show that small pieces of animal pole tissue recently exposed to MIF go on to produce morphogenetic signals perhaps distinct from MIF. We discuss the role of these signals in establishing and modifying the spatial pattern of cellular differentiation in the mesoderm of Xenopus.

Gurdon (1987), reviewing the role of inductive interactions in early development, distinguishes between heterogenous (foreign) and homogenous sources of inducing factors. A factor that induces mesoderm from competent amphibian ectoderm has recently been described, which, while not strictly homogenous in coming from the natural anatomical locality and time in the life cycle for this signal, is at least a gene product of the organism in which it is being studied (Smith, 1987a). The hope that this factor is identical, or is closely related, to a natural agent of mesoderm induction is based on two alternative ideas. First, culture-adapted cells derived from late stages of development might express embryo-specific proteins as accidental ‘breakthrough’ products. Alternatively, they may synthesize substances having normal roles in the adult body that were also used at earlier stages of development as intercellular signals (see Slack, Darlington, Heath & Godsave, 1987).

The present mesoderm-inducing factor (MIF) is secreted by the Xenopus XTC cell line (Pudney, Varma & Leake, 1973). It leads to differentiation of mesodermal structures characterizing ‘dorsal’ or axial parts of the normal body (Smith, 1987a). It does this via a pathway that leads responding tissue to participate in just those sequences of mechanical and adhesive activities that distinguish the precursors of these mesodermal regions in vivo (Symes & Smith, 1987). There is currently no reason, however, to think of the factor as a direct stimulus to the determination of any one mesodermal cell type. There is indeed no reason to think of the natural process that specifies tissue as mesodermal (Nieuwkoop, 1969) as simultaneously causing responding cells to differentiate as any of the cell types ultimately produced. Elaboration of pattern within the responding tissue probably requires further sequences of signalling interactions coming under the general concept of homeogenetic induction. If, as suggested for some cases studied in amphibia (Kaneda, 1981; Kurihara & Sasaki, 1981; Kaneda & Suzuki, 1983), the propagation of an induced state is accompanied by alteration in the cell types specified, such a process could be part of the mechanism for establishing positional information (Wolpert, 1969). The most elegant and incisive recent experiments on the conditions and time course required for primary mesodermal induction have, nevertheless, utilized the onset of muscle-specific transcription as a molecular marker for successful interaction (Gurdon, Fairman, Mohun & Brennan, 1985; Gurdon & Fairman, 1986; Warner & Gurdon, 1987).

Here, we report the results of several different kinds of experiment, on whole embryos and on explants, using partially purified preparations of MIF, and also poly (A)+ RNA from MIF-producing cells (see also Woodland & Jones, 1987). Together with the studies of cell movement and cell cycle kinetics in explanted responding tissue (Symes & Smith, 1987), our results make it seem increasingly likely that the XTC cell MIF is, or is closely related to, a natural initiator of mesoderm formation. They are also helping us to begin the difficult task of understanding the process whereby the natural mesoderm becomes organized or patterned, at or soon after its inception.

Embryos

Embryos of Xenopus laevis were obtained by artificial fertilization. They were chemically dejellied, washed and transferred to 10% normal amphibian medium (NAM: Slack, 1984) until required. They were staged according to Nieuwkoop & Faber (1967).

Partially purified mesoderm-inducing factor

Serum-free conditioned medium from the XTC cell line was obtained as described by Smith (1987a), except that the cells were grown on glass roller bottles. The medium was heated to 100°C (Smith, 1987a), concentrated tenfold by ultrafiltration and exchanged into 20 mM-Tris pH 8·0. Mesoderm-inducing activity was partially purified by ion exchange chromatography on DEAE-Sepharose, where activity eluted at approximately 0·5M-NaCl, followed by hydrophobic interaction chromatography on phenyl-Sepharose, where activity eluted with approximately 30–40% ethylene glycol (Smith & Yaqoob. unpublished data). This material was active at concentrations of 0·1–0·2% in an assay in which internal cells of the blastocoel roof are exposed to mesoderm-inducing activity at stage 8 and subsequently tested for mesodermal differentiation (Smith. 1987a). We define one unit of mesoderm-inducing activity as the minimum quantity which must be present in 1 ml medium for induction to occur. Our partially purified preparation of MIF therefore contained 500–1000 units ml−1, representing about 7000units mg−1 protein.

0·5 ml of this partially purified preparation was dialysed for 3 h against 67 % NAM adjusted to pH 8-0, after addition of bovine serum albumin (BSA) to 0·5 mg ml−1 to reduce nonspecific loss of protein. The final dialysate was active at 0·3–0·5% (250unitsml−1) in the above assay. Control dialysates were made of the dialysis medium and of 30 times concentrated XL-conditioned medium, previously adjusted to 30% ethylene glycol and 0·5mgml−1 BSA. XL cells (Anizet, Huwe, Pays & Picard, 1981) produce little mesoderm-inducing activity (Smith, 1987a). Dialysates were microfuged and stored at –70°C as 50μl aliquots. Each experiment was carried out with dialysate that had been kept for not more than 72 h at 4°C after thawing. Intracellular and blastocoelic injections were performed with micro-pipettes of appropriate tip size, using pulses of air at 250gcm−2 or by means of a completely oil-filled micrometer-syringe apparatus. Injection was performed under 10% NAM containing 5% Ficoll (Sigma) to clamp the vitelline membrane to the cell surface, and injected embryos were kept in Ficoll for 30 min before being transferred to 10% NAM.

RNA isolation

RNA was extracted from XTC or XL cells as described by Krumlauf, Holland, McVey & Hogan (1987). Poly (A)+ RNA was isolated by oligo(dT)-cellulose chromatography as described by Maniatis, Fritsch & Sambrook (1982) and quantified according to its A260. Proteins of up to 100000 molecular mass were seen after in vitro translation of our RNA preparations (data not shown).

For microinjection, poly (A)+ RNA was dissolved in sterile distilled water at 1 mg ml−1 and micropipettes were heated overnight at 180°C to destroy RNase activity.

Blastula animal cap explants and combinations

Animal cap regions were excised in 67% NAM. pH 7·3. According to the experiment, they were then cultured in the same medium, or in heated XTC-conditioned medium diluted with 67 % NAM to concentrations known to produce strong ‘dorsal’ mesodermal induction in stage-8 animal caps (see Smith, 1987a). In different experiments, animal caps were excised between the 128 to 256-cell stage and late stage 8. It was found that a larger cell sheet than represented by the ‘60° solid angle around the animal pole’ of Dale. Smith & Slack (1985) can be cultured without spontaneous production of mesoderm, if care is taken to exclude the thickened zone of white opalescent material appearing at the margin from stage 8. In some experiments, freshly excised animal caps were fused with others, or with matched pieces of stage-11 gastrula ectoderm, in ‘sandwich’ combinations that were sealed by the weight of one donor embryo in a shallow agar well for 10min. These then rounded up to become a sealed, hamburger-shaped wall of tissue. In most culture experiments the ionic strength was lowered to 20% NAM about 1h after excision, as this prevented excessive contraction of unsealed explants with consequent extrusion of inner tissue, perhaps the equivalent of exogastrulation in whole embryos.

Lineage labelling

Fertilized eggs were injected with 20–30 nl of 100 mg ml−1 rhodamine–lysine–dextran (RLDx: see Gimlich & Cooke, 1983; Gimlich & Braun, 1985). The resulting embryos were uniformly labelled with the lineage marker (see Smith & Slack, 1983; Dale et al. 1985).

Indirect immunofluorescence for Xenopus muscle

Muscle differentiation was detected on PEDS wax-embedded sections by indirect immunofluorescence with the monoclonal antibody 12/101 (Kintner & Brockes, 1984). The technique was identical to that of Dale et al. (1985) and Smith (1987a) except that material was fixed in 4 % fresh formaldehyde in PBS to preserve RLDx lineage labelling.

Preparation of stained sectioned material

Gastrulae were fixed overnight in 2% paraformaldehyde, 2% glutaraldehyde in 0·08 M-cacodylate buffer, pH 7·3, leaving the vitelline membrane intact. They were then rinsed in buffer alone for 2h, and slowly dehydrated to 95 % ethanol over 3 or 4h after puncture of a small hole with a tungsten needle to minimize distortions during dehydration. The embryos were embedded in Historesin (LKB) via a preinfiltration in resin:alcohol (50:50). 2μm sagittal sections were dried onto clean glass slides, stained with 0·1 % toluidine blue in 1 % borax at 70°C for about 2min, rinsed and dried for mounting.

Embryos and explants at axial larval stages were fixed and embedded in 58°C m.p. wax for Feulgen staining with light green/orange G counterstain (Cooke, 1979, 1981).

Examination of stage-9 to -12 embryos by fixation-dissection

Embryos to be examined by dissection were demembranated and a small hole was punctured in the blastocoel roof. They were then placed in a fixative of 1 % K2Cr2O7, 2% glacial acetic acid in water and examined against black wax in oblique lighting under a Wild dissecting stereomicroscope. After 5–10min. the embryos could be cracked in desired planes by sharpened tungsten needles, and the presence of tissue layers and cell social behaviour could be examined on a gross scale. Ectoderm, postinvoluted mesoderm with the inner ‘blastoporal lip’ (Keller, 1975, 1976) and endoderm could easily be identified under these conditions.

The effect of MIF is stably recorded within minutes of exposure at the ‘inner’ surface of blastula cells

Freshly excised blastula animal caps were exposed to diluted XTC-conditioned medium followed after various times by thorough washing and culture in 67 % NAM. In most experiments, the Ca2+ and Mg2+ concentrations in the initial inducing medium were one third the normal, to delay contraction and adhesion of the explant cells so that washing would be effective. Explants exposed to MIF for more than 10 min showed by their convergent extension movements and subsequent differentiation that mesoderm had developed, as described by Smith (1987a) and Symes & Smith (1987). In three such experiments, the test animal caps were excised at stage 7– (128-cell stage). Both faces of their single layer of cells were fully exposed to the wash that followed 20 min of culture in MIF. 17 out of 21 cases showed a strongly induced appearance, and histology confirmed that notochord and massive, organized somite tissue had resulted (Fig. 1A,B). This suggests that MIF is stably bound at the cell surface, or causes the activation of a stable second messenger, within minutes. This initial phase of recording the signal is unlikely to depend upon RNA synthesis since transcription is barely detectable before the mid-blastula transition (Newport & Kirschner, 1982), which occurs about 2h after this short exposure to MIF.

Fig. 1.

Differentiation in explanted animal caps after various conditions of early exposure to MIF. (A,B) Sections transverse and longitudinal to axial structure in explants exposed to MIF for 20 min immediately after excision from the 128-cell-stage embryo. All explants were extensively washed after exposure and cultured in 67 % NAM until controls reached stage 35. (C) Differentiation indistinguishable from that of controls (67 % NAM only) and (D) differentiation corresponding to the ‘weakest’ grade of mesoderm induction, as observed after culture in limiting dilutions of MIF. These appearances follow continuous culture in the same MIF preparation that gives rise to A and B but beginning 30 min after dissection from the donor, when little or no original ‘inner’ blastomere membrane was exposed to the medium, nc, notochord; s, somite; ab, abnormal, wrinkled epidermis; ep, properly stretched, bilayered epidermis over a fluid-inflated cavity, not seen in complete absence of induction. Scale bar, 300μm.

Fig. 1.

Differentiation in explanted animal caps after various conditions of early exposure to MIF. (A,B) Sections transverse and longitudinal to axial structure in explants exposed to MIF for 20 min immediately after excision from the 128-cell-stage embryo. All explants were extensively washed after exposure and cultured in 67 % NAM until controls reached stage 35. (C) Differentiation indistinguishable from that of controls (67 % NAM only) and (D) differentiation corresponding to the ‘weakest’ grade of mesoderm induction, as observed after culture in limiting dilutions of MIF. These appearances follow continuous culture in the same MIF preparation that gives rise to A and B but beginning 30 min after dissection from the donor, when little or no original ‘inner’ blastomere membrane was exposed to the medium, nc, notochord; s, somite; ab, abnormal, wrinkled epidermis; ep, properly stretched, bilayered epidermis over a fluid-inflated cavity, not seen in complete absence of induction. Scale bar, 300μm.

The single layer of cells in the stage 7– animal cap is polarized, with the outer surface being derived from the egg membrane and the lateral and basal membranes being newly assembled (see Gerhart, 1980). We investigated which surface(s) could respond to MIF by fusing two animal pole regions such that after 30 min healing they formed a sphere, of which the external surface was composed entirely of the outer domain. On electron microscopical and electrophysical grounds (Kalt, 1971; Slack & Warner, 1973; Regen & Steinhardt, 1986) the inner cell membrane domains should be inaccessible to macro-molecules introduced into the culture medium.

Ten double explants of this type were transferred to XTC-conditioned medium when healing was complete. None formed mesoderm, all differentiating as ‘atypical epidermis’. As a control, at the same time the fused explants were transferred to MIF, nine additional double explants were constructed from sibling embryos in the presence of MIF. All formed mesoderm. We conclude that only the inner membrane domain is capable of mediating the response to MIF.

This observation makes it possible to expose different proportions of cells in an explant to MIF. When animal pole regions are dissected from embryos, over the next hour or so they tend to curl into a solid ball, in which the original outer surface encloses those parts of the cells which would have faced the blastocoel (see Symes & Smith, 1987). In three experiments, therefore, stage 7+ explants were prepared in 67 % NAM and allowed to heal progressively into spheres. At intervals, samples of five to seven of these were transferred to highly inducing concentrations of XTC-conditioned medium and allowed to develop to control stage 40. In all 38 cases, explants first experiencing MIF when considerable ‘inner’ cell surface was still visible went on to show convergent extension movements and the appearance of induction. However, as healing time increased, the intensity of response decreased. When only a small focus of inner cell membrane was apparent, at 30–40min after excision, only 3 out of 18 explants showed any kind of induction (Fig. 1C,D). This was a ‘weak’ or ‘ventral-type’ induction, resembling those obtained from newly dissected animal pole explants in response to fibroblast growth factor (Slack et al. 1987) or low concentrations of XTC-conditioned medium (Symes & Smith, unpublished observations). In each experiment, a control group of stage-9 explants, made from synchronous embryos of the same egg batch, was exposed to MIF at the same time as the final healed explants. All 15 were strongly induced, confirming that the cells were still competent to respond to MIF.

Injection of MIF into the Xenopus blastocoel disorganizes gastrulation and alters the subsequent body pattern

In the remaining experiments (other than those involving poly(A)+ RNA), partially purified MIF has been microinjected into the blastocoels of embryos, either to study its effect on pattern formation in the mesoderm or to provide a source of animal pole tissue which, unable to heal into a sphere, has received homogenous exposure to MIF. Control dialysates consisting of saline containing only 0·5 mg ml−1 BSA, or of concentrated XL-conditioned medium, are without effect when used at a volume of 200 nl that will effectively irrigate the cavity and assume a uniform concentration to the lining cells. The effective concentration of MIF has thus been varied by delivering this constant volume diluted as necessary with XL-derived dialysate, to give doses of 0·003 to 0-1 units per embryo (see Materials and methods for details and origin of units).

Irrespective of dose or stage of injection (from stage 7− to 9+; 5h to 30min before gastrulation) MIF has the same striking effect on cell behaviour. The primary response occurs at stage 10 and is complete within 10 min; this is after the onset of normal mesoderm involution but several hours before its completion. The change in cell behaviour is consistent with the entire blastocoel roof being induced to form mesoderm. This results in changes in cell shape, locomotion and adhesion that correspond to those occurring in a spatially and temporally organized fashion around the marginal zone (Keller, 1976, 1986; Keller & Schoenwolf, 1977).

We have studied these early events and their effects on development by fix-dissection and histology of MIF-injected and control embryos from stage 9 onwards. The observations are summarized in Figs 2 and 3. These illustrate how cells of the inner layer of the blastocoel roof initially become less adhesive and more motile. Soon afterwards, this inner layer contracts due to deepening of cell shapes and maximization of cell contact area and a prominent interface separates it from the outer layer of cells. Rather than migrate over this substrate, the newly involuted normal mesoderm fuses with the inner blastocoel wall, while peripheral endodermal cells migrate to make contact with the entire blastocoel lining. The volume of the embryo is preserved by the appearance of a new cavity within the endoderm in the vegetal region.

Fig. 2.

Thin-sectional appearances at gastrula stages and the organization of somite muscle seen by immunofluorescence, after blastocoelic injection of MIF or control (XL-derived) dialysate. A–C show the marginal regions of stage-10 (early) gastrulae. A is a control embryo. B and C are, respectively, at very early and slightly more advanced stages in the minutes-long sequence of behaviour change in the inner blastocoel roof and wall, following injection with MIF at stage 7. Inner cells first become more loosely adhesive and surface-active, and then form a recognizably separate layer without localized, organized involution. This is followed by loss of the normal layered structure at the junction of endoderm, involuting mesoderm and blastocoel wall (arrows). D and E are higher power views of, respectively, the normal and abnormal blastocoel roof at the slightly later stage 10+. The inner, looser layers of cells in (E), consisting of ectopic mesoderm, how have the appearance of newly involuted mesoderm, except that there is no uninduced ectoderm available for them to migrate over. The large-celled yolky blastocoel floor appears at bottom. (F) Immunofluorescence for somite muscle in the tail of a normal larva, in grazing parasagittal section. Note the regular positioning of the fusiform cells in chevron-shaped blocks. (G) Image as in F but from an embryo that received a low dose (0·004 units) of MIF into the blastocoel (see Fig. 4A, second embryo). Note inferior extension and regularity of cell positioning, and the crazing and disorganization of the pattern. (H) Muscle immunofluorescence in the ventral site of the abnormally occluded blastocoel of an embryo that only just completed gastrulation and tailbud formation after MIF injection. An ectopic layer of striated muscle is visible. This forms from what would have normally produced anteroventral ectoderm. Scale bar, 300 μm for A–C, 100μm for D,E, 125 μm for F,G and 80 μm for H.

Fig. 2.

Thin-sectional appearances at gastrula stages and the organization of somite muscle seen by immunofluorescence, after blastocoelic injection of MIF or control (XL-derived) dialysate. A–C show the marginal regions of stage-10 (early) gastrulae. A is a control embryo. B and C are, respectively, at very early and slightly more advanced stages in the minutes-long sequence of behaviour change in the inner blastocoel roof and wall, following injection with MIF at stage 7. Inner cells first become more loosely adhesive and surface-active, and then form a recognizably separate layer without localized, organized involution. This is followed by loss of the normal layered structure at the junction of endoderm, involuting mesoderm and blastocoel wall (arrows). D and E are higher power views of, respectively, the normal and abnormal blastocoel roof at the slightly later stage 10+. The inner, looser layers of cells in (E), consisting of ectopic mesoderm, how have the appearance of newly involuted mesoderm, except that there is no uninduced ectoderm available for them to migrate over. The large-celled yolky blastocoel floor appears at bottom. (F) Immunofluorescence for somite muscle in the tail of a normal larva, in grazing parasagittal section. Note the regular positioning of the fusiform cells in chevron-shaped blocks. (G) Image as in F but from an embryo that received a low dose (0·004 units) of MIF into the blastocoel (see Fig. 4A, second embryo). Note inferior extension and regularity of cell positioning, and the crazing and disorganization of the pattern. (H) Muscle immunofluorescence in the ventral site of the abnormally occluded blastocoel of an embryo that only just completed gastrulation and tailbud formation after MIF injection. An ectopic layer of striated muscle is visible. This forms from what would have normally produced anteroventral ectoderm. Scale bar, 300 μm for A–C, 100μm for D,E, 125 μm for F,G and 80 μm for H.

Fig. 3.

The altered structure during the first half of normal gastrulation, and the morphology at larval stages, after intrablastocoelic injection of MIF. (A) Sagittal sections at control stage 10–, 1012 and 1112 in normal (sham-injected) and MIF-injected embryos. Presumptive dorsal side, where normal mesoderm involution occurs first, is to the right. The upper control drawing represents the onset of stage 10, up to which point control and MIF-injected development are morphologically identical. Second and third rows represent typical stage-1012 and 1112 embryos, with normal on the left and MIF-injected on the right, but with a split representation of the latter at stage 111 to show events typical of higher (left half) and lower (right half) doses. Tissue that is specified as mesoderm and has involuted or undergone the analogous change in cell behaviour is shown stippled throughout. Ectoderm and uninvoluted mesoderm is unshaded, and endoderm is hatched. The stage of development of external blastoporal lips and archenteron is shown in the outlines. At the time of MIF-induced abnormality, the normal mesoderm has just started to create an interface by migrating up the outer shell of preinvoluted mesoderm and presumptive ectoderm. The endodermal mass has a sharply limited border just ahead of the emerging mesoderm, because it has preferential affinity for that tissue. The two principal sequences of abnormal behaviour concern mesoderm, and then a consequent one by endoderm. There is a massive, synchronous ‘recruitment’ of mesodermal cells out of both the deep presumptive ectoderm and, we believe, the presumptive mesoderm as yet unrecruited by the normal process. Normal mesoderm already in a new layer then ceases its involution and movement, for lack of appropriate space and substrate. The normal development of a sharp pre- and post-involution interface (arrowheads), the deepening of the archenteron, and the advance of the blastopore over the yolk plug, all driven by mesoderm involution (Keller, 1986), are halted. Meanwhile the anterior edge of the endoderm mass alters its normal sharp outline, and its cells rapidly and abnormally achieve contact with the entire lining of the former blastocoel. The volume of the embryo is preserved by the appearance of an abnormal cavity within the endoderm in the vegetal region. After lower MIF doses (about 0 ·006 units), dorsal lip activity reappears and a pre- and post-involution interface forms by stage 111212 (arrowheads), whereas with doses above 0 ·02 units this never occurs, and the ectopic mesoderm may reveal its ‘dorsal’ status by its semiorganized convergent behaviour. (B) Camera lucida drawings of normal and a range of MIF-injected embryos at stage 35. The least abnormal case has all major head parts and all germ layers present, but heart and blood do not form. Increasing doses reduce the anterior axial pattern, the pronephros disappears and somite segmentation is disrupted. A significant accessory trunk-tail formation is often present, centred on the ectopic mesoderm region but running into what remains of the embryo’s ‘own’ axis anteriorly where there is no brain. At intermediate doses where gastrulation has only just recovered, the tailbud may be the only externally recognizable structure, although massive, normally positioned notochord and somite tissue are found internally. The ectoderm and nervous system have few cells and often disintegrate. Above this dose, gastrulation never occurs and a thick pad of disorganized mesoderm lies between the thin shell of abnormal ciliated epidermis (stippled) and the unenclosed endodermal mass. e, eyecup; ev, ear vesicle; g, gills and pharynx; pn, pronephros; s, somite segments; fb, forebrain above cement gland; bp, blastopore. Scale bars, 1mm.

Fig. 3.

The altered structure during the first half of normal gastrulation, and the morphology at larval stages, after intrablastocoelic injection of MIF. (A) Sagittal sections at control stage 10–, 1012 and 1112 in normal (sham-injected) and MIF-injected embryos. Presumptive dorsal side, where normal mesoderm involution occurs first, is to the right. The upper control drawing represents the onset of stage 10, up to which point control and MIF-injected development are morphologically identical. Second and third rows represent typical stage-1012 and 1112 embryos, with normal on the left and MIF-injected on the right, but with a split representation of the latter at stage 111 to show events typical of higher (left half) and lower (right half) doses. Tissue that is specified as mesoderm and has involuted or undergone the analogous change in cell behaviour is shown stippled throughout. Ectoderm and uninvoluted mesoderm is unshaded, and endoderm is hatched. The stage of development of external blastoporal lips and archenteron is shown in the outlines. At the time of MIF-induced abnormality, the normal mesoderm has just started to create an interface by migrating up the outer shell of preinvoluted mesoderm and presumptive ectoderm. The endodermal mass has a sharply limited border just ahead of the emerging mesoderm, because it has preferential affinity for that tissue. The two principal sequences of abnormal behaviour concern mesoderm, and then a consequent one by endoderm. There is a massive, synchronous ‘recruitment’ of mesodermal cells out of both the deep presumptive ectoderm and, we believe, the presumptive mesoderm as yet unrecruited by the normal process. Normal mesoderm already in a new layer then ceases its involution and movement, for lack of appropriate space and substrate. The normal development of a sharp pre- and post-involution interface (arrowheads), the deepening of the archenteron, and the advance of the blastopore over the yolk plug, all driven by mesoderm involution (Keller, 1986), are halted. Meanwhile the anterior edge of the endoderm mass alters its normal sharp outline, and its cells rapidly and abnormally achieve contact with the entire lining of the former blastocoel. The volume of the embryo is preserved by the appearance of an abnormal cavity within the endoderm in the vegetal region. After lower MIF doses (about 0 ·006 units), dorsal lip activity reappears and a pre- and post-involution interface forms by stage 111212 (arrowheads), whereas with doses above 0 ·02 units this never occurs, and the ectopic mesoderm may reveal its ‘dorsal’ status by its semiorganized convergent behaviour. (B) Camera lucida drawings of normal and a range of MIF-injected embryos at stage 35. The least abnormal case has all major head parts and all germ layers present, but heart and blood do not form. Increasing doses reduce the anterior axial pattern, the pronephros disappears and somite segmentation is disrupted. A significant accessory trunk-tail formation is often present, centred on the ectopic mesoderm region but running into what remains of the embryo’s ‘own’ axis anteriorly where there is no brain. At intermediate doses where gastrulation has only just recovered, the tailbud may be the only externally recognizable structure, although massive, normally positioned notochord and somite tissue are found internally. The ectoderm and nervous system have few cells and often disintegrate. Above this dose, gastrulation never occurs and a thick pad of disorganized mesoderm lies between the thin shell of abnormal ciliated epidermis (stippled) and the unenclosed endodermal mass. e, eyecup; ev, ear vesicle; g, gills and pharynx; pn, pronephros; s, somite segments; fb, forebrain above cement gland; bp, blastopore. Scale bars, 1mm.

Although the initial appearance of embryos injected with MIF does not depend upon the amount received, their further development does, for there is a dose-dependent ‘recovery’ of normal gastrulation. We have no explanation of how this occurs but it is clearly of great interest. Embryos injected with the lowest doses of MIF (0·003 units) recover to form an embryo whose only external abnormality is a reduced head. However, somite morphogenesis is significantly disorganized in such embryos (Fig. 2G) and no blood is formed. As the dose of MIF is increased, the embryos recommence normal gastrulation activity at successively later stages and form only posterior or tailbud structures. Heart and then pronephros development are removed from the body pattern. 47 embryos have been examined in this series of experiments, 35 using conventional histology and 12 by immunofluorescence using an antibody to Xenopus globins (Cooke & Smith, 1987).

Embryos are unable to recommence normal gastrulation after being injected with more than 0·02 units MIF. Real embryonic mesodermal pattern is not detectable and the abnormal single-layered epidermal remnant is sloughed off. However, ectopic mesoderm is invariably present at the site of the obliterated blastocoel, as it is in embryos injected with lower doses of MIF which have managed to recover normal gastrulation movements (Fig. 2H). At the highest doses of MIF this ectopic mesoderm includes muscle, but never notochord. Interestingly, when isolated animal pole regions are exposed to MIF there is a dose–response relationship such that concentrations greater than 10–20 units ml−1 frequently result in notochord formation (Symes & Smith, unpublished observations; see Smith, 1987a; Symes & Smith, 1987). The MIF concentration range in the blastocoel in these experiments is about 15–500 units ml−1 and thus the required dose range in intact embryos is elevated with respect to that in isolated animal pole tissue. This suggests that the normal mechanisms of mesoderm formation and patterning include a widely disseminated inhibitor, which down-regulates the threshold of response in embryos, but which is absent in animal pole explants (see also Smith, 1987a). We shall present further evidence for this elsewhere.

MIF behaves like a secreted protein that acts via cell surface receptors and a second messenger system

Equal volumes of MIF or of control dialysates were injected into five separate batches of sibling embryos, either intracellularly, into the animal hemisphere at the 1- to 4-cell stage, or into the blastocoel at the 128-cell stage. More than 150 embryos were involved, with five groups of 15–25 receiving intracellular injection but smaller groups receiving intrablastocoelic injection to confirm the familiar positive response (see above). Ectopic mesoderm is formed after intrablastocoelic injection of approximately 0·003 units MIF. However, this dose can be injected into the animal hemisphere of the precleavage zygote or the 2- to 4-cell-stage embryo without visible effect on development. Even doses of 0·005 unit into the single cell are without effect, though the control (XL-derived) dialysate is slightly toxic, producing some miscleavage and pregastrula arrest at the required injection volume.

It is possible that proteins with special function early in development are particularly rapidly degraded, so that intracellularly injected MIF would be destroyed before the onset of mesodermal competence at the 32- to 64-cell stage (Jones & Woodland, personal communication). Accordingly, in three of the experiments, three or four animal blastomeres at the 16- to 32-cell stage were given an aggregate dose of 0·003 units MIF from a very fine (and thus slowly delivering) micropipette. Embryos showing delayed cleavage after this treatment were discarded. Subsequently MIF was injected into the blastocoel of stage 7– embryos from the same egg batch under identical conditions but with only 75 % of the total aggregate delivery time per embryo. A striking MIF effect (see Figs 2, 3) was observed in all ten of the latter embryos, but 21 out of 25 intracellularly injected embryos were normal. The four abnormal embryos each had large single patches of cytolysis, possibly resulting from original overfill of one blastomere.

We conclude that microinjected MIF protein in an undamaged cell is ineffective, and is not secreted from the cell to act via extracellular receptors.

In contrast, like Woodland & Jones (1987), we find that intracellular injection of poly(A)+ RNA from XTC cells, but not XL cells, is effective. Approximately 10nl of lmgml−1 RNA, or water, was injected into the animal hemisphere of embryos at the 2- to 4-cell stage. These embryos were allowed to reach the mid-blastula stage (stage 8), when the animal pole regions were dissected out and allowed to develop until control stage 38 in 67% NAM. Nine explants from embryos injected with water were sectioned and examined by immunofluorescence with an anti-muscle antibody. All formed atypical epidermis. Seven explants were taken from embryos injected with XL poly(A)+ RNA, of which six formed atypical epidermis (Fig. 5A,B) and one formed some mesenchyme and mesothelium. Of the 14 explants from embryos injected with XTC poly(A)+ RNA, five formed atypical epidermis, and one mesenchyme and mesothelium but eight formed large masses of muscle (Fig. 5C,D). One even formed notochord.

These results suggest that newly synthesized MIF is capable of being secreted from blastomeres and is effective once outside the cell.

Blastocoel roof tissue exposed to MIF acts as ‘organizer’ mesoderm when grafted to normal host ventral marginal zones

Animal pole ectoderm was dissected from stage-9 Xenopus embryos that had received, at earlier blastula stages, intrablastocoelic injections of MIF. These pieces of tissue were washed and grafted to the ventral marginal zones of stage-9 embryos and the embryos were allowed to develop. In 65 out of 80 cases, from five egg batches, where donor tissue had received at least 0·025 units of MIF between 30 and 90min beforehand, the grafted embryos developed secondary axes with spinocaudal to hindbrain levels of neuraxis (Fig. 4), or else complete rostrocaudal pattern. Similar grafts from normal embryos or from embryos injected with control solutions are without effect (see control embryo, Fig. 4A); the grafted tissue integrates into the posterior trunk or tail region, visible as an abnormally pigmented patch of epidermis.

Fig. 4.

The effect of blastocoelic MIF injection on morphogenesis of embryos, and on properties of grafted blastocoel roof tissue. (A) A group of one normal and five abnormal siblings at stage 37. The normal individual is second from bottom. The upper group show external appearances after low (top) and moderate (2nd and 3rd) doses of MIF. Even individuals such as that at top have no heart or blood and little pronephric development, while somite disorganization (see Fig. 2G) and reduced epidermal and CNS cell populations are evident. As the syndrome becomes more severe, only posterior axial structure and notochord is recognizable, while unorganized muscle or significant accessory tail and CNS formation may occur anteroventrally. Higher MIF doses abort gastrulation.

Fig. 4.

The effect of blastocoelic MIF injection on morphogenesis of embryos, and on properties of grafted blastocoel roof tissue. (A) A group of one normal and five abnormal siblings at stage 37. The normal individual is second from bottom. The upper group show external appearances after low (top) and moderate (2nd and 3rd) doses of MIF. Even individuals such as that at top have no heart or blood and little pronephric development, while somite disorganization (see Fig. 2G) and reduced epidermal and CNS cell populations are evident. As the syndrome becomes more severe, only posterior axial structure and notochord is recognizable, while unorganized muscle or significant accessory tail and CNS formation may occur anteroventrally. Higher MIF doses abort gastrulation.

Fig. 5.

Injection of poly(A)+ RNA from XTC. but not XL. cells into animal pole blastomeres causes mesoderm induction. Poly(A)+ RNA from XL cells (A.B) or XTC cells (C,D) was injected into the animal hemisphere of Xenopus embryos at the 2- to 4-cell stage. At the mid-blastula stage (stage 8) the animal pole regions of these embryos were dissected out and allowed to develop in 67 % NAM. (A) DAPI staining. (B) The same section stained with 12/101; no muscle is formed. (C) DAPI staining. (D) The same section stained with 12/101; large amounts of muscle are formed in response to XTC poly(A)+ RNA. Scale bar in (D) is 200pm and applies to all frames.

Fig. 5.

Injection of poly(A)+ RNA from XTC. but not XL. cells into animal pole blastomeres causes mesoderm induction. Poly(A)+ RNA from XL cells (A.B) or XTC cells (C,D) was injected into the animal hemisphere of Xenopus embryos at the 2- to 4-cell stage. At the mid-blastula stage (stage 8) the animal pole regions of these embryos were dissected out and allowed to develop in 67 % NAM. (A) DAPI staining. (B) The same section stained with 12/101; no muscle is formed. (C) DAPI staining. (D) The same section stained with 12/101; large amounts of muscle are formed in response to XTC poly(A)+ RNA. Scale bar in (D) is 200pm and applies to all frames.

These experiments suggest that MIF-treated animal pole ectoderm acts as ‘organizer’ tissue, capable both of respecifying ventral mesoderm and of causing neural induction (Spemann & Mangold, 1924; Gimlich & Cooke, 1983; Smith & Slack, 1983; Jacobson, 1984). It is unlikely that the secondary axes are caused by passive release of MIF from the treated explants because in seven cases when MIF was injected into the blastocoels of stage-11 gastrulae and one hour later a larger panel of the thinner stage-11·5 blastocoel roof was used as a graft, no secondary axes were produced. Stage-11 tissue is no longer competent to respond to MIF (Symes & Smith, unpublished observations), and this suggests that induced stage-9 animal pole tissue causes pattern respecification by producing further inductive signals.

Further evidence that induced animal pole ectoderm acts as organizer tissue comes from its cellular behaviour. At the time of grafting the donor tissue is of quite normal appearance for its tissue of origin (see above). About 1h after grafting, however, when gastrulation would be commencing in the donor, active surface convergence and invagination begin at the graft site, as occurs with grafted dorsal lip tissue (Cooke, 1972). Furthermore, it is possible to count the number of induced animal pole cells in the graft, and as few as 100 are required for successful double axis formation. This indicates that there is a large host contribution to the secondary axes, and we are currently investigating this by cell lineage labelling, as well as examining the effects of varying graft size and dose of MIF on the patterns produced.

Homeogenetic induction

We have investigated whether animal pole ectoderm exposed to MIF can itself induce mesoderm by combining stage-8 MIF-treated ectoderm with RLDx-labelled uninduced ectoderm (Fig. 6). Twelve combinations were prepared in which one component was derived from an embryo which had received an intrablastocoelic injection of MIF one hour previously, at stage 7. The dissected animal cap region was washed before combining it with an RLDx-labelled uninduced explant at the same stage.

Fig. 6.

Explants combining MIF-treated and lineage-labelled animal cap tissue. (A) Preparation of explant combinations. The left-hand side shows how one donor embryo is lineage labelled with RLDx and allowed to develop to stage 8 (about 7 h). The centre column shows how competent animal cap tissue at stage 7 is exposed to MIF and excised with careful washing after about 1 h at stage 8+. The right-hand side shows how the same tissue, but past its competent period at stage 11+, is similarly exposed to MIF. Experimental combinations are prepared from left and centre components, control combinations from left and right. (B) Upper (muscle immunofluorescence) and lower (lineage label) images from a sectioned combination at control larval stages. The bulk of muscle develops from the directly MIF-exposed component, but morphogenesis frequently results in such muscle being overlain by lineage-labelled epidermis. In C–H, muscle immunofluorescent images from sections are on the left, and their paired lineage label images on the right. (C,D) Even within the mesodermal tissue that fills the explant combinations, nonmuscle regions including lineage-labelled cells frequently abutt onto muscle. Freely diffusing transfer of the original MIF does not seem to cause systematic muscle induction adjacent to treated tissue. (E,F and G,H) Two examples of the incorporation of lineage-labelled cells into the periphery of a region of muscle development. Arrowheads indicate cells labelled with RLDx that are also stained with anti-muscle antibody. Scale bar: 1 mm in drawings of A, and 65pm on photomicrographs B–H.

Fig. 6.

Explants combining MIF-treated and lineage-labelled animal cap tissue. (A) Preparation of explant combinations. The left-hand side shows how one donor embryo is lineage labelled with RLDx and allowed to develop to stage 8 (about 7 h). The centre column shows how competent animal cap tissue at stage 7 is exposed to MIF and excised with careful washing after about 1 h at stage 8+. The right-hand side shows how the same tissue, but past its competent period at stage 11+, is similarly exposed to MIF. Experimental combinations are prepared from left and centre components, control combinations from left and right. (B) Upper (muscle immunofluorescence) and lower (lineage label) images from a sectioned combination at control larval stages. The bulk of muscle develops from the directly MIF-exposed component, but morphogenesis frequently results in such muscle being overlain by lineage-labelled epidermis. In C–H, muscle immunofluorescent images from sections are on the left, and their paired lineage label images on the right. (C,D) Even within the mesodermal tissue that fills the explant combinations, nonmuscle regions including lineage-labelled cells frequently abutt onto muscle. Freely diffusing transfer of the original MIF does not seem to cause systematic muscle induction adjacent to treated tissue. (E,F and G,H) Two examples of the incorporation of lineage-labelled cells into the periphery of a region of muscle development. Arrowheads indicate cells labelled with RLDx that are also stained with anti-muscle antibody. Scale bar: 1 mm in drawings of A, and 65pm on photomicrographs B–H.

These combinations formed elongated structures surrounded by epidermis derived from both components. A cement gland was frequently present at one end, and this was invariably labelled with RLDx and so derived from the uninduced animal pole ectoderm. In sections of the explants, large masses of muscle were identified by indirect immunofluorescence with a fluorescein-conjugated second anti-body (Fig. 6B,C,E,G). The muscle was sometimes accompanied by notochord, and at least 90% of this axial mesoderm was derived from the unlabelled MIF-treated ectoderm. Lineage-labelled tissue, derived from the uninduced ectoderm, usually abutted the muscle cells (Fig. 6B-D). Significant numbers of lineage-labelled cells, both as individuals and as small groups, had contributed to peripheral parts of the muscle mass (Fig. 6E-H), but most of the lineage-labelled cells differentiated as mesenchyme, resembling the tissue seen in ‘weak’ or ‘ventral-type’ inductions (see above). It is probable that this tissue induces the cement gland from overlying ectoderm; in our hands, cement gland never forms spontaneously from uninduced animal pole ectoderm.

The mesenchymal tissue formed by lineage-labelled animal pole ectoderm must have arisen through a form of induction because fused pairs of uninduced animal cap regions never form mesoderm. The inducing signal is unlikely to be MIF diffusing from the induced animal pole tissue because in 12 experiments when RLDx-labelled blastula animal pole ectoderm was combined with ectoderm that had been exposed to MIF at stage 11 (Fig. 6A), the labelled tissue formed atypical epidermis. Stage-11 ectoderm is no longer competent to respond to MIF, and in 11 cases the unlabelled cells themselves formed epidermis and mixed with the labelled cells that had not been exposed to MIF. There was one unusual case where a little muscle was formed from the MIF-treated gastrula ectoderm.

These results suggest that passive transfer of MIF cannot account for the formation of muscle and mesenchyme by the uninduced component of experimental combinations. It is probable that these mesodermal cell types arise through a form of homeogenetic induction, in which newly induced ectoderm itself produces a mesoderm-induction signal. This is discussed below.

The discovery of soluble factors capable of inducing isolated animal pole ectoderm to form mesoderm (Smith, 1987a; Slack et al. 1987) has made possible a series of novel and informative experiments on mesodermal induction. In a previous paper, for example, we have shown that an early response to induction is the onset of gastrulation movements (Symes & Smith, 1987). The results described in the present paper emphasize some of the differences between induction and other cell signalling systems (see Gurdon, 1987) but also indicate some similarities. In addition, our data pose interesting questions about the specification of mesoderm and the initiation of pattern formation within it.

Mesoderm induction and other cell signalling systems

If poly(A)+ RNA from the XTC cell line is micro-injected into the animal pole region of early Xenopus embryos, the prospective ectoderm of the blastula becomes capable of differentiating as mesoderm

(Woodland & Jones, 1987; see our Fig. 5). By contrast, when partially purified MIF is microinjected into the same cells it is without effect, although it will induce mesoderm if applied extracellularly, by injection into the blastocoel. This suggests that the activity of microinjected poly(A)+ RNA is due to secretion of MIF and that MIF acts via extracellular receptors. In this respect, the mesoderm-induction signal resembles, for example, the polypeptide growth factors like epidermal growth factor (EGF) (Carpenter, 1985) rather than steroid hormones (reviewed by Tata, 1984).

In two respects, however, mesoderm induction clearly differs from growth factor action. The first is in the timing of the response. Gurdon et al. (1985) and Symes & Smith (1987) have previously demonstrated that the time of muscle-specific gene activation and the onset of gastrulation movements do not depend upon the time of mesoderm induction. Here we show that the change in structure of the blastocoel roof after MIF injection begins at the same stage irrespective of the time of injection or the concentration of MIF. This change appears to be the abnormal, ectopic counterpart .of those changes normally causing involution and then convergent extension of mesoderm. The time of response is therefore set by the age of the responding tissue rather than by the time of induction. This is a phenomenon unique to inductive interactions (Gurdon, 1987).

The effect of MIF can be ‘remembered’ by responding cells within 10min. This swift response is not unprecedented. Some growth factors, like EGF, need to be present in the culture medium continuously to exert their effect (Shechter, Hernaez & Cuatracasas, 1978) but others, like platelet-derived growth factor (PDGF), rapidly induce a long-term cellular memory which is retained in the absence of the factor (Pledger, Stiles, Antoniades & Scher, 1977; Singh et al. 1983). Acquisition of this ‘memory’ requires RNA synthesis (Smith & Stiles, 1981) and may depend upon the transcription of oncogenes such as c-myc and c-fos as well as other genes (Cochran, Reffel & Stiles, 1983; Armelin et al. 1984; Greenberg & Ziff, 1984). What is unusual about the rapid response to MIF is that it can occur in the virtual absence of transcription; that is, before the mid-blastula transition (Newport & Kirschner, 1982). This may indicate that the signal is remembered in some other way, perhaps by binding of the factor to the cell surface. Alternatively, the extremely low level of transcription that can be detected before the MBT (Kimelman, Kirschner & Scherson, 1987) may be sufficient to record the effect of MIF.

Above and below the control embryo are two individuals with highly organized but anteriorly incomplete second axial patterns, following the grafting operation described in B. (B) An operation in which a piece of blastocoel roof tissue from a donor blastula, previously injected with MIF or a control solution, is grafted into the ventral marginal zone of a late blastula host. Grafts (100–300 cells) from control injected donors or from those receiving less than about 0·06 units MIF, integrate into host posteroventral regions with no disturbance to morphogenesis (see lower larval sketch). Grafts treated with higher doses of MIF, even after careful washing before implantation, act like dorsal blastoporal lip tissue and organize the development of second axial patterns (see upper larval sketch and A,C). The host blastula is represented in sagittal section with its own prospective dorsal lip site to left. (C) Anterior tissue structure in horizontal sections from the two secondary axes following grafting shown in (A). The level of anterior structure reached in these depends upon graft size and original donor MIF dosage. Both those shown are incomplete, one reaching pronephric levels but showing a small notochord piece in its otherwise fused somite axis, the other reaching ear vesicle and hind brain induction levels, without notochord. Note highly organized somite segmentation, which does not occur anywhere within donor embryos after the doses of MIF used in these experiments (see Fig. 2G). eg, cement gland; nc, notochord tissue; ev, ear vesicle; pn, pronephros; s, somites; tb, tailbud. Scale bar: 1 mm for drawings of B.

Mesodermal specification and pattern formation

The establishment of the spatial pattern of cellular differentiation within the mesoderm of Xenopus depends upon signals received before and during gastrulation as well as on the gastrulation movements themselves (Smith, Dale & Slack, 1985; Smith, 1987b; Keller, 1986; Cooke & Smith, 1987). The work described in this paper uses partially purified MIF to investigate these signals. The experiments fall into two classes, those on intact embryos and those on explants.

Injection of MIF into the blastocoel of intact embryos

Microinjection of partially purified MIF into the blastocoel of intact embryos causes all the cells lining the blastocoel to experience a homogenous concentration of inducing factor. As a result the cells undergo changes in cell shape, locomotion and adhesion which mimic those occurring during gastrulation in the marginal zone (see also Symes & Smith, 1987).

These changes in cell behaviour begin shortly after the onset of normal mesoderm involution. However, a large body of evidence (reviewed by Cooke, 1985; see also Cooke, 1987) suggests that the time at which the movements of involution occur in different regions of the normal mesoderm is characteristic of each region, and reflects its preliminary specification for body position and mode of differentiation. On this hypothesis, the time of change of behaviour of ectopic mesoderm might equally reflect its positional character. This time may be estimated very precisely by frequent inspection of synchronous sibling embryos, and it is thus possible to ask whether it depends upon the time of receipt, or concentration, of an inducing signal. Preliminary results, based on several hundred injected embryos, show that the greatest time difference that can be observed is less than 30min. This is short compared with the hours-long progression of normal mesoderm involution (Keller, 1976; Gerhart & Keller, 1986). Thus differences in timing and intensity of a single endogenous mesoderm-inducing factor may be insufficient to establish the time course of gastrulation and the spatial pattern of cellular differentiation.

One consequence of the change in behaviour of blastocoel roof tissue in response to MIF is the inhibition of gastrulation. However, although the timing of the change and the initial appearance of the cells do not depend upon the dose of MIF, there does occur a dose-dependent recovery of normal gastrulation movements, giving rise to an organized axis and induced nervous system (Cooke & Smith, unpublished data; see Fig. 3).

There are two effects of injected MIF on cell differentiation. Consistent with our observation that higher doses of MIF tend to induce dorsal structures and lower doses ventral (Symes & Smith, unpublished observations; see Grunz, 1983), injection of MIF suppresses blood and pronephros differentiation; it is as if the concentration of MIF within the embryo is nowhere low enough to permit such cell types to form. On the other hand, even when high doses of MIF are injected into the blastocoel, the blastocoel roof cells form muscle but never noto-chord. At equivalent concentrations of MIF isolated animal pole regions frequently form notochord, and this suggests that the embryo contains an inhibitor of mesoderm formation as well as an activator.

Explant experiments

One conclusion from the experiments in which MIF is injected into whole embryos is that the factor is capable of modifying pattern within the mesoderm but is not a molecule which determines particular pathways of cell differentiation; it is not, for example, the notochord-inducing factor, or the muscle-inducing factor.

Further evidence for this view comes from experiments using animal pole explants. A significant observation is that the proportion of cells in an explant that are exposed to MIF determines its subsequent appearance and cellular constitution (Fig. 1). For example, when an explant is exposed to a high concentration of MIF immediately after dissection it forms axial mesoderm, including notochord and muscle. However, if only a small area of such an explant is induced, by allowing it to round up before being exposed to MIF, it does not form smaller amounts of axial mesoderm, but structures containing mesenchyme and mesothelium. These explants resemble the ‘ventral’ explants obtained when FGF is used as an inducer (Slack et al. 1987) but are also observed in response to low concentrations of MIF or when ectoderm near the end of its competent period is exposed to high concentrations (Symes & Smith, unpublished observations). These results suggest that a small group of induced cells can coordinate meso-dermal differentiation throughout an entire animal pole explant.

The organizing ability of induced animal pole ectoderm is also observed when such tissue is fused with a lineage-labelled uninduced explant. A few cells of the uninduced explant form muscle, in what might be regarded as a strong form of homeogenetic induction – the induction of like by like (see Slack, 1983). But a more interesting observation is that many of the lineage-labelled cells differentiate as mesenchyme, which in turn induces overlying ectoderm cells to form cement gland. The mesenchymal cells arise through a form of homeogenetic induction in which mesoderm induces mesoderm, but which also permits the establishment of regional pattern, a phenomenon made possible by the relatively large size of the explants used in these experiments. This provides further evidence for the notion that MIF initiates mesoderm formation and patterning rather than acting as a factor that induces specific cell types.

The existence of homeogenetic induction in amphibian development has previously been inferred from grafting experiments in the slower-developing newt Cynops pyrrhogaster (Kaneda, 1981; Kaneda & Suzuki, 1983). Furthermore, using carp swimbladder as a mesoderm-inducing factor, Kurihara & Sasaki (1981) have also demonstrated that ectoderm induced to become muscle and notochord will induce adjacent ectoderm to become mesenchyme and mesothelium. A normal role for homeogenetic mesoderm induction in Xenopus can be inferred from previously published work, although an explicit experimental study is desirable. Thus clonal analysis of particular early blastomeres (Gimlich & Cooke, 1983; Dale & Slack, 1987a) together with study of the differentiation of blastula animal cap explants (this paper and Dale et al. 1985) shows that groups of cells which would never be in direct contact with the endodermal source of induction nevertheless make a large contribution to mesoderm and receive their inducing signal relatively late in development.

Further patterning of the mesoderm of Xenopus is believed to arise through ‘dorsalization’, an interaction distinct from both mesoderm induction and homeogenetic induction, in which dorsal ‘organizer’ mesoderm induces ‘ventral’ mesoderm to adopt an intermediate level, such as muscle (see Dale & Slack, 1987b). Dorsalization is one of the interactions revealed by the ‘organizer graft’ (Smith & Slack. 1983) and our results (Fig. 4) show that animal pole ectoderm exposed to MIF is also active in this assay. The other interaction demonstrated by the organizer graft, neural induction (Gimlich & Cooke, 1983; Smith & Slack, 1983; Jacobson, 1984), also occurs.

The impression from these results is that pattern formation within the mesoderm of Xenopus is not a single-step process. It may be initiated by an inducing factor (for which the present partially purified molecule is a good candidate) and polarity of the pattern may be set by different concentrations, or by the localized origin, of that factor in vivo (e.g. Gimlich & Gerhart, 1984). Further building up and refinement of the pattern could result from subsequent homeo-genetic induction and dorsalization, probably involving the serial production of, and response to, additional signalling molecules. These may act to ‘fine-tune’ the early specification, thus ensuring the remarkable constancy of pattern seen in the mesoderm of Xenopus (Cooke, 1981; Cooke & Smith, 1987).

We thank Jenny Hanson, Brigid Hogan and Robb Krumlauf for help and advice with RNA preparation, and Jeremy Brockes for the monoclonal antibody 12/101. We are most grateful to Liz Jones and Hugh Woodland for communicating some of their unpublished observations.

Anizet
,
M. P.
,
Huwe
,
B.
,
Pays
,
A.
&
Picard
,
J. J.
(
1981
).
Characterization of a new cell line, XL2, obtained from Xenopus laevis and determination of optimal culture conditions
.
In Vitro
17
.
267
274
.
Armelin
,
H. A.
,
Armelin
,
M. C. S.
.
Kelly
.
K.
.
Stewart
,
T.
,
Leder
,
P.
,
Cochran
,
B. H.
&
Stiles
.
C. D.
(
1984
).
Functional role for c-myc in mitogenic response to platelet-derived growth factor
.
Nature. Lond
.
310
,
655
660
.
Carpenter
,
G.
(
1985
).
Epidermal growth factor: biology and receptor metabolism
.
J. Cell Sci. Suppl
.
3
.
1
9
.
Cochran
,
B. H.
.
Reffel
,
A. C.
&
Stiles
,
C. D.
(
1983
).
Molecular cloning of gene sequences regulated by platelet-derived growth factor
.
Cell
33
.
939
947
.
Cooke
,
J.
(
1972
).
Properties of the primary organization field in the embryo of Xenopus laevis. III. Retention of polarity in cell groups excised from the region of the early organizer
.
J. Embryol. exp. Morph
.
28
.
47
56
.
Cooke
,
J.
(
1973
).
Properties of the primary organization field in the embryo of Xenopus laevis. V. Regulation after removal of the head organizer in normal early gastrulae and in those already possessing a second implanted organizer
.
J. Embryol. exp. Morph
.
30
.
283
300
.
Cooke
,
J.
(
1979
).
Cell number in relation to primary pattern formation in the embryo of Xenopus laevis. The cell cycle during new pattern formation in response to implanted organizers
.
J. Embryol. exp. Morph
.
51
.
165
182
.
Cooke
,
J.
(
1981
).
Scale of body pattern adjusts to available cell number in amphibian embryos
.
Nature, bond
.
290
,
775
778
.
Cooke
,
J.
(
1985
).
The system specifying body position in the early development of Xenopus, and its response to early perturbations
.
J. Embryol. exp. Morph
.
89
Supplement
,
69
87
.
Cooke
,
J.
(
1987
).
Dynamics of the control of body pattern in the development of Xenopus laevis. IV. Timing and pattern in the development of twinned bodies after re-orientation of eggs in gravity
.
Development
99
,
417
428
.
Cooke
,
J.
&
Smith
,
J. C.
(
1987
).
The midblastula cell cycle transition and the character of mesoderm in u.v.-irradiated nonaxial Xenopus development
.
Development
99
,
197
210
.
Dale
,
L.
&
Slack
,
J. M. W.
(
1987a
).
Fate map for the 32-cell stage of Xenopus laevis
.
Development
99
,
527
551
.
Dale
,
L.
&
Slack
,
J. M. W.
(
1987b
).
Regional specification within the mesoderm of early embryos of Xenopus laevis
.
Development
100
,
279
295
.
Dale
,
L.
,
Smith
,
J. C.
&
Slack
,
J. M. W.
(
1985
).
Mesoderm induction in Xenopus laevis: a quantitative study using a cell lineage label and tissue-specific antibodies
.
J. Embryol. exp. Morph
.
89
,
289
312
.
Gerhart
,
J. C.
(
1980
).
Mechanisms regulating pattern formation in the amphibian egg and early embryo
.
In Biological Regulation and Development
, vol.
2
(ed.
R. F.
Goldberger
), pp.
133
293
.
New York
:
Plenum
.
Gerhart
,
J.
&
Keller
,
R.
(
1986
).
Region-specific cell activities in amphibian gastrulation
.
A. Rev. Cell Biol
.
2
,
201
229
.
Gimlich
,
R. L.
&
Braun
,
J.
(
1985
).
Improved fluorescent compounds for tracing cell lineage
.
Devi Biol
.
109
,
509
514
.
Gimlich
,
R. L.
&
Cooke
,
J.
(
1983
).
Cell lineage and the induction of second nervous systems in amphibian development
.
Nature, Lond
.
306
,
471
473
.
Gimlich
,
R. L.
&
Gerhart
,
J. C.
(
1984
).
Early cellular interactions promote embryonic axis formation in Xenopus laevis
.
Devi Biol
.
104
,
117
130
.
Greenberg
,
M. E.
&
Ziff
,
E. B.
(
1984
).
Stimulation of 3T3 cells induces transcription of the c-fos proto-oncogene
.
Nature, Lond
.
311
,
433
438
.
Grunz
,
H.
(
1983
).
Changes in the differentiation pattern of Xenopus laevis ectoderm by variation of the incubation time and concentration of vegetalizing factor
.
Wilhelm Roux Arch, devl Biol
.
192
,
130
137
.
Gurdon
,
J. B.
(
1987
).
Embryonic induction - molecular prospects
.
Development
99
,
285
306
.
Gurdon
,
J. B.
& FAIRMAN. S
. (
1986
).
Muscle gene activation by induction and the non requirement for cell division
.
J. Embrvol. exp. Morph
.
97
Supplement
.
75
84
.
Gurdon
,
J. B.
.
Fairman
.
S.
,
Mohun
,
T. J.
&
Brennan
.
S.
(
1985
).
The activation of muscle-specific actin genes in Xenopus development by an induction between animal and vegetal cells of a blastula
.
Cell
41
,
913
922
.
Jacobson
,
M.
(
1984
).
Cell lineage analysis of neural induction: origins of cells forming the induced nervous system
.
Devi Biol
.
102
,
122
129
.
Kalt
,
M. R.
(
1971
).
The relationship between cleavage and blastocoel formation in Xenopus laevis. II. Electron microscopic observations
.
J. Embryol. exp. Morph
.
26
.
51
66
.
Kaneda
,
T.
(
1981
).
Studies on the formation and state of determination of the trunk organizer in the newt Cynops pyrrhogaster. III. Tangential induction in the dorsal marginal zone
.
Dev. Growth Differ
.
23
,
553
564
.
Kaneda
,
T.
&
Suzuki
,
A. S.
(
1983
).
Studies on the formation and state of determination of the trunk organizer in the newt, Cynops pyrrhogaster. IV. The association of the neural-inducing activity with the mesodermization of the trunk organizer
.
Wilhelm Roux Arch, devl Biol
.
192
,
8
12
.
Keller
,
R. E.
(
1975
).
Vital dye mapping of the gastrula and neurula of Xenopus laevis. I. Prospective areas and morphogenetic movements of the superficial layer
.
Devl Biol
.
42
.
222
241
.
Keller
,
R. E.
(
1976
).
Vital dye mapping of the gastrula and neurula of Xenopus laevis. II. Prospective areas and morphogenetic movements of the deep layer
.
Devl Biol
.
51
,
118
137
.
Keller
,
R. E.
(
1986
).
The cellular basis of amphibian gastrulation
.
In Developmental Biology: A Comprehensive Synthesis
, vol.
2
(ed.
L.
Browder
), pp.
241
327
.
New York
:
Plenum Press
.
Keller
,
R. E.
&
Schoenwolf
,
G. C.
(
1977
).
An SEM study of cellular morphology, contact, and arrangement as related to gastrulation in Xenopus laevis
.
Wilhelm Roux Arch, devl Biol
.
182
,
165
186
.
Kimelman
,
D.
,
Kirschner
,
M.
&
Scherson
,
T.
(
1987
).
The events of the midblastula transition in Xenopus are regulated by changes in the cell cycle
.
Cell
48
,
399
407
.
Kintner
,
C. R.
&
Brockes
,
J. P.
(
1984
).
Monoclonal antibodies identify blastemal cells derived from dedifferentiating muscle in newt limb regeneration
.
Nature, Lond
.
308
,
67
69
.
Krumlauf
,
R.
,
Holland
,
P. W. H.
,
Mcvey
,
J. H.
&
Hogan
,
B. L. M.
(
1987
).
Developmental and spatial patterns of expression of the mouse homeobox gene Hox2.1
.
Development
99
,
603
617
.
Kurihara
,
K.
&
Sasaki
,
N.
(
1981
).
Transmission of homoiogenetic induction in presumptive ectoderm of newt embryo
.
Dev. Growth Differ
.
23
,
361
369
.
Maniatis
,
T.
,
Fritsch
,
E. F.
&
Sambrook
,
J.
(
1982
).
Molecular Cloning. A Laboratory Manual
.
New York
:
Cold Spring Harbor
.
Newport
,
J.
&
Kirschner
,
M.
(
1982
).
A major developmental transition in early Xenopus embryos: I. Characterization and timing of cellular changes at the midblastula stage
.
Cell
30
,
675
686
.
Nieuwkoop
,
P. D.
(
1969
).
The formation of mesoderm in Urodelean amphibians. I. Induction by the endoderm
.
Wilhelm Roux Arch, devl Biol
.
162
,
341
373
.
Nieuwkoop
,
P.
&
Faber
,
J.
, eds (
1967
).
“Normal Table of Xenopus laevis (Daudin)’’
, 2nd edn.
Amsterdam
:
North-Holland
.
Pledger
,
W. J.
,
Stiles
,
C. D.
,
Antoniades
,
H. N.
&
Scher
,
C. D.
(
1977
).
Induction of DNA synthesis in BALB/c-3T3 cells by serum components: Reevaluation of the commitment process
.
Proc. natn. Acad. Sci. U.S.A
.
74
,
4481
4485
.
Pudney
,
M.
,
Varma
,
M. G. R.
&
Leake
,
C. J.
(
1973
).
Establishment of a cell line (XTC-2) from the South African clawed toad, Xenopus laevis
.
Experientia
29
,
466
467
.
Regen
,
C. M.
&
Steinhardt
,
R. A.
(
1986
).
Global properties of the Xenopus blastula are mediated by a high-resistance epithelial seal
.
Devi Biol
.
113
,
147
154
.
Shechter
,
Y.
,
Hernaez
,
L.
&
Cuatracasas
,
P.
(
1978
).
Epidermal growth factor: Biological activity requires persistent occupation of high-affinity cell surface receptors
.
Proc. natn. Acad. Sci. U.S.A
.
75
,
5788
5791
.
Singh
,
J. P.
,
Chaikin
,
M. A.
,
Pledger
,
W. J.
,
Scher
,
C. D.
&
Stiles
,
C. D.
(
1983
).
Persistence of the mitogenic response to platelet-derived growth factor (competence) does not reflect a long-term interaction between the growth factor and the target cell
.
J. Cell Biol
.
96
,
1497
1502
.
Slack
,
C.
&
Warner
,
A.
(
1973
).
Intracellular and intercellular potentials in the early amphibian embryo
.
J. Physiol., Land
.
232
,
313
330
.
Slack
,
J. M. W.
(
1983
).
From Egg to Embryo
.
Cambridge, London
:
Cambridge University Press
.
Slack
,
J. M. W.
(
1984
).
Regional biosynthetic markers in the early amphibian embryo
.
J. Embryol. exp. Morph
.
80
,
289
319
.
Slack
,
J. M. W.
,
Darlington
,
B. G.
,
Heath
,
J. K.
&
Godsave
,
S. F.
(
1987
).
Mesoderm induction in early Xenopus embryos by heparin-binding growth factors
.
Nature, Load
.
326
,
197
200
.
Smith
,
J. C.
(
1987a
).
A mesoderm-inducing factor is produced by a Xenopus cell line
.
Development
99
,
3
14
.
Smith
,
J. C.
(
1987b
).
Cellular interactions in establishment of regional patterns of cell fate during development
.
In Developmental Biology: A Comprehensive Synthesis
, vol.
5
(ed.
L.
Browder
) (in press).
Smith
,
J. C.
,
Dale
,
L.
&
Slack
,
J. M. W.
(
1985
).
Cell lineage labels and region-specific markers in the analysis of inductive interactions
.
J. Embryol. exp. Morph
.
89
Supplement
,
317
331
.
Smith
,
J. C.
&
Slack
,
J. M. W.
(
1983
).
Dorsalization and neural induction: properties of the organizer in Xenopus laevis
.
J. Embryol. exp. Morph
.
78
,
299
317
.
Smith
,
J. C.
&
Stiles
,
C. D.
(
1981
).
Cytoplasmic transfer of the mitogenic response to platelet-derived growth factor
.
Proc. natn. Acad. Sci. U.S.A
.
78
,
4363
4367
.
Spemann
,
H.
&
Mangold
,
H.
(
1924
).
Uber Induktion von Embryonenanlagen durch Implantation artfremder Orgamsatoren
.
Wdhelm Roux Arch. EntwMech. Org
.
100
,
599
638
.
Symes
,
K.
&
Smith
,
J. C.
(
1987
).
Gastrulation movements provide an early marker of mesoderm induction in Xenopus laevis
.
Development
101
,
339
349
.
Tata
,
J. R.
(
1984
).
The action of growth and developmental hormones: evolutionary aspects
.
In Biological Regulation and Development
. vol.
3b
(ed.
R. F.
Goldberger
&
K.
Yamamoto
), pp.
1
58
.
New York
:
Plenum
.
Warner
,
A.
&
Gurdon
,
J. B.
(
1987
).
Functional gap junction are not required for muscle gene activation by induction in Xenopus embryos
.
J. Cell Biol
.
104
,
557
564
.
Wolpert
,
L.
(
1969
).
Positional information and the spatial pattern of cellular differentiation
.
J. theor. Biol
.
25
,
1
47
.
Woodland
,
H. R.
&
Jones
,
E. A.
(
1987
).
The development of an assay to detect mRNAs that affect early development
.
Development
101
,
925
930
.