Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
599 Development 106, 599-610 (1989) Printed in Great Britain © The Company of Biologists Limited 1989 Lithium changes the ectodermal fate of individual frog blastomeres because it causes ectopic neural plate formation STEVEN L. KLEIN and SALLY A. MOODY Department of Anatomy and Cell Biology, University of Virginia Health Sciences Center, Box 439 Medical Center, Charlottesville, VA 22908, USA Summary Amphibian blastulae that are treated with lithium (Li) develop into embryos that consist almost exclusively of head structures. This dramatic change in embryogenesis may occur either because Li selectively kills trunk progenitors or because Li causes trunk progenitors to become head progenitors. To distinguish between these possibilities, we compared the fates of individual frog blastomeres between Li-treated embryos and normal embryos using lineage tracers. The results demonstrate that Li causes ventral midline cells, which normally populate large amounts of trunk, to produce many head structures, including the brain. Examination of fluorescently labeled clones in living Li-treated gastrulae shows that: (1) the ectodermal members of the clones migrate normally, and chordamesodermal involution begins normally; (2) the chordamesoderm's later in- volution is altered such that it is confined to the vegetal hemisphere; (3) accordingly, the neural plate forms in the vegetal hemisphere, circumscribing the blastopore, which normally gives rise to the cloaca; and (4) the ectodermal progeny of the ventral midline blastomeres that are near the blastopore populate the brain because they are induced by the stalled chordamesoderm to form part of the ectopic neural plate. These results demonstrate that Li, administered during a short developmental window at early cleavage stages, ultimately alters ectodermal fate because it changes the pattern of chordamesodermal involution during gastrulation, which in turn changes the site of neural plate formation. Introduction olution of this important issue may indicate the mode of action of Li-teratogenesis, the way in which cleavagestage manipulations alter cell fate, and the mechanisms that control cell fate during normal development. In this study, we marked single blastomeres of early cleavage stage Xenopus laevis embryos with lineage tracers, exposed the embryos to Li, and compared the distribution of their progeny to that of normal embryos (Moody, 1987a,b). We show that Li causes large changes in cell fate, especially in ectodermal descendants, and that these changes result from alterations in the pattern of gastrulation. Lithium (Li), a known teratogen (Briggs et al. 1986), suppresses trunk formation and enhances head formation when it is administered briefly to cleavage-stage frog embryos (Kao et al. 1986; Breckenridge et al. 1987; Condie & Harland, 1987; Regen & Steinhardt, 1987; Kao & Elinson, 1988,1989; Cooke & Smith, 1988; Slack etal. 1988; Klein & King, 1988; Busa & Gimlich, 1989). The gross morphology of Li-treated embryos (Fig. 1A) clearly illustrates that a major change in embryonic cell fate has occurred. This change may result from alterations in cleavage stage intercellular communication (Kao et al. 1986; Kao & Elinson, 1988), because Li is effective only when administered during cleavage stages (Kao et al. 1986; Breckenridge et al. 1987; Cooke & Smith, 1988) and because it blocks the inositol trisphosphate-diacylglycerol second messenger pathway (Berridge et al. 1982; Sherman et al. 1981; Berridge, 1987; Busa & Gimlich, 1989). One of the most striking fate changes is a doubling of the number of neurons (Breckenridge etal. 1987). However, previous studies have not indicated whether Li causes non-neuronal progenitor cells to produce neurons, or whether the normal neuronal ancestors overproduce their normal progeny. Res- Key words: cell fate, gastrulation, lithium treatment, lineage tracers. Materials and methods Embryos of Xenopus laevis were obtained, dejellied, selected and injected with lineage tracers as described by Moody (1987a). One blastomere per embryo was labeled by microinjection of lnl of either horseradish peroxidase (HRP, Boehringer Mannheim) or fluorescein-dextran-amine (FDA, Molecular Probes). Embryos were incubated in 300mM-LiCl for a 6min interval during the 32-, to 128-cell stage, followed by extensive rinsing in Steinberg's solution. This treatment produced a majority (up to 90 %) of head-only embryos with normal faces and heads (Fig. 1A), which met 600 5. L. Klein and S. A. Moody the criteria of DAI 7 and 8 (dorsoanterior enhancement index) of Kao & Elinson (1988), in which DAI 5 are normal and DAI 10 are radially symmetric dorsoanterior embryos. When control embryos reached stage 33-43 (Fig. ID; Nieuwkoop & Faber, 1964), the Li-embryos were fixed, cryostat sectioned and processed for either histochemistry (diaminobenzidine reaction for HRP; Moody, 1987a) or for fluorescence (FDA) by conventional methods. Single identified blastomeres along the dorsal and ventral midline of 16-, and 32-cell embryos were injected with the lineage tracers. The positions of these blastomeres are illustrated in Figs 2 and 3. The dorsal midline blastomeres (Dl.l and D2.1) of the normal 16-cell embryo are major head progenitors. Both blastomeres contribute progeny to all head mesoderm and endoderm, and Dl.l contributes heavily to all head ectoderm, including the brain (Moody, 1987a). Of the ventral midline blastomeres, V2.1 contributes virtually no progeny to the head, and Vl.l's only head progeny populate branchial arch mesoderm and non-central nervous system ectoderm (Moody, 1987a). The distribution of the labeled progeny within each organ of the Li-embryo was mapped from complete serial sections, as described previously for normal embryos (Moody, 1987a). Fate maps identical to those constructed for normal embryos were made for Li-treated embryos. In order to minimize population variations (see Moody, 1987ft), at least 10 embryos per blastomere were analyzed. Although these DAI 7/8 embryos were 'dorsoanterior enhanced' (Kao & Elinson, 1988), a small amount of rostral trunk persisted. For our analysis, we considered an organ to be a trunk structure if it was located posterior to the otocyst in transversely sectioned material. Complete fate maps were constructed for 86 HRP-labeled Li-embryos. The relative proportion of labeled progeny in each organ was assessed in coded specimens as described previously (Moody, 1987a). Quantitative estimates of cell fate were made by assigning a numerical value to each region of each organ as follows: no labeled cells = zero; five or fewer labeled cells = 0-5; many labeled cells = 1-5; almost completely labeled = 3-0. The relative contribution of the blastomere to each organ was determined by summing the value of each organ region and averaging the numerical value of the organ over all specimens in which that blastomere had been injected. Using this procedure, the average amount of central nervous system (CNS) that is produced by a normal VI. 1 is: forebrain, 0-17 (S.E. =0-17); midbrain, 006 (0-06); hindbrain, 0-05 (0-05); spinal cord, 1-3 (0-13). The average amount of CNS produced by a Li-treated VI. 1 is: forebrain, 0-86 (0-64); midbrain, 1-84 (0-48); hindbrain, 3-63 (0-49); anterior spinal cord, 3-47 (0-35). The posterior spinal cord was not present in the DAI 7/8 embryos. The difference between the fate of the Li-treated blastomere and the normal blastomere was determined by subtracting the normal average value from the Li-average value (Fig. 2). Differences of ±1 were considered to be significant because they distinguish between a few and many labeled progeny. To determine whether the CNS progeny of ventral blastomeres were actually neurons, many FDA-labeled embryos were prepared for indirect immunofluorescent localization of neurofilament proteins. Frozen tissue sections were incubated in a monoclonal antibody directed against the phosphorylated high and middle molecular weight atxon-specific neurofilament proteins (Sternberger-Meyer; SMI 31 at a 1:50000 dilution), and a goat anti-mouse secondary antibody conjugated to RITC (HyClone at a 1:100 dilution). We examined whether Li treatment altered the pattern of gastrulation by monitoring the position of FDA-labeled clones in both living and fixed specimens. During gastrulation, neurulation and tail bud stages, the location of the superficial members of the FDA-labeled clones of living embryos (33 normal and 34 Li-treated embryos that developed to DAI 8) were examined continuously with a Wild M5 stereomicroscope equipped for epifluorescence and an image intensifier (KS-1380; Videoscope International, Ltd). The positions of both the superficial and deep members of the clones also were examined by dissecting 43 normal and 42 Liembryos that were fixed between the beginning of gastrulation and the tail bud stages. Finally, in order to confirm the ectopic position of the neural plate, 35 uninjected normal and 43 uninjected Liembryos were fixed and examined between neurula and tailbud stages. The blastopore lip of these gastrulae were marked with a spot of the vital dye Nile blue sulfate (Kirschner & Hara, 1980). The dye was applied to the dorsal blastopore lip of the normal embryos, and to a portion of the Li-blastopore lip. However, we could not identify the specific region of the Li-blastopore lip that was labeled because the lip is not visibly polarized in Li-embryos. Results Lithium-treated ventral blastomeres populate the central nervous system The changes in cell fate that the Li-treatment produced were most obvious in the CNS. The two dorsal blastomeres, which contribute to CNS normally, continued to do so. However, the two ventral blastomeres, which normally make significant contributions to only the dorsal spinal cord, contributed numerous progeny to the CNS. The VI. 1 blastomere (ventral animal midline cell) showed the most dramatic change in cell fate. In the normal embryo, the ectodermal progeny of VI. 1 populate large amounts of epidermis in both the head and the trunk; they populate virtually none of the brain (Fig. IE), and only a small amount of the dorsal spinal cord (Moody, 1987a). But, in the Li-treated embryo, Vl.l's progeny populated large areas of brain and retina (Fig. IB, C). Many of these progeny had a neuronal morphology, with long axon-like processes, most of which were in established axonal tracts such as the ventral longitudinal tract of the forebrain and the ventral midbrain commissure (Fig. 1C). The presence of neurofilaments in these FDA-labeled fibers verified that they were axons (Fig. 1C, inset). Thus, in Litreated embryos, the progeny of VI. 1 occupy an abnormal location (head rather than trunk) and have an abnormal phenotype (neuron rather than epidermis). To estimate the extent of the fate changes in the ectodermal derivatives, and to determine whether less obvious changes in fate occurred in other organ systems, the relative contribution that each blastomere made to each organ system was given a numerical value, which was compared to values similarly derived from the normal fates of the same blastomeres. Fig. 2 illustrates several significant fate changes in each of the midline blastomeres. VI. 1 populated increased amounts of head and trunk epidermis, otocyst, cranial ganglia, midbrain, hindbrain and anterior spinal cord. Additionally, VI. 1 produced smaller amounts of hind- Lithium changes neural plate location 601 Fig. 1. (A) Examples of Li-treated embryos in which the trunk failed to develop (DAI 8 of Kao & Elinson, 1988). Each has a well-developed face and head including cement gland (c), stomqdeum (s), olfactory pits (o) and eyes (e). (B) A section through the level of the eye of a Li-treated embryo in which the VI.1 blastomere was injected with HRP. Many of the VI.1 progeny populate the forebrain (b) and retina (r). n, neurocoele; nt, notochord. (C) A fluorescence photomicrograph of a section through the midbrain of a Li-treated embryo in which VI. 1 was labeled with FDA. Many FDA-labeled axons are in the ventral commissure (arrowheads). The inset shows that these same fibers also are immunoreactive for neurofilament proteins. The arrowheads are in the same positions as those in the FDA micrograph, n, neurocoele; g, fifth cranial ganglion. (D) A control embryo of the same age as those in A (stage 41 of Nieuwkoop & Faber, 1964). (E) A section through the level of the eye of a control embryo in which VI.1 was labeled with HRP. No labeled progeny are present at this axial level. The dark spots are melanin granules, not HRP. r = retina; b, forebrain; n, neurocoele. gut. V2.1 (ventral vegetal midline cell) produced increased amounts of midbrain, hindbrain, head somite, foregut and anterior spinal cord, and reduced amounts of trunk epidermis, trunk neural crest, trunk somites, nephrotome and hindgut. Dl.l (dorsal animal midline cell) produced increased amounts of head epidermis, forebrain and midbrain, and reduced amounts of foregut and hindgut. D2.1 (dorsal vegetal midline cell), which normally produces very small amounts of ectoderm (Moody 1987a), did not alter its ectodermal fate (except for a reduced contribution to the greatly stunted Li-spinal cord). However, D2.1 produced smaller amounts of mesodermal and endodermal structures in both head and trunk regions. Dl.l normally contributes many progeny to the CNS, and after Li-treatment this contribution was increased significantly (Fig. 2). In contrast, VI. 1 and V2.1 normally contribute very few progeny to the CNS. The large increase in their CNS contribution after Li- treatment (Fig. 2) may have resulted either from extra cell divisions of the few ventral progeny that normally populate the CNS, or from an actual change in the fate of descendants that normally produce other phenotypes. To distinguish between these possibilities, we mapped the fate of the daughters of these two ventral midline blastomeres. In normal embryos, the equatorial daughters of VI.1 and V2.1 (VI.1.2 and V2.1.2, respectively) both contribute to CNS, whereas the polar daughters (VI. 1.1 and V2.1.1, respectively) have no CNS descendants (Moody, 19876). Specifically, VI. 1.1 has no descendants in the forebrain or midbrain and contributes an average amount of only 0-045 (i.e. a total of less than 5 cells in eleven embryos) to the hindbrain, and V2.1.1 has no descendants in any brain region (c.f. Dale & Slack, 1987). Following Li-treatment, the equatorial daughters increased their CNS progeny more than the polar daughters (Fig. 3). In fact, V2.1.2 was responsible for the entire change seen in V2.1's fate 602 S. L. Klein and S. A. Moody « m 3 -* o k— r^ 3 o o en rnu re ^ - rr o 3 a i-- au 3o to •— OJ m •o m i o « rT a L m > =5 en T re "D r^ 3 re i i --1 0 -2.0 a t— t _ Structures Fig. 2. Histograms showing the extent to which each structure of the Li-treated embryo contains progeny of the indicated blastomere. The diagram in the left corner of each graph shows the location of the injected blastomere (blackened) in a 16cell embryo with the animal pole up and dorsal midline to theright.The graphs show the contribution of the Li-blastomere minus the contribution of the normal blastomere (normal data from Moody, 1987a). Positive values indicate a greater than normal contribution by the Li-blastomere and negative values indicate a smaller than normal contribution by the Liblastomere. The dotted lines at +1 and at - 1 indicate significant differences from normal. The structures are arranged in roughly anterior (left) to posterior (right) order; the group of structures on the left of each graph is in the head and that on the right is in the trunk. The interface between the head and the trunk was considered to be the level of the otocyst. According to this criterion, DAI 7/8 embryos contain a small amount of anterior trunk structures. Gl, gland; Epi, epidermis; Olfactory, olfactory pit; Cr, cranial; D head Mes, dorsal head mesoderm; Br, branchial; Hd, head; Tr, trunk; Cd, cord; Lat Plate, lateral plate mesoderm. map, indicating that Li-treatment caused the progeny of V2.1 that normally populate the CNS to produce extra CNS. However, the polar daughter of VI. 1 (i.e. VI. 1.1) also made a large contribution to all parts of the CNS. This is a significant change in cell fate because VI. 1.1 normally produces only a few of the spinal cord's Rohon-Beard neurons (Moody, 19876; Moody, 1989), which are derived from the neural crest and not from the neural plate (Du Shane, 1938; Chibon, 1966). Thus, the fate changes of VI. 1.1 show that Li-treatment causes a non-CNS progenitor to produce CNS. Changes in ectodermal fates are correlated with changes in the pattern of gastrulation Because Li-treated ventral midline blastomeres changed their fate to become major neuronal progenitors, it seemed likely that Li perturbs a cell-cell communication that either directly influences cell fate (as suggested by Kao et al. 1986; Kao & Elinson, 1988; Slack et al. 1988; Busa & Gimlich, 1989), or that indirectly influences cell fate by altering the pattern of gastrulation such that novel cell populations, i.e. the progeny of ventral blastomeres, come under the influence of Spemann's Organizer (i.e. the presumptive chordamesoderm; Spemann & Mangold, 1924; Spemann, 1938). To distinguish between these possibilities, we studied the gastrulation movements of fluorescently labeled clones of Dl.l and VI. 1 in living Li-embryos that developed to DAI 8. In all cases, the migration of the superficial normal and Li-clones of each blastomere was indistinguishable. For example, the progeny of both normal and Li-Vl.l migrated ventrally so that at the end of gastrulation they each occupied the same mostly unilateral strip between the ventral tip of the animal pole and the ventral Up of the blastopore (Fig. 4A, B, F and summarized in Fig. 5). Lithium changes neural plate location 603 0.0 Structures Fig. 3. Differences from normal contributions of the daughters of VI. 1 (A) and V2.1 (B) to the Li-CNS (normal data from Moody, 19876). V2.1's equatorial daughter cell, V2.1.2, accounts for almost all of the change of fate seen for the mother cell (left panel in B), and Vl.l's equatorial daughter cell, VI.1.2, accounts for more of the change in cell fate seen for the mother cell (left panel in A). However, Vl.l's anterior daughter (VI.1.1) changes fate to produce greatly increased amounts of CNS. See the legend of Fig. 2 for abbreviations. Nevertheless, the processes of gastrulation and neurulation were retarded and abnormal in the Liembryo. Marking the blastopore lip with a spot of Nile blue sulfate toward the end of gastrulation (stages 11-5-12) labeled the posterior third to half of normal embryos (c.f. Keller, 1975), but labeled the rostral tip of the DAI 8 Li-head (data not shown). This result suggested that the Li-head formed around the blastopore, which normally becomes the cloaca (Fig. 5). This proposal was confirmed by observing neural plate formation in Li-embryos. First, a dark line formed around the yolk plug and extended to the equator, where the cement gland formed (Fig. 4C). The ectoderm within this line formed an ovoid plateau (Fig. 4G). The part of the plateau away from the cement gland contained a large number of VI. 1 progeny, and the side of the plateau near the cement gland contained a small VI.1 subclone (Fig. 4E, G). The part of the plateau near the cement gland contained many D l . l progeny. The edges of the plateau eventually raised up and fused to form a short tube with the cement gland at one end. The head subsequently formed from this region of the Li-embryo. The ectodermal plateau that formed around the blastopore, and contained many VI. 1 progeny (Fig. 4E, G), was obviously the Li-neural plate (c.f. Kao & Elinson, 1988). Fig. 5 summarizes the position of the fluorescent Dl.l and VI. 1 clones that we observed in living embryos at different stages of development. Examination of the interior of Li-embryos during gastrulation and neurulation demonstrated that the neural plate formed around the blastopore, instead of along the dorsal midline, because the chordamesoderm involuted abnormally. Although the involuting chordamesoderm began to migrate beneath the dorsal ectoderm as in the normal embryo, its leading edge stopped when it reached the equator (Figs4D, 6B, C); normally, the leading edge migrates to the animal pole (Fig. 6A; and Nieuwkoop & Faber, 1964). Then, the involuting ventral mesoderm, instead of migrating beneath the ventral ectoderm (Keller, 1975), joined the posterior chordamesoderm deep to the yolk plug (Fig. 7). The combined (i.e. dorsal and ventral) mesodermal finger moved into the center of the embryo to become the 'internal proboscis' described by Kao & Elinson (1988) (Figs 4D, E, 6B, C). The chordamesoderm folded onto itself; its anterior portion was directly beneath the dorsal vegetal ectoderm and its posterior portion was beneath the periblastoporal ectoderm and in the embryo's interior (Figs 4D, 6B, C). The anterior ventral mesoderm was in the proboscis and the posterior ventral mesoderm was beneath the periblasto- 604 S. L. Klein and S. A. Moody Fig. 4. (A, B) The vegetal hemisphere of the gastrula showing the location of the FDA-labeled VI.1 clone. Ventral is to the left. (A) Li-embryo, (B) normal embryo (stage 11-5), y, yolk plug. (C) The vegetal hemisphere of a Li-neurula showing the neural plate, delineated by the dark line, around the yolk plug (arrow). The cement gland (c) is at the rostral end of the line, and indicates the dorsum of the embryo. (D) The interior of the same embryo shown in C following a midsagittal dissection. The yolk plug (arrow in C) was lifted from the surface of the neural plate (between the arrowheads). The animal pole (an) is to the bottom of the figure and the cement gland, i.e. the dorsal side of the embryo, is to the right. The involuted chordamesoderm occupies two areas within the original vegetal hemisphere. Its leading edge (+) extends beneath the dorsal ectoderm about 90° from the yolk plug, and the remainder forms a portion of the 'internal proboscis' (bracket). The neural plate forms superficial to both areas of chordamesoderm and invaginates beneath the yolk plug to form the core of the proboscis. (E) Fluorescence photomicrograph of the same embryo shown in D. FDA-labeled VI.1 progeny are located in the ventral epidermis (e), in the ventral half of the internal proboscis (p), in the neural core of the proboscis (arrow) and in the neural plate (arrowheads). (F) A side view of a Li-neurula showing the location of the FDA-labeled VI. 1 clone. A large external yolk plug is at the top of the figure (between the arrowheads) and the cement gland (c) is to the right of the yolk plug. (G) A top view of the same embryo shown in F following the removal of the yolk plug. Beneath the yolk plug is the neural plate (border outlined by arrowheads) containing many FDA-labeled VI. 1 progeny. A small portion of the clone (arrow) is in the anterior part of the neural plate. poral ectoderm. The neural plate formed from the dorsal and ventral ectoderm that was superficial to the stalled, vegetal, chordamesoderm (Figs 4D, E, 6B, C). The posterior ventral mesoderm and some of the posterior chordamesoderm were located in the presumptive head. This altered pattern of gastrulation is summarized in Fig. 7. The neural plate formed from the ectoderm that was superficial to the chordamesoderm. The leading edge of the chordamesoderm denned the anterior extent of the neural plate in all embryos regardless of the extent of dorsoanterior enhancement. Neurulae in which the neural plate was of normal or near-normal length developed to DAI 5 or 6 (normal or stunted tail). In most of these neurulae, the chordamesoderm had migrated beneath the dorsal ectoderm from the dorsal lip of the blastopore to the anterior pole. In those neurulae in which chordamesodermal involution was incomplete, a small knob of mesoderm projected into the archenteron from the caudal pole. Conversely, neurulae with a round neural plate circumscribing the blastopore developed to DAI 9 or 10 (radially symmetric dorsoanterior). In these neurulae, the chordamesoderm had migrated only a small distance from the blastopore beneath dorsal ectoderm; the remainder of the chordamesoderm formed a large internal proboscis. Thus, the extent of dorsoanterior enhancement was inversely related to the amount of chordamesoderm beneath ectoderm, and was directly related to the size of the internal proboscis. Discussion The determination of embryonic cells probably occurs in several steps, and can be influenced by several extracellular cues. Among the interactions considered Lithium changes neural plate location 605 influence gastrulation, and that the altered pattern of gastrulation changes cell fate secondarily. Fig. 5. Diagram summarizing the gastrulation movements of the superficial members of the VI.1 (dots) and Dl.l (stippling) clones in normal and Li-treated embryos. In each embryo, the animal pole is to the left. The ectodermal progeny migrate normally during gastrulation so that they occupy a strip between the animal pole (left) and the lip of the blastopore (right), y, yolk plug. The members of each of the Li-clones that are near the blastopore populate the CNS because the neural plate (np, outlined by the thin internal line in the neurula) forms from the periblastoporal ectoderm. Some of the periblastoporal members of the VI. 1 clone migrate onto the dorsum of the neurula (dorsally located dots on the caudal part of the normal neurula and in the anterior neural plate of the Li-neurula). The Li-larva forms upside down; the vegetal members of the VI.1 and Dl.l clones that normally populate the caudal embryo populate the Li-head, and the animal members of the VI.1 and Dl.l clones that normally populate the head populate the caudal regions of the Li-embryo. The VI. 1 progeny that produce the dorsal epidermis of the normal larva are 'removed' to show the contributions of Dl.l to the deeper dorsal structures, i.e. brain and spinal cord. D, dorsal; V, ventral. to be influential are intercellular communications via growth factors, morphogens and/or inductors (Slack, 1983). It has been well documented that administration of Li during cleavage stages produces dramatic changes in embryogenesis (Kao et al. 1986; Breckenridge et al. 1987; Condie & Harland, 1987; Regen & Steinhardt, 1987; Kao & Elinson, 1988,1989; Cooke & Smith, 1988; Slack et al. 1988; Klein & King, 1988; Busa & Gimlich, 1989), and several laboratories have concluded that these changes indicate that cell fate is influenced directly by an early Li-sensitive event (Kao et al. 1986; Kao & Elinson, 1988, 1989; Slack et al. 1988; Busa & Gimlich, 1989). To test this proposal directly, we used lineage tracers to precisely compare the fates of individual identified blastomeres between normal and Li-treated embryos. We demonstrate that incubation in Li causes significant changes in cell fate, especially in ventral, midline blastomeres. The most dramatic changes were the gross overproduction, and novel expression, of CNS progeny. These changes in cell fate can be attributed to alterations in the pattern of gastrulation, which brought novel cell populations into contact during primary embryonic induction. Therefore, we propose that Li primarily perturbs cleavage-stage events that ultimately Changes in cell fate The DAI 7/8 embryos that were examined in the present study provide a very useful tool for cell fate studies because they primarily consist of only a properly organized head. The absence of body structures would result if the Li-treatment selectively killed the trunk progenitors or if the treatment caused trunk progenitors to produce head structures. Fate maps of 16-cell embryos can help to distinguish between these possibilities because the potential to make head versus trunk is largely segregated by this stage. Head structures are populated largely by dorsal blastomeres and trunk structures are populated largely by ventral blastomeres (Moody, 1987a). The fate maps show that Li causes the ventral blastomeres to produce significantly decreased amounts of trunk and significantly increased amounts of head structures. Additionally, continuous observation of fluorescent clones during gastrulation and neurulation detected no obvious cell death in over 95 % of the embryos. Therefore, Li-treatment appears to effect specific changes in cell fate, rather than cause the selective death of trunk progenitors. Because each frog blastomere gives rise to a very large number of progeny, and because their progeny reside in a several organs and spatial locations, their fates are difficult to quantify. Only a few cell populations consist of sufficiently small numbers to enable cell counting (e.g. Moody, 1989). Therefore, we could only estimate the changes in clone size in each organ by constructing semiquantitative fate maps of Li-embryos in a manner identical to that used for normal embryos (Moody, 1987a,fr). Comparison of the Li- and the normal semiquantitative fate maps show relative differences from normal cell fate. Because these differences are only estimates, they should be interpreted conservatively, and we consider only large changes (>1) to be significantly different from normal. In spite of these potential difficulties in quantification, many striking changes in cell fate were observed after Li-treatment. The 16-cell fate maps show that Li-treatment causes an animal hemisphere cell (Dl.l), which normally contributes large numbers of progeny to the CNS, to increase significantly its contribution to forebrain. Litreatment also causes two ventral blastomeres, which normally contribute ectodermal progeny only to epidermis and to small areas of the dorsal spinal cord, to significantly increase their contribution to brain and spinal cord. In fact, the anterior blastomere (VI. 1) becomes a major progenitor of neurons in the brain. These increases in CNS descendants may have resulted either from extra cell divisions of the normal neuronal progenitors (as in nematode reiterative lineage mutants; reviewed in Sternberg & Horvitz, 1984), or from the recruitment of extra CNS progenitors from blastomeres that do not normally populate the CNS (as in Drosophila neurogenic mutants; reviewed in CamposOrtega, 1988). We could not distinguish between these possibilities directly with the 16-cell fate maps, because 606 5. L. Klein and S. A. Moody Fig. 6. Midsagittal view of normal (A) and Li- (B and C) embryos at the end of gastrulation. Both normal and Liembryos are oriented the same way; the vegetal pole is to the left (y, yolk plug). (A) The leading edge of the normal chordamesoderm (+) migrates about 180°, to the animal pole (*), and the ectoderm between it and the yolk plug becomes the neural plate (dark ectoderm between arrowheads). The anterior thickening of the neural plate will form the brain. The anterior somites are segmented already (bracket). (B) The leading edge of the Lichordamesoderm (+) migrates about 90°, to the equator. The remainder of the Li-chordamesoderm moves into the center of the embryo (0, its leading edge). The neural plate is between the white arrowheads. The anterior part of the Li-neural plate forms from the ectoderm that is between the chordamesoderm's leading edge and the yolk plug (y), whereas the posterior part of the neural plate forms from the ectoderm on the ventral side of the yolk plug. The rostral somites are visible in the vegetal hemisphere (bracket). (C) An example of an older Li-embryo with a more extensive internal proboscis. The first few somites of the Li-embryo (bracket) form from the mesoderm in the vegetal dorsal quadrant. each of these blastomeres normally has a small, but demonstrable, contribution to the CNS. However, fate maps of the 32-cell stage can be used to make this distinction because at this stage some blastomeres normally have no CNS progeny (Jacobson & Hirose, 1981; Gimlich & Cooke, 1983; Jacobson, 1984; Dale & Slack, 1987; Moody, 19876,1989). Analyses of the CNS contribution by Li-treated 32-cell-stage ventral midline blastomeres demonstrate that the largest increase in CNS progeny descend from the equatorial daughters, i.e. from blastomeres that normally contribute at least a few cells to CNS. Thus, Li appears to cause those blastomeres that produce neurons normally to produce them in greater numbers, possibly by extra cell divisions. Additionally, Li causes at least one non-CNS progenitor to contribute progeny to the CNS. Normally, Vl.l.l's only ectodermal progeny are epidermis and neural crest (including the Rohon-Beard neurons, which are embedded in the dorsal roof of the neural tube, Moody, 19876); after Li-treatment, this blastomere produces large numbers of neurons throughout the brain and spinal cord. These fate changes represent a shift from the ventral to the dorsal ectodermal phenotype. They were demonstrable because the ectodermal derivatives display a very different phenotype (epidermis vs CNS) depending on their location in the tail bud embryo (ventral vs dorsal). Comparable changes in mesodermal and endodermal fates were not detected. However, because most mesoderm and endoderm display similar phenotypes in both ventral and dorsal locations, such changes may not be detectable by fate mapping. Thus, our fate maps do not enable one to conclude whether mesoderm and endoderm are shifted from ventral to dorsal locations. However, our analyses did demonstrate changes in the posterior-anterior locations of mesodermal and endodermal descendants. For example, at least one blastomere (V2.1) that normally contributes to only trunk mesoderm and endoderm (Moody, 1987a; Dale & Slack, 1987), populates these tissues in the head of the Li-treated embryo. Additionally, the Li-head was populated by blastomeres (Dl.l and VI. 1) that normally contribute to these tissues in both the head and the trunk. This change in posterior-anterior position of mesoderm and endoderm may represent a specific alteration in cell fate (Cooke & Smith, 1988), but more probably results from the reorganization of the anterior-posterior axis of the Li-embryo that occurs because the reduction in the extent of gastrulation Lithium changes neural plate location Fig. 7. Diagram illustrating the formation of the internal proboscis. The midsagittal plane of the embryo is depicted and oriented as in Fig. 6. Asterisk indicates the animal pole. The mesoderm is blackened; the chordamesoderm is originally on the dorsal side (top in A) and the ventral mesoderm is originally on the ventral side (bottom in A). (A,B) Li-gastrulation begins normally. The leading edge of the chordamesoderm (+) migrates beneath the dorsal ectoderm. (C) The leading edge of the chordamesoderm stops when it reaches the equator and the rest of the chordamesoderm (O) moves into the interior of the embryo. (D) The posterior chordamesoderm unites with the ventral mesoderm beneath the yolk plug (y), and the combined mesodermal finger moves into the center of the embryo. (E) At the end of gastrulation, the chordamesoderm is folded onto itself. The anterior portion of the chordamesoderm is directly beneath the dorsal vegetal ectoderm and the posterior portion is beneath the periblastoporal ectoderm and in the embryo's interior. The anterior portion of the ventral mesoderm is in the proboscis and the posterior portion is beneath the periblastoporal ectoderm. (F) The neural plate (np) forms from the ectoderm that is superficial to mesoderm. c, cement gland. causes the head to form in the vegetal hemisphere. Because some of the 'posterior' mesoderm and endoderm also are located in the vegetal hemisphere at the end of gastrulation, it differentiates into head, rather than trunk, structures (i.e. V2.1 produces less hindgut, nephric tubules and trunk somite, and produces more foregut and head somite; see Fig. 2). Thus, our data suggest that this change in endodermal and mesodermal 607 fate is the secondary result of alterations in the pattern of gastrulation. In addition to changes in the axial position of mesodermal and endodermal descendants, there is evidence that some mesodermal and endodermal progeny are lost from the Li-treated embryo. Most of the loss seems to result from the failure of the cells to develop rather than to cell death. As mentioned above, extensive cell death was not detected. In less than 5 % of the embryos, the yolk plug remains external after gastrulation (e.g. Fig. 4F), and falls off at hatching; in only this small proportion of embryos does cell death obviously contribute to the reduction in trunk structures. Conversely, nearly all blastomeres show significant decreases in hindgut progeny, and D2.1 shows large decreases in mesodermal and endodermal progeny without any corresponding increases in other organs. Possibly some of the D2.1 stem cells die, or they do not complete the normal number of cell divisions. From these changes in cell fate, it appears that the trunk of the Li-embryo fails to develop due to a combination of the following: (1) the trunk mesodermal and endodermal progeny of V2.1 populate head, rather than trunk, structures, (2) the hindgut progeny of VI.1 and Dl.l fail to differentiate, and (3) some of the mesodermal and endodermal progeny of D2.1 either die or divide less than normal. It has been proposed that the Li-phenotype results because Li causes ventral cells to behave as Spemann's Organizer (Kao & Elinson, 1988). However, our fate maps show that none of the midline blastomeres produces detectably increased amounts of the Organizer's differentiated tissues (notochord or somite). Although V2.1 produces more head somite, it produces correspondingly less trunk somite, suggesting a change in axial position rather than a change in phenotype. In contrast, fate maps of three cases of radially symmetric dorsoanterior embryos (DAI 9/10) demonstrate that V2.1 produces significantly more notochord than normal (data not shown), supporting the conclusion of Kao & Elinson (1988) that the ventral marginal zone may behave as an additional source of Spemann's Organizer in DAI 9/10 Li-embryos. Effects on gastrulation movements The dramatic changes in cell fate that were caused by Li-treatment during cleavage stages may indicate that Li alters directly a cleavage-stage event that normally influences cell fate. That is, Li may cause cells to 'transfate' (e.g. as in leech embryos, Weisblat & Blair, 1984). Alternatively, Li may change the spatial relationship between the ectoderm and the neural inductive chordamesoderm during gastrulation. The latter seemed the more likely possibility, especially because ventrally transplanted presumptive chordamesoderm causes ventral midline blastomeres to produce CNS (Gimlich & Cooke, 1983; Jacobson, 1984). Continuous observation during gastrulation showed that the superficial ectodermal cells of the Li-treated VI. 1 clone migrate normally. All of the cells of the VI. 1 clone that move migrate toward the ventral lip of the 608 S. L. Klein and S. A. Moody blastopore; none of them migrate onto the dorsal side of the embryo, into possible contact with a normally located inductor. However, direct observation and vital dye marking showed that the neural plate of the Liembryo forms ectopically. VI. 1 progeny are in the neural plate, not because they migrate improperly, but because the neural plate forms in the vegetal hemisphere, in part, from the ventral ectoderm that normally contains VI. 1 progeny. Previous studies also have shown that the neural tissue of Li-treated embryos forms near the blastopore (Kao & Elinson, 1988; Cooke & Smith, 1988). The ectopic neural plate is coincident with ectopic chordamesoderm, which also is confined to the vegetal hemisphere. Although chordamesodermal involution begins normally, the leading edge of the chordamesoderm stops at the equator. The remainder of the chordamesoderm and the adjacent ventral mesoderm subsequently move into the interior of the embryo. Thus, at the end of gastrulation, the anterior portion of the chordamesoderm is deep to the dorsal vegetal ectoderm, and the posterior portion is beneath the periblastoporal (i.e. dorsal and ventral) ectoderm and in the internal proboscis (summarized in Fig. 7). The neural plate forms from the vegetal ectoderm that is superficial to both regions of chordamesoderm; some of this ectoderm derives from dorsal blastomeres and some from ventral blastomeres. The ventral periblastoporal ectoderm, which includes VI. 1 and V2.1 progeny, is induced to form the posterior neural plate (Figs 4C-G and 5), which probably differentiates into the spinal cord, hindbrain and midbrain. The dorsal periblastoporal ectoderm, which includes D l . l progeny, is induced to form the anterior neural plate (Fig. 5), which probably differentiates into the rest of the midbrain and the forebrain. This novel location of the neural plate accounts for the observation that VI. 1 populates more of the caudal CNS than normal, whereas Dl.l populates more of the rostral CNS than normal (see Fig. 2). The periblastoporal location of the neural plate also explains why VI. 1 and V2.1 produce increased amounts of spinal cord, whereas D2.1 produce decreased amounts. A normal secondary migration probably relocates some of the VI. 1 progeny into the forebrain and retina. Members of the VI. 1 clone migrate from the posterior neural plate, around the blastopore and into the anterior neural plate on the dorsal side of the embryo (Figs 4G, 5). In normal embryos, some ventral cells migrate along this route at the end of gastrulation to populate dorsal epidermis (Fig. 5; Keller, 1975). The normal occurrence of this secondary migration further suggests that the Li-treatment alters the involution of the mesoderm without changing the movement of ectodermal cells. The reason that the posterior chordamesoderm moves into the interior of the embryo, rather than involuting beneath the dorsal ectoderm, may indicate the primary action of Li-treatment upon gastrulation movements. A lot of recent evidence shows that the migration of the ventral mesoderm of Li-treated em- bryos resembles that normally seen in dorsal mesoderm. The ventral mesoderm of Li-treated embryos undergoes precocious involution so that the normal dorsal-ventral asymmetry of involution is reduced (Regen & Steinhardt, 1988). Additionally, ventral mesoderm explanted from Li-treated embryo becomes elongated as the cells undergo the movements of convergent extension normally seen only in dorsal mesoderm (Regen & Steinhardt, 1988; Slack etal. 1988; Kao & Elinson, 1989). Considering that both dorsal and ventral mesoderm of Li-treated embryos elongate, it seems paradoxical that these embryos are stunted. The reduction in overall length reflects the fact that the chordamesoderm is folded onto itself. Therefore, the folding of the chordamesoderm may be largely responsible for producing the morphology of the Li-treated embryo. It may be that the Li-treatment also prevents the leading edge of the chordamesoderm from entering the animal hemisphere so that the posterior chordamesoderm is passively pushed into the interior by the normal convergent extension of the dorsal cells (Keller et al. 1985). Alternatively, the abnormal migratory properties of the ventral mesoderm may cause it to adhere to the posterior chordamesoderm and pull the chordamesoderm into the interior of the embryo. The crucial role of ventral mesoderm is also indicated by the observation that injections of low doses of Li into ventral blastomeres cause dorsoanterior enhancement (Kao etal. 1986; Kao & Elinson, 1989; Busa & Gimlich, 1989). • In addition to acquiring the migratory properties of dorsal mesoderm, the ventral mesoderm of Li-treated embryos apparently acquires the ability to act as the Organizer, because transplanted ventral marginal zones from Li-treated gastrulae induce axis formation in u.v.irradiated axis-deficient embryos (Kao & Elinson, 1988). However, our fate maps show that ventral midline blastomeres of DAI 8 embryos do not produce increased amounts of the Organizer's differentiated tissues (see above). It may be that some of the chordamesoderm in the internal proboscis is sufficiently close to the ventral ectoderm to induce it to enter the CNS, either by itself or in combination with the ventral mesoderm. Regardless of whether the inductive signal is derived from the dorsal and/or ventral mesoderm, the reason that the CNS derives from only vegetal ectoderm is that it is the only ectoderm that comes into contact with the mesoderm. These observations suggest that the Li-fate map differs from normal because Li alters the involution of the chordamesoderm and the ventral mesoderm. The ectodermal fates are shifted toward head formation because the vegetally stalled mesoderm induces the formation of the nervous system from the vegetal portion of the ectoderm on both the dorsal and ventral sides of the blastopore. Accordingly, the dorsal and ventral ectoderm of the vegetal hemisphere produce the CNS, whereas the dorsal and ventral animal ectoderm, which does not come into contact with the chordamesoderm (Fig. 4D), produces the epidermis of the remainder of the body (e.g. Li-Vl.l produced much more Lithium changes neural plate location 609 trunk epidermis than normal, whereas Li-V2.1 produced much less trunk epidermis than normal, Figs 2 and 5). These changes represent both phenotypic and spatial alterations in the ectodermal fates. The observation that ventral cells are competent to populate CNS is not novel. Previous studies have demonstrated that transplantations that bring ventral ectoderm into contact with presumptive chordamesoderm cause this ectoderm to produce nervous system (Spemann & Mangold, 1924; Spemann, 1938; Gimlich & Cooke, 1983; Jacobson, 1984). However, in the present study, the manipulation that brought the ventral ectoderm into contact with the inductive mesoderm was performed hours before gastrulation, and consisted only of brief exposure to Li. This result shows that the pattern of gastrulation is influenced by a Li-sensitive event that occurs soon after fertilization. This event may be involved in mesoderm induction, as suggested by Kao & Elinson (1988, 1989) and by Slack et al. (1988), and may then affect neural induction secondarily. Action of lithium Many biochemical studies have shown that Li blocks the inositol trisphosphate-diacylglycerol second messenger pathway by inhibiting the conversion of inositol phosphate to inositol (Sherman et al. 1981; Berridge et al. 1982; Berridge, 1987). This pathway transduces extracellular signals to intracellular calciumfluxesand, thus, to changes in cell physiology. The observation that Li alters pattern formation suggests that cleavage-stage intercellular communications that are mediated by the inositol trisphosphate pathway normally play a role in later morphogenesis. In fact, recent studies have shown that the detrimental effects of Li on embryogenesis are reversed by injecting inositol into Li-treated embryos (Busa & Gimlich, 1989). Additionally, Li reduces the production of particular proteins that are synthesized during cleavage stages (Klein & King, 1988). Thus, a Lisensitive cleavage-stage event, which may involve intercellular communication, may normally modulate the physiological, and thus developmental, state of individual blastomeres. The present study shows that this event influences the pattern of mesodermal involution and thus ultimately influences the induction of the nervous system. On-going examinations are expected to reveal the nature of these cleavage-stage interactions and the mechanism by which they influence gastrulation movements. We acknowledge the excellent technical assistance of Daniel Best and Kathryn Kersey. Supported by NIH grants HD23324 (SLK), and NS 23158 (SAM). References BERRIDGE, M. J. (1987). Inositol trisphosphate and diacylglycerol: Two interacting second messengers. In A. Rev. Biochem., vol. 56 (ed. C. C. Richardson, P. D. Boyer, I. B. Dawid & A. Meister). pp. 159-193. Annual Reviews, Palo Alto, Calf. BERRIDGE, M. J., DOWNES, C. P. & HANLEY, M. R. (1982). Lithium amplifies agonist-dependent phosphatidylinositol responses in brain and salivary glands. Biochem. J. 206, 587-595. BRECKENRIDGE, L. J., WARREN, R. L. & WARNER, A. E. (1987). Lithium inhibits morphogenesis of the nervous system but not neuronal differentiation in Xenopus laevis. Development 99, 353-370. BRIGGS, G. G., FREEMAN, R. K. & YAFFE, S. J. (1986). Drugs in Pregnancy and Lactation: A Reference Guide to Fetal and Neonatal Risk. Williams and Wilkins, Baltimore, MD. BUSA, W. B. & GIMLICH, R. L. (1989). Li+-induced teratogenesis in frog embryos prevented by a polyphosphoinositide or diacylglycerol analogue. Devi Biol. 132, 315-324. CAMPOS-ORTEGA, J. A. (1988). Cellular interactions during early neurogenesis of Drosophila melanogaster. Trends in Neuroscience 11,400-405. CHTBON, P. (1966). Analyse experimental de la re"gionalisation et des capacity morphog6n€riques de la crfite neurale chez Pamphibien urodele Pleurodeles waltlii. Michah. Mem. Soc. Fr. Zool. 36, 1-107. CONDIE, B. G. & HARLAND, R. M. (1987). Posterior expression of a homeobox gene in early Xenopus embryos. Development. 101, 93-105. COOKE, J. & SMITH, E. J. (1988). The restrictive effects of early exposure to lithium upon body pattern in Xenopus development, studied by quantitative anatomy and immuno- fluorescence. Development 102, 85-99. DALE, L. & SLACK, J. M. W. (1987). Fate map for the 32-cell stage of Xenopus laevis. Development 99, 527-551. Du SHANE, G. P. (1938). Neural fold derivatives in the amphibian pigment cells, spinal ganglia and Rohon-Be'ard cells. /. exp. Zool. 78, 485-503. GIMUCH, R. L. & COOKE, J. (1983). Cell lineage and the induction of second nervous systems in amphibian development. Nature, Lond. 306, 471-473. JACOBSON, M. (1984). Cell lineage analysis of neural induction: Origins of cells forming the induced nervous system. Devi Biol. 102, 122-129. JACOBSON, M. & HIROSE, G. (1981). Clonal organization of the central nervous system of the frog. II. Clones stemming from individual blastomeres of the 32- and 64-cell stages. /. Neurosci. 1, 271-284. KAO, K. R. & ELINSON, R. P. (1988). The entire mesodermal mantle behaves as Spemann's Organizer in dorsoanterior enhanced Xenopus laevis embryos. Devi Biol. 127, 64-77. KAO, K. R. & ELINSON, R. P. (1989). Dorsalization of mesoderm by lithium. Devi Biol. 132, 81-90. KAO, K. R., MASUI, Y. & ELTNSON, R. P. (1986). Lithium-induced respecification of pattern in Xenopus laevis embryos. Nature, Lond. 322, 371-373. 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. Devi Biol. 42, 222-241. KELLER, R. E., DANILCHIK, M., GIMUCH, R. & SHIS, J. (1985). The function and mechanism of convergent extension during gastrulation of Xenopus laevis. J. Embryol. exp. Morph. 89 Supplement, 185-209. KIRSCHNER, M. W. & HARA, K. (1980). A new method for local vital staining of amphibian embryos using ficoll and 'crystals' of rule red. Mikroskopie 36, 12-15. KLEIN, S. L. & KING, M. L. (1988). Correlations between cell fate and the distribution of proteins that are synthesized before the midblastula transition in Xenopus. Roux's Arch, devl Biol. 197, 275-281. MOODY, S. A. (1987a). Fates of the blastomeres of the 16-cell stage Xenopus embryo. Devi Biol. 119, 560-578. MOODY, S. A. (1987ft). Fates of the blastomeres of the 32-cell stage Xenopus embryo. Devi Biol. 122, 300-319. MOODY, S. A. (1989). Quantitative lineage analysis of the origin of frog primary motor and sensory neurons from cleavage stage blastomeres. /. Neurosci. (In Press). NiEUWKoop, P. D. & FABER, J. (1964). Normal Table o/Xenopus (Daudin). Amsterdam: North-Holland. REGEN, C. M. & STEINHARDT, R. A. (1988). Lithium dorsalizes but also mechanically disrupts gastrulation of Xenopus laevis. Development 102, 677-686. 610 S. L. Klein and S. A. Moody SHERMAN, W. R., LEAVTIT, A. L., HONCHAR, M. P., HALLCHER, L. M. & PHILLIPS, B. E. (1981). Evidence that lithium alters phosphoinositide metabolism: Chronic administration elevates primary D-myo-inositol-1-phosphate in cerebral cortex of the rat. ]. Neurochem. 36, 1947-1951. SLACK, J. M. W. (1983). From Egg to Embryo, Developmental and Cell Biology Series 13 Cambridge University Press, Cambridge. SLACK, J. M. W., ISAACS, H. V. & DARLINGTON, B. G. (1988). Inductive effects of fibroblast growth factor and lithium ion on Xenopus blastula ectoderm. Development 103, 581-590. SPEMANN, H. (1938). Embryonic Development and Induction Hafner, New York (reprinted 1967). SPEMANN, H. & MANGOLD, H. (1924). Uber induktion von embryonalanlagen durch implantation artfremder organisatoren. Arch. Mikr. Anal. EntwMech. 100, 599-638. STERNBERG, P. W. & HORVITZ, H. R. (1984). The genetic control of cell lineage during nematode development. A. Rev. Genet. 18, 489-524. WEISBLAT, D. A. & BLAIR, S. S. (1984). Developmental indeterminacy in embryos of the leech Helobdella triserialis. Devi Biol. 101,326-335. (Accepted 14 April 1989)