Suramin prevents transcription of dorsal marker genes in Xenopus laevis embryos, isolated dorsal blastopore lips and activin A induced animal caps

Mechanisms of Development, 43 (1993) 121-133

121

© 1993ElsevierScientificPublishers Ireland, Ltd. 0925-4773/93/$06.00 MOD 00181

Suramin prevents transcription of dorsal marker genes in Xenopus laevis embryos, isolated dorsal blastopore lips and activin A induced animal caps Ralf Oschwald a Joachim H. Clement a Walter Kn6chel a and Horst Grunz b a Abteilung Biochemie, Universitiit Ulm, Oberer Eselsberg, Ulm, Germany and b Universitiit GH Essen, FB 9 (Biologic), Abteilung Zoophysiologie, Essen, Germany

(Received22 April 1993;revisionreceivedand accepted28 June 1993)

Suramin, a polyanionic compound which is known to interact with the receptors of growth factors inhibits the expression of dorsal marker genes in whole embryos and isolated dorsal blastopore lips. Suramin also prevents activin A induced dorsalization of animal cap explants from blastula stage embryos, but it simultaneously evokes a shift of the differentiation pattern from dorsal mesodermal structures (notochord, somites) to ventral mesodermal derivatives (mesothelium and erythroid precursor cells). The results are consistent with the assumption that the dorsal vegetal zone (Nieuwkoop center) primarily releases more general/ventral mesodermalization signals. They further suggest a dual role of activin A in early embryogenesis. While the maternal component may contribute to a more general/ventral type of induction, increasing concentrations of the zygotic component along with the activation of primary response genes may contribute to the dorsalization of the organizer. Suramin; Mesoderm induction; Animal cap; Activin A; XFD-1

Introduction

Suramin is a polysulphonated compound which has clinically been used in treatment of African trypanosomis and onchocerciasis. It prevents the binding of many growth factors to cell surface receptors including members of the FGF, TGF-/3 and Wnt families (Betsholtz et al., 1986; Coffey et al., 1987; Papkoff and Schryver, 1990; Chakrabarti et al., 1992; Pinedo and Van Rijswijk, 1992). Results from extensive research during past years now clearly suggest that different members of these multigene families are main actors in the concert of interacting signals leading to the formation and regional specification of mesoderm in the amphibian embryo (reviewed in Kn6chel and Tiedemann, 1989; Smith, 1989; Slack et al., 1989; Jessel and Melton, 1992; Sive, 1993). While members of the FGF family and BMP-4 mainly contribute to ventral mesoderm, activin A has been shown to induce dorsal structures like notochord and somites. It was also shown

Correspondence to: W. Kn6chel, Abt. Biochemie, Universit~it Ulm, Oberer Eselsberg, 7900 Ulm, Germany.

that suramin applied to blastula stage embryos inhibits convergent extension movements during gastrulation and that resulting embryos are completely lacking anterior structures (Gerhart et al., 1989). Previous experiments revealed that suramin shifts the differentiation of the organizer from dorsal mesoderm (notochord) into heart structures (Grunz, 1992). It is generally accepted that the dorsal blastopore lip (Spemann's organizer) is formed by the interaction between the dorsal vegetal zone (the so-called Nieuwkoop center) and the neighbouring animal cap area. During gastrulation the organizer differentiates into head mesoderm, notochord and somites. The chordamesoderm in turn causes the neuralization of the overlaying neuroectoderm resulting in the formation of the central nervous system. Furthermore, it was shown that notochord formation in isolated dorsal blastopore lip can be prevented by suramin up till the middle gastrula (Grunz, 1993). From these results it was concluded that primarily a more general or ventral mesodermal signal is transferred from the dorsal vegetal zone. The dorsalization, which enables the blastopore lip to differentiate into head mesoderm and notochord and in turn to acquire neuralizing activity, takes

122 place during gastrulation. If activin A or a closely related homologue acts as the molecular trigger for this dorsalization event, these molecules should be effectively inhibited by suramin. Indeed, a suramin caused inhibition of dorsal mesoderm formation in recombinants of animal and vegetal explants and also in activin A induced animal cap explants has been reported (Slack, 1991). Results from our present study extend this observation, i.e., by using other experimental conditions (higher concentrations of the inducer, longer times of culturing) there is primarily a shift from dorsal to ventral mesodermal induction. Closely related with the determination of the organizer into dorsal mesodermal structures (notochord, head mesoderm and somites) is the activation of a number of genes in the dorsal blastopore lip at the beginning of gastrulation, Such early response genes including goosecoid (Cho et al., 1991), Xlim-1 (Taira et al., 1992) and XFD-1 (Kn6chel et al., 1992) (also described as XFKD-1 (Dirksen and Jamrich, 1992) and pintallavis (Ruiz i Altaba and Jessel, 1992)) are selectively expressed in the amphibian organizer area and are thought to be important factors for the establishment of the dorsal axis of the embryo. XFD-1 is a fork head related gene which is activated at the blastula stage. Its transcripts are localized in a rather thin stripe of ceils in the dorsal area of the blastopore lip. During gastrulation and neurulation the gene is transcribed within forming notochord and the neural floor plate. In the animal cap assay the gene is activated by incubation with activin A but not with bFGF. Therefore, in the present study we used different marker genes, amongst them XFD-1, to study the effect of suramin on the development of whole embryos, isolated dorsal blastopore lips and of animal cap explants induced by activin A. We show that suramin prevents the expression of XFD-1 concomitant with a shift from dorsal to ventral mesodermal structures,

Results

Suramin interferes with dorsal mesoderm induction and completely inhibits the formation of anterior neural structures in vivo To study the effects of suramin treatment on whole embryos in vivo we have exposed early blastula stage

embryos for a limited time period to varying concentrations of suramin. After untreated control embryos had developed to stage 12 or 32, respectively, embryos were fixed and analyzed by whole mount in situ hybridization for the expression of four different marker genes (Fig. 1). We used: (1) XFD-1, a forkhead related Xenopus gene which is activated after midblastula transition and transcribed in notochord and in neural floor plate of gastrula and neurula stage embryos (Kn6chel et al., 1992; Dirksen and Jamrich, 1992; Ruiz i Altaba and Jessell, 1992); (2) a neural specific class II /3-tubulin being activated in neuroectoderm during gastrulation (clone 24-10; Richter et al., 1988; Good et al., 1989; Oschwald et al., 1991); (3) a mesoderm specific, sarcomeric a-cardiac actin a l (Stutz and Spohr, 1986) co-expressed in heart and skeletal muscle during embryogenesis and (4) embryonic a-globin (pXGL 19.1 (Widmer et al., 1981) identical to aT3 (Banville and Williams, 1985)) being initially expressed in blood islands derived from ventral mesoderm. Fig. lc shows that 50 tzM suramin treatment does not detectably interfere with XFD-1 expression at the posterior dorsal axis. In contrast, high concentrations of suramin (500/zM) lead to a suppression of XFD-1 expression, although the embryos still exhibit a rather normal morphological appearance and show a roundish blastoporus (Fig. ld). At stage 32, suramin treated embryos are completely lacking anterior structures, i.e., they are morphologically devoid of heads and eyes (Fig. lh,l,p). Nevertheless, at 50 /~M suramin concentration both the CNS specific/3-tubulin and the a-actin genes are normally expressed at the posterior dorsal axis, i.e., we observe spinal cords and a segregation of somites (Fig. lg,k). This correlates with the early expression of XFD-1 and its posterior versus anterior dominance. As shown by/3-tubulin expression, no anterior neural structures are visualized (Fig. lg). 500 p~M suramin treatment inhibits in most cases axis formation resulting in a lack of/3-tubulin and for a great part also of a-actin gene expression (Fig. lh,1). This reflects that posterior dorsal mesoderm formation is also inhibited and, again, correlates with the early suppression of the XFD-1 gene. However, there is absolutely no suppression of the a-globin gene; on the contrary, we often observed an increased transcription of this gene in otherwise totally deformed embryos

Fig. 1. Whole mount in-situ hybridization of embryos ( - / + suramin). (a-d) XFD-1; (e-h)/3-tubulin; (i-l) a-actin; (m-p) a-globin. Embryos in (a-d) are stage 12, all other embryos are stage 32/33. (a,b,e,f,i,j,m,n) show control embryos, (c,g,k,o) show embryos treated with 50 tzM Suramin and (d,h,l,p) show embryos treated with 500 ~M Suramin. Suramin treatment started at stage 7 and lasted for 5 h. Embryos were washed and raised in 0.75 × Steinberg solution until the desired developmental stage.

123

d~

qD

mm

•

OI

124 (Fig. lp). This was substantiated by RNase protection experiments using E F - l a (PSting et al., 1990) as internal control (data not shown). The ratio of signal intensities obtained after autoradiography with o~-globin and E F - l a probes was twice as high in embryos treated with 500 ~M suramin as compared to that from untreated embryos. In summary, these experiments demonstrate that suramin interferes with dorsal but not with ventral mesoderm formation. This especially holds true for the most anterior located part of involuted dorsal mesoderm, the prechordal plate, which is absent in suramin treated embryos but which is normally required for the induction of anterior neural structures.

onic a-globin gene (Fig. 21) which would be indicative for a complete ventralization of the dorsal lip. This means, that at stage 10 when dorsal lips were excised, this tissue is already determined to an extent which excludes re-programming of cells to the most ventral type, i.e., formation of blood cells.

Suramin shifts the mesoderm inducing activity of EDF in isolated animal caps from dorsal to ventral mesoderm derived tissues

Suramin interferes with the fate of dorsal blastopore lip

To study the ventralizing effect of suramin at the initial phases of mesoderm formation we have induced animal cap explants excised from stage 8 embryos with recombinant human erythroid differentiation factor (being identical with activin A (Murata et al., 1988)) in

In the light of recent reports about changes of the fate of Spemann's organizer by suramin treatment (Grunz, 1992), we next have analyzed the expression of the four above mentioned marker genes in isolated dorsal blastopore lips which normally develop to chordamesoderm and brain structures. In untreated explants we observe 5 h after excision by whole mount in situ hybridization the expression of XFD-1. As within the control embryo, only those cells forming the dorsal midline and giving rise to notochord show transcription of this gene (Fig. 2a,c). After culturing the lips together with control embryos up to stage 37, we observe expression of a-actin and /3-tubulin but not of the aglobin gene (Fig. 2d,g,j). This reflects the development of chordamesoderm with secondary induction of neuroectoderm to neural structures. Exposure of the isolated dorsal lips to suramin suppresses transcription of all marker genes being expressed in these structures in a time dependent manner. Moreover, XFD-1 which is already strongly expressed at early gastrula, when dorsal lips are excised, is completely deactivated, when controls are at late gastrula, and the transcripts being present at early gastrula are totally degraded (Fig. 2b). a-Actin and /3-tubulin gene expression is decreased at longer exposure time to suramin (Fig 2e,f,h,i). These results fully support the recently reported change of the Spemann's organizer from dorsal to ventral tissue (Grunz, 1992). Since cardiac and skeletal muscle actin genes are not differentially expressed in the early embryo, it is possible that oz-actin gene expression observed at treatment with intermediate suramin concentrations reflects formation of the cardiac muscle. The previously described rhythmic contractions together with the histological identification of heart muscle structures (Grunz, 1992)would, indeed, agree with this assumption. However, even at longest exposure to suramin we did not observe an activation of the embry-

the absence or presence of suramin. It is well known that activin A is a potent mesoderm inducing substance when used in the animal cap assay. At very low concentrations (below 1 ng/ml) it preferentially induces ventral tissues but exceeding different treshold concentrations there is a complete shift of induced tissues towards dorsal structures including a dorsal axis with notochord and somites. Thus, at a concentration of 75 ng/ml we morphologically observe a typical elongation of induced explants (Fig. 3a,c) and histological analysis demonstrates the formation of notochord and somites (Fig. 4a,b). The morphological appearance of induced explants which have been treated with suramin is completely different (Fig. 3b,d). They do not elongate and more closely resemble explants which have been induced with bFGF, i.e., they form less pigmented bubble-like structures. Furthermore, a histological analysis revealed that these explants did not contain notochord but they frequently showed mesothelium lined cavities filled with roundish cells which have the typical appearance of immature red blood cells (Fig. 4c,d). Formation of these structures is clearly indicative for ventral mesodermalization. So it can be concluded that suramin does not completely inhibit the action of activin but changes the type of induced tissue from dorsal to ventral mesoderm derived structures. These findings were further substantiated by whole mount hybridization studies using the above mentioned marker genes. While neither of these genes is expressed in control explants in the absence of mesoderm inducers (data not shown), all the four genes are transcribed in EDF induced animal caps (Fig. 5). Previous RNase protection and Northern blot studies have already shown that XFD-1 is activated by activin A (Kn6chel et al., 1992; Dirksen and Jamrich, 1992; Ruiz i Altaba and Jessell, 1992); we here demonstrate that transcripts of this gene can also be visualized by whole mount in situ hybridization of isolated caps (Fig. 5a). Expression of

125

a

d

e

f

g

h

i

I

1

t

Fig. 2. Whole mount in situ hybridization of dorsal blastopore lips ( - / + suramin). (a-c) XFD-1; (d-f) fl-tubulin; (g-i) a-actin; (j-l) c~-globin. All lips were excised at stage 10 using fine glass needles. Lips in (a) and (b) were cultivated for 5 h until control embryos (c) reached stage 12; other lips were cultivated for 3 days until control embryos reached stage 37. (a,d,gj) untreated blastopore-lips (controls); (b) blastopore lip raised in Barth-solution containing 150 IxM suramin for 5 h; (e,h,k) lips treated for 2 h with suramin; (f,i,1) lips treated for 6 h with suramin. For suramin-treatment lips were raised for 1.75 h in Barth solution containing 300 tzM suramin; for the rest of the time up to 6 h the lips were transferred to 150 IxM suramin.

a-actin reflects the formation of solnites and that of /3-tubulin is probably due to secondary neural inductions between induced dorsal mesoderm and still not

finally determined ectodermal cells (Fig. 5f, d). Expression of embryonic a-globin is characteristic for red blood cell formation (Fig. 5h), but our hybridization

126

Fig. 3. Morphology of E D F induced animal caps ( - / + suramin). (a) animal caps incubated for 4.5 h in 75 n g / m l E D F and grown for a total of 24 h. (b) animal caps simultaneously treated for 4.5 h with 75 n g / m l E D F and 150/~mol suramin. They were incubated for additional 3 h in 50 p~mol suramin without EDF, transferred to suramin free solution and grown for a total of 24 h. (c) animal caps treated as in (a) after 96 h. (d) animal caps treated as in (b) after 96 h. Bars: 0.5 m m (A, B) or 1 m m (C, D).

127 r e s u l t was r a t h e r u n e x p e c t e d , since we a n d o t h e r s h a d o r i g i n a l l y f a i l e d in t h e d e m o n s t r a t i o n o f h e m a t o p o i e t i c cells in i n d u c e d a n i m a l caps by u s i n g e m b r y o n i c /3-

A

g l o b i n a n t i s e n s e p r o b e s or g l o b i n a n t i b o d i e s as m o l e c u lar m a r k e r s ( K n 6 c h e l et al., 1989; G r e e n et al., 1990). T h u s , a l t h o u g h t h e m o r p h o l o g i c a l a p p e a r a n c e was

13

so

C

D

Fig. 4. Histological sections of EDF induced animal caps ( - / + suramin). (a,b) sections of animal caps as shown in Fig. 3c. Both explants clearly show the formation of notochord, although at different amounts. (c,d) sections of animal caps as shown in Fig. 3d. The sections show mesothelium lined cavities with blood like cells but no notochord. Bar: 0.1 mm. n: neural, bl: blood like cells, no, notochord, so: somites.

128 clearly indicative for hematopoietic progenitor cells (Grunz, 1983), it was generally questioned whether these cells represent red blood cell precursors. However, the localized expression of the embryonic a-globin gene in whole embryos and the demonstration of corresponding transcripts in cells of E D F induced explants now clearly suggests, that these cells are at least determined as hematopoietic cells, even if their further differentiation including the expression of embryonic fl-globin chains may require additional factors. When explants were induced in the presence of suramin, we failed to detect XFD-1 transcripts by whole mount hybridizations (Fig. 5b). It may be argued that the sensitivity of this method is not sufficient to detect minor quantities of corresponding transcripts. However, RNase protection experiments with RNAs extracted from induced animal caps using XFD-1 antisense R N A as labelled probe led to the same result, i.e., there is no transcription of the XFD-1 gene in suramin treated explants (Fig. 6). In addition, after long time cultivation of suramin treated explants there is a striking decrease in the copy number of tubulin and actin transcripts as compared to induced explants grown in the absence of suramin (Fig. 5d-g). On the other hand, there is a significant increase of embryonic o~-globin transcripts (Fig. 5h,i) which is consistent with the histological results. In summary, these observations fully agree with the above presented findings obtained with whole embryos: suramin inhibits expression of marker genes for dorsal but not for ventral mesoderm. Moreover, they show that suramin changes the mesoderm inducing activity of activin A in animal cap explants by a shift of induction from dorsal to ventral type tissues.

Discussion Treatment of early blastula stage embryos with increasing concentrations of suramin leads to transcriptional suppression of dorsal marker genes, XFD-1, a-actin and nerve specific fl-tubulin, but not of a ventral marker gene, a-globin. In agreement with a recent report (Slack et al., 1992) we detected in many

cases even an increase in red blood cell formation. Resulting embryos are lacking head structures and, at higher suramin concentrations, also a dorsal axis. This observation correlates well to previous experiments, when suramin was injected into the blastocoel; the failure to develop a dorsal axis a n d / o r anterior structures was clearly aligned to an inhibition of convergent extension movements during gastrulation (Gerhart et al., 1989). Transcriptional inactivation of the early response gene XFD-1 now suggests that the inhibition of gastrulation movements is accompanied or even caused by a change in the genetic program of prospective dorsal cells. This, in turn, might be explained by the ability of suramin to block receptors for ligands being involved in the process of dorsalization. Furthermore, no XFD-l-transcripts could be detected in isolated dorsal blastopore lip (Spemann's organizer), after it had been treated with suramin from excision (early gastrula) till late gastrula stage. Since XFD-1 gene is activated in normal embryos after midblastula transition, this observation suggests a down-regulation of this gene in suramin treated lip and a continuous requirement for extracellular signalling in the maintanance of its transcription. Results obtained with molecular markers are in good agreement with previously reported histological results which showed that notochord formation is inhibited by suramin (Grunz, 1992, 1993). Suramin shifted the differentiation pattern of isolated dorsal blastopore lip from dorsal nlesodermal (notochord, somites) to ventral mesodermal structures (especially heart structures). We show here that this ventralization coincides with an inactivation of the XFD-1 gene and a suppression of the a-actin gene. Moreover, these lips have lost their capacity to induce neural structures in competent ectoderm (Grunz, 1992) and, therefore, do not show expression of the CNS specific /3-tubulin gene in adherent ectodermal cells. Down-regulation of the XFD-1 gene together with the suppression of notochord formation in isolated dorsal blastopore lips raised the question, whether suramin can also prevent XFD-1 expression and notochord formation in animal caps treated with dorsal mesoderm inducing growth factors like activin A. In normogenesis, mesodermalization of the marginal zone is mediated by growth factors which are released from the vegetal half and interact with cells of the animal

Fig. 5. Whole mount in situ hybridizationof EDF induced animal caps (-/+ suramin). (a-c) XFD-1; (d,e)/3-tubulin; if,g) a-actin; (h,i) a-globin. Animal caps in (a) and (b) were grown until control embryos(c) had reached stage 11-12; all other animal caps were grown until control embryos had reached stage 37. All animal caps were induced with 75 ng/ml EDF for 4.5 h in the absence or presence of suramin. (a,d,f,h) caps not treated with suramin; (b,e,g,i) caps tredted for 4.5 h with 150/xM suramin, 3 h with 50/xM suramin and grown in suramin free medium until the desired stage.

129

a

d

e

f

g

130 .tz E m t~ -Q

*" ~

Ii

a U.I "~ en ~

Ii

IZl U.l ~ en t~

............... ..........

4 - - XFD-1

EF-I~ Fig. 6. RNase protection analysis for XFD-1 transcripts in suramin treated caps. RNase-protection analysis was performed with 30 /zg total RNA from gastrula stage embryos RNA or with RNA extracted from each 6 animal caps either treated with 75 ng/ml EDF (caps/EDF) or with 75 ng/ml EDF and 150 /~Mol suramin (caps/EDF/suramin) for 4.5 h. After additional incubation for 3 h in Barth solution, RNAs were extracted and hybridized with 32p_

labelled antisense RNA transcribed from a 484 bp EcoRI/BgIII fragment of XFD-1. For control we used 32P-labelled antisense transcripts from a 311 bp PvuII/PstI fragment of the E F - l a sequence pXEF7 (PSting et al., 1990) elongated by 250 nueleotides of vector DNA. Hybridization and RNA digestions were performed as

described (Melton et al., 1984).

hemisphere (Nieuwkoop, 1969; Asashima, 1975; Grunz and Tacke, 1986; Slack et al., 1987; Slack, 1991; Smith, 1989). One of these factors is thought to belong to the F G F family. Basic and acidic F G F induce in uncommitted ectoderm mainly ventral mesodermal structures (blood like cells, coelomic epithelia). On the other hand, certain members of the TGF-/3 superfamily, especially activin A, induce competent ectoderm to form dorsal mesodermal structures like notochord and somites (Asashima et al., 1990; Smith et al., 1990; Eijnden-Van Raaij et al., 1990; Thomsen et al., 1990; Tiedemann et al., 1992). It could be shown that activin A induces the expression of XFD-1 in competent ectoderm, but b F G F did not (Kn6chel et al., 1992; Dirksen and Jamrich, 1992; Ruiz i Altaba and Jessell, 1992). Recombinant erythroid differentiation factor (EDF, identical with activin A) induces dorsal mesoderm derived tissues (notochord and somites) in isolated biastula ectoderm. The simultaneous treatment of the ectoderm with activin A and suramin results in the formation of ventral mesodermal structures, mainly mesothelium and erythroid progenitor cells. It was recently reported that 40 /xM suramin inhibits the

inducing activity of very low concentrated activin A (1 U / m l ) (Slack, 1991). Our findings extend this result by the observation that the initial event is a shift of the induction from dorsal to ventral structures. At a first glance, this reminds of previous results obtained with varying concentrations of the vegetalizing factor or activin A, because different threshold concentrations are known to evoke in a graded fashion completely different mesodermal derivatives (Grunz, 1983; Green et al., 1990). High concentrations of activin A preferentially induce dorsal mesodermal (notochord, somites), low concentrations mainly ventral mesodermal structures. Thus, a simple explanation for the observed shift from dorsal to ventral mesoderm may be given by the assumption that increasing concentrations of suramin lead to a decrease in activin A activity caused by a graded inhibition of the activin A receptor. However, it is worth being noted, that we did not observe a suppression of a-globin gene transcription when increasing the suramin concentration to 0.5 mM or lowering the E D F concentration to 5 n g / m l (data not shown); o n the contrary, these explants exhibited rather strong hybridization signals also in the presence of suramin.

Even at 1 ng/ml we observed a-globin transcripts in both E D F and E D F / s u r a m i n treated c a p s . A reduction in signal intensity was only observed at less than 1 n g / m l E D F in the presence of 150 /zM suramin or higher which roughly corresponds to the data reported by Slack (1991), but decreasing the E D F concentration below 0.1 n g / m l also resulted in a failure of induction. The observation of a-globin gene expression in caps induced with 1 or 5 n g / m l E D F in the presence of 150 /xM suramin also corresponds to the results obtained with whole embryos; suramin treatment at a concentration of 0.5 mM leads to embryos lacking head and dorsal axis structures, but there is absolutely no decrease of a-globin gene transcripts. Therefore, alternative explanations have to be considered, like selective inhibition of certain subtypes of activin A receptors which are characterized by a different response to varying threshold concentrations. The recently reported heterogeneity of such receptors caused by distinct genes and alternative splicing (Attisano et al., 1992) strongly supports this hypothesis, even if only two receptors with equal affinities have so far been reported to be expressed in early Xenopus embryos (Kondo et al., 1991; Mathews et al., 1992; HemmatiBrivanlou and Melton, 1992). It should also be noted that the ventralizing effect of suramin treatment observed in animal cap explants strongly depends on the presence of activin A, because neither suramin treated controls nor uninduced controls reacted with the different probes used as markers. Whereas we cannot exclude the interference of suramin with additional factors or their receptors in whole embryos or in dorsal blastopore lips, observed ventralization of animal cap

131 explants does not require addition of ventralizing factors like bFGF or BMP-4, although the type of tissues induced by these substances is indistinguishable from that induced by activin A in the presence of suramin (Green et al., 1990; K6ster et al., 1991, Jones et al., 1992; Dale et al., 1992). We show that activin A in high concentrations induces notochord and somites, while the same concentration of activin A in the presence of suramin causes the differentiation of erythroid progenitor cells. These results indicate that the formation of dorsal or ventral mesoderm cannot simply be explained by binding of distinct growth factors to their specific receptors. It rather suggests a different mode of action of a given factor by the utilization of different signal transduction pathways or by the interaction with other factors. In this context it has to be mentioned that activin A like activity is already present in the oocyte (Asashima et al., 1991), although activin A transcripts are not detected before gastrulation (Thomsen et al., 1990). The maternal component which is apparently secreted by follicle cells (G.D. Guex et al., lecture at Asilomar conference, 1992) probably represents a more general early mesodermalization signal of unspecified or more ventral character. This is consistent with the fact that low concentrations of activin do always induce ventral mesoderm. The dorsalizing activity which acts during middle to late gastrula stages may already be assigned to increasing amounts of zygotically expressed activin, Dorsalization probably depends upon the expression of early response genes, like goosecoid, Xlim and XFD-1. These genes are exclusively expressed in the organizer area during normogenesis and may control other genes, which code for further inducing factors resulting in the dorsalization of the organizer during gastrulation. Finally the formation of dorsal mesoderm (head mesoderm and notochord) is essential for the induction of neural structures (central nervous system). This view is supported by several facts. Suramin inhibits notochord formation and XDF-1 expression in the organizer area during normogenesis together with a shift from dorsal to ventral mesodermal derivatives. This means, the organizer must have received signals of a ventral character. The dorsalizing activity (induction of notochord and somites) of activin A in the animal cap assay is also inhibited by suramin which analogously results in the formation of ventral mesodermal structures. These results firstly suggest that from the dorsal vegetal zone (Nieuwkoop center)primarily a more general (ventral mesoderm inducing) signal is transmitted to the dorsal animal zone (presumptive organizer). Secondly, low concentrations of a maternal activin component probably induce primarily ventral mesodermal derivatives. In the next step, additional genes are activated within the organizer field by dorsal mesodermalizing factors which may include activin A at elevated concentrations,

In summary, we suggest a dual role for activin A in early embryogenesis. The maternal component contributes a more general signal with mainly ventral character whereas the zygotic component is involved in the dorsalization process. By dorsalization the organizer then aquires the potency to induce the overlying neuroectoderm to form the central nervous system during subsequent developmental stages.

Materials and Methods

Biological material Xenopus eggs were obtained by injecting female frogs with 600 U HCG (Schering AG, Berlin) into the dorsal lymph sack. After in vitro fertilization the embryos were dejellied for 4 to 6 min using 4% cysteinechloride in Steinberg solution (58.2 mM NaC1, 0.67 mM KCI, 0.34 mM Ca(NO3) 2, 0.81 mM MgSO 4) (pH 7.4). Embryos were carefully washed in Steinberg solution and raised until the desired developmental stages (stage classification according to Nieuwkoop and Faber, 1956). Prior to further experiments, the vitelline membrane was mechanically removed using fine forceps.

Suramin-treatment Embryos at stage 7 were incubated for 5 h in Steinberg solution containing 50 /~M or 500 /~M suramin (Germanin ®, Naganol ®, Bayer AG, Leverkusen) respectively, and were then transfered to 0.75 × Steinberg solution until they had reached the desired stage. Especially for gaining later stages it has proven to be useful to transfer the embryos after suramin treatment into 2% ficoll, 0.3 x Steinberg which facilitates gastrulation. After beginning of neurulation they were transferred to 0.75 x Steinberg solution and raised until stage 32/33. Animal caps were excised by using fine glass needles from early to midblastula stage embryos and incubated for 4.5 h with 75 ng/ml recombinant human erythroid differentiation factor (EDF) in Barth solution (44 mM NaC1, 0.5 mM KC1, 0.41 mM MgSO4, 0.17 mM Ca(NO3) 2, 0.2 mM CaCI 2, 0.6 mM NaHCO 3, 0.05 mM NazHPO 4, 0.07 mM KHzPO4) in the absence or presence of 150 /xM suramin. After incubation for additional 3 h in Barth solution (in the absence or presence of 50 /zM suramin) caps were transferred to Holtfreter solution (60 mM NaC1, 0.67 mM KC1, 0.83 mM CaC12) and raised until control embryos had reached stage 12 or stage 37, respectively. Blastopore lips with a small portion of adjacent ectoderm were excised by using fine glass needles from embryos at stage 10 and incubated for 1.75 h in Barth solution in the absence or presence of 300/zM suramin. Suramin-treated lips were transferred to Barth solution containing 150 p~M suramin and incubated for addi-

132 tional 15 min or 4.25 h, respectively. Then all blastopore lips were transferred to Holtfreter solution and raised until control embryos had reached stage 37. In the experiment using XFD-1 as probe the lips were incubated for 5 h in Barth solution in the absence or presence of 150 /.~M suramin. They were then transferred to Barth solution and cultivated until control embryos had reached stage 12. All embryos, caps and blastopore lips were finally fixed in M E M F A (100 m M MOPS, 2 m M E G T A , 1 m M MgSO4, 3.7% formaldehyde) for at least 2 h, then transferred to methanol and stored at - 2 0 ° C . Results as shown in Figs. 1-5 are typical for at least 10 specimens taken for each experiment.

Hybridization probes Probes used for whole mount hybridizations were transcribed in anti sense direction using a commercially available digoxygenin R N A labelling kit (Boehringer). The following f r a g m e n t s / v e c t o r s / R N A polymerases were used: (1) a 484 bp EcoRI/BglII fragment of XFD-1 (KnGchel et al., 1 9 9 2 ) / p S P T 1 9 / T 7 - R N A polymerase; (2) the complete 1744 bp class II isotype /3-tubulin clone 24-10 (Richter et al., 1988; G o o d et al., 1 9 8 9 ) / b l u e s c r i p t / T 3 RNA-polymerase; (3) a 0.95 kb PstI/HindlII a-cardiac actin a l fragment (Stutz and Spohr, 1 9 8 6 ) / p S P T 1 9 / T 7 RNA-polymerase; (4) a 252 bp PstI/BamHI fragment of the larval a-globin clone p X G L 19.1 (Widmer et al., 1981; identical to a T 3 ; Banville and Williams, 1 9 8 5 ) / p S P T 1 8 / T 7 R N A polymerase.

Whole mount in situ hybridization Whole mount in situ hybridization was done as recently described (Harland, 1991) with some minor modifications. We found it very convenient to perform the whole procedure in round-bottomed 2.0 ml Eppendorf-tubes. Volume of required solutions was reduced to 1.5 ml without any loss in signal intensities. Whenever possible we used a rocking platform set to a very low speed to perform the different incubation and equilibration steps. After final fixation of embryos in M E M F A for at least 1 h, they were dehydrated in methanol and partially depigmented in 30% H 2 0 2 / methanol ( 1 / 2 ) over 2 days with several changes of the solution. After this treatment they were dehydrated in a methanol series. Photographs were taken either at this stage or after making the embryos transparent in benzyl b e n z o a t e / b e n z y l alcohol ( 2 / 1 ) for at least 1 h.

RNase protection A 484 bp EcoRI/BglII fragment from the 5'-end of Xenopus laevis XFD-1 sequence (KnGchel et al., 1992)

was cloned in pSPT19. In vitro transcription was performed with a commercially available kit (Boehringer). [32p]CTP labelled antisense R N A was hybridized with R N A s as indicated. Hybridization and RNase digestions were performed as described (Melton et al., 1984). After RNase digestion the protected fragments were separated on a 6% urea-polyacrylamide gel.

Acknowledgements We thank Dr. M. Asashima, Y o k o h a m a University, for the gift of recombinant human activin A (EDF) and S. Effenberger for skillful technical assistance. This work was supported by grants from the Deutsche Forschungsgemeinschaft to W.K. (Kn 200/3-3) and to H.G. (Gr 439/5-2), by the Forschungspool of the Univ e r s i t y / G H Essen and by Landesforschungsschwerpunktprogramm Baden-Wfirttemberg.

References Asashima, M. (1975) Roux's Arch. Dev. Biol. 177, 301-308. Asashima,M., Nakano, H., Shimada, K., Konoshita, K., Ishii, K., Shibai, H. and Ueno, N. (1990) Roux's Arch. Dev. Biol. 198, 330-335. Asashima,M., Nakano, H., Uchiyama, H., Sugino, H., Nakamura, T., Eto, Y., Ejina, D., Nishimatsu, S.-I., Ueno, N. and Kinosha, K. (1991) Proc. Natl. Acad. Sci. USA 88, 6511-6514. Attisano,L., Wrana, J.L., Cheifetz, S. and Massague, J. (1992) Cell 68, 97-108. Banville, D. and Williams, J.G. (1985) Nucl. Acids Res. 13, 54075421. Betsholtz, C., Johnsson, A., Heldin, C.-H. and Westermark, B. (1986) Proc. Natl. Acad. Sci. USA 83, 6440-6444. Chakrabarti, A., Matthews, G., Colman, A. and Dale, L. (1992) Development 115, 355-370. Cho, K.W.Y., Blumberg, B., Steinbeisser, H. and DeRobertis, E.M. (1991) Cell 67, 1111-1120. Coffey, R.J., Leof, E.B., Shipley, G.D. and Moses, H.L. (1987) J. Cell. Physiol. 132, 143-148. Dale, L., Howes, G., Price, B.M.J. and Smith, J.C. (1992) Development 115, 573-585. Dirksen,M.L. and Jamrich, M. (1992) Genes Dev., 6, 599-608. Eijnden-VanRaaij, van den A.J.M., Zoelent, van E.J., Nimmen, van K., Koster, C.H., Shock, G.T., Durston, A.J. and Huylebroeck, D. (1990) Nature, 345, 732-734. Gerhart, J., Danilchik, M., Doniach, T., Roberts, S., Rowning, B. and Stewart, R. (1989) Development, Suppl. 37-51. Good, P.J., Richter, K. and Dawid, I.B. (1989) Nuc. Acids Res. 17, 8000. Green, J.B.A., Howes, G., Symes, K., Cooke, J. and Smith, J.C. (1990) Development, 108, 173-183. Grunz, H. (1983) Roux's Arch. Dev. Biol. 192, 130-137. Grunz, H. and Tacke, L. (1986) Roux's Arch. Dev. Biol. 195,467-473. Grunz, H. (1992) Mech. Dev. 38, 133-142. Grunz, H. (1993) Dev. Growth Differ. 35, 21-28. Harland, R.M. (1991). In: Methods in Cell Biology, Eds. Kay, B.K. Peng, H.B., Academic Press, San Diego, 36, 685-695. Hemmati-Brivanlou,A. and Melton, D.A. (1992) Nature 359, 609614.

133 Jessel, T.M. and Melton, D.A. (1992) Cell 68, 257-270. Jones, C.M., Lyons, K.M., Lapan, P.M., Wright, C.V.E. and Hogan, B.L.M. (1992) Development 115, 639-647. Kn6chel, W. and Tiedemann, H. (1989) Cell Differ. Dev. 26, 163-171. Kn6chel, W., Grunz, H., Loppnow-Blinde, B., Tiedemann, H. and Tiedemann, H. (1989) Blut 59, 207-213. Kn6chel, S., Lef, J., Clement, J., Klocke, B., Hille, S., K6ster, M. and Kn6chel, W. (1992) Mech. Dev. 38, 157-165. K6ster, M., Plessow, S., Clement, J., Lorenz, A., Tiedemann, H. and Kn6chel, W. (1991) Mech. Dev. 33, 191-200. Kondo, M., Tasahiro, K., Fuji, G., Asano, M., Miyoshi, R., Yamada, R., Muramatsu, M. and Shiokawa, K. (1991) Biochem. Biophys. Res. Commun. 181,684-690. Mathews, L.S., Vale, W.W. and Kintner, C.R. (1992) Science 255, 1702-1705. Melton, D.A., Krieg, P.A., Rebagliati, M.R., Maniatis, T., Zinn, K., and Green, M.R. (1984) Nucleic Acids Res. 12, 7035-7056. Murata, M., Eto, Y., Shibai, H., Sakai, M. and Muramatsu, M. (1988) Proc. Natl. Acad. Sci. USA 85, 2434-2438. Nieuwkoop, P.D. and Faber, J. (1956) Normal Table of Xenopny laevis Daudin, North-Holland, Amsterdam. Nieuwkoop, P.D. (1969) Roux's Arch. Entw.Mech. Org., 162, 341373. Oschwald, R., Richter, K. and Grunz, H. (1991) Int. J. Dev. Biol. 35, 399-405. Papkoff, J. and Schryver, B. (1990) Mol. Cell. Biol. 10, 2723-2730. Pinedo, H.M. and van Rijswijk, R.E.N. (1992) J. Clin. Oncol. 10, 875-877.

P6ting, A., Danker, K., Hartmann, L., K6ster, M., Wedlich, D. and Kn6chel, W. (1990) Differentiation 44, 103-110. Richter, K., Grunz, H. and Dawid, I.B. (1988) Proc. Natl. Acad. Sci. USA, 85, 8086-8090. Ruiz i Altaba, A. and Jessel, T.M. (1992) Development 116, 81-93. Sive, H.L; (1993) Genes Dev. 7, 1-12. Slack, J.M.W., Darlington, B.G., Heath, S.K. and Godsave, S.F. (1987) Nature, 326, 197-200. Slack, J.M., Darlington, B.G., Gillespie,, L.L., Godsave, S.F., Isaacs, H.V. and Paterno, G.D. (1989) Development 107, 141-148. Slack, J.M.W. (1991) Development 113, 661-669. Slack, J.M.W., Isaacs, H.V., Johnson, G.E., Lettice, L.A., Tannahill, D. and Thompson, J. (1992) Development Suppl. 143-149. Smith, J.C. (1989) Development 105, 665-667. Smith, J.C., Price, B.M.J., Van Nimmen, K. and Huylebroeck, D. (1990) Nature, 345, 729-731. Stutz, F. and Spohr, G. (1986) J. Mol. Biol. 187, 349-361. Taira, M., Jamrich, M., Good, P.J. and Dawid, I.B. (1992) Genes Dev. 6, 356-366. Thomsen, G., Woolf, T., Whitman, M., Sokol, S., Vaughan, J., Vale, W. and Melton, D.A. (1990) Cell 63, 485-493. Tiedemann, H., Lottspeich, F., Davids, M., Kn6chel, S., Hoppe, P. and Tiedemann, H. (1992) FEBS Lett. 300, 123-126. Widmer, H.J., Andres, A.-C., Niessing, J., Hosbach, H.A. and Weber, R. (1981) Dev. Biol., 88, 325-332.