Anteroposterior patterning and organogenesis of Xenopus laevis ...

Anteroposterior patterning and organogenesis ofXenopus laevisrequirea correct dose of germ cell nuclear factor (xGCNF)

Robert David, Thomas O. Joos, Christine Dreyer*

Max-Planck-Institut fu¨r Entwicklungsbiologie, Spemannstrasse 35/V, D-72076 Tu¨bingen, Germany

Received 29 April 1998; revised version received 28 September 1998; accepted 29 September 1998

Abstract

The germ cell nuclear factor ofXenopus laevis(xGCNF; NR6A1) is a nuclear orphan receptor that is predominantly expressed duringneurula and late tailbud stages. As a strategy to analyze the role of xGCNF in embryogenesis, we have induced a gain of function byoverexpression of full-length (fl) GCNF and a functional inhibition by a dominant-negative (dn) GCNF. Early events of embryogenesisincluding gastrulation and neurulation were not affected and the expression of several early mesodermal markers was normal. Yet specificdefects were observed upon organogenesis. Ectopic posterior overexpression of the full-length xGCNF caused posterior defects anddisturbed somite formation. In contrast, expression of dnGCNF interfered with differentiation of the neural tube and affected the differ-entiation of anterior structures, including the cement gland and the eyes. Embryos affected by dnGCNF were rescued by coexpression offlGCNF. After expression of dnGCNF, mRNA encoding the retinoic acid receptor xRARg2 was selectively suppressed anteriorly. From thedistinct phenotypes obtained, we conclude that GCNF has an essential function in anteroposterior differentiation during organogenesis.1998 Elsevier Science Ireland Ltd. All rights reserved

Keywords:Germ cell nuclear factor; Nuclear orphan receptor; Organogenesis; Dominant negative receptor; Germ cell nuclear factoroverexpression;Xenopus laevis

1. Introduction

Nuclear orphan receptors share the domain structure ofthe nuclear hormone receptors and their ability to regulatethe expression of specific target genes by binding to aresponse element in the promoter region. The germ cellnuclear factor (GCNF; NR6A1) is a nuclear orphan receptorthat is to date the first and only member of the 6th subfamilyof the nuclear hormone receptors. The name GCNF givescredit to the fact that the murine receptor was first detectedin male germ cells (Chen et al., 1994; Hirose et al., 1995;Katz et al., 1997). AXenopusGCNF (xGCNF) was foundindependently and the mRNA encoding the xGCNF is by farmore abundant in the embryo during neurulation and earlyorganogenesis than it is in male germ cells of the adult.Zygotic transcription reaches a maximum during neurula-tion and an anteroposterior gradient of transcripts is estab-

lished during gastrulation and neurulation. Transcriptsspecific for xGCNF are present in anterior ectoderm anddorsal mesoderm, and are especially abundant in the pre-sumptive neuroectoderm during neurulation (Joos et al.,1996). Subsequently the murine GCNF was also describedin embryos and in the developing brain (Su¨sens et al., 1997;Bauer et al., 1997). A human GCNF has also recently beendescribed (Su¨sens and Borgmeyer, 1996; Kapelle et al.,1997; Lei et al., 1997).

Because of the late expression of xGCNF, a role for thisorphan receptor in the early events of mesoderm inductionand neural induction is unlikely. The expression patternrather suggests a function for xGCNF in anteroposteriorpatterning during neurulation and early organogenesis. Dur-ing neurulation and the subsequent tailbud stages, a numberof differentiation events proceed in the craniocaudal direc-tion: In the ectoderm the neural plate segregates from theepidermis between neurula stages 13.5 and 15 and it thick-ens simultaneously, except for a thin stripe along the dorsalmidline (Nieuwkoop and Faber, 1967). Interestingly, this

Mechanisms of Development 79 (1998) 137–152

0925-4773/98/$ - see front matter 1998 Elsevier Science Ireland Ltd. All rights reservedPII S0925-4773(98)00181-6

* Corresponding author. Tel.: +49 7071 601471; fax: +49 7071 601449;e-mail: [email protected]

stripe contains less xGCNF mRNA than the rest of theneural plate (Joos et al., 1996). Similarly, segregation ofthe neural crest from the neural plate and from the neuraltube, as well as the differentiation of the spinal cord proceedin craniocaudal direction. In the axial and paraxial meso-derm, vacuolization of the notochord and somitogenesisalso proceed in craniocaudal direction. (Nieuwkoop andFaber, 1967).

Based on its expression pattern, xGCNF could beinvolved in the regulation of organogenic processes withrespect to the anteroposterior axis. If this is the case, orga-nogenesis would be expected to be sensitive to the dose offunctional GCNF at a given time and location. We havetherefore explored whether ectopic overexpression of thefunctional full-length xGCNF might disturb organogenesis.Furthermore, we have assessed the influence of a dominantnegative xGCNF (dnGCNF) on development, particularlyin the dorsoanterior structures that normally express thereceptor.

2. Results

2.1. xGCNF is a nuclear protein in embryos

As transcription of xGCNF mRNA is spatiotemporallyregulated in embryos ofX. laevis (Joos et al., 1996), wehave studied the expression of xGCNF during embryogen-esis at the protein level, with application of a polyclonalantiserum. On developmental Western blots, a band of theexpected Mr of approximately 50× 103, was detected fromthe late gastrula stage 12.5 onward. The level of this antigenincreases during neurulation and does not change signifi-cantly between the neurula, stage 19, and late tailbud,stage 28 (Fig. 1A). The antigen obviously persists afterthe level of the xGCNF mRNA has decreased (Joos et al.,1996). Whole mount immunofluorescence staining demon-strates that xGCNF is found in most cell nuclei during neur-ulation and tailbud stages. Nevertheless, the pattern ofprotein expression is dynamic, with almost all regions ofthe embryo expressing xGCNF at a certain developmentalstage (Fig. 1B and data not shown). Fig. 1B–D show alongitudinal section of a tailbud embryo at stage 24 stainedfor xGCNF (B), for xRARg (C), and for DNA to identifycell nuclei (D). Anteriorly, all nuclei contain xGCNF,except for those in the eye anlagen and staining is weak inpart of the forebrain (Fig. 1B). In the notochord (data notshown), and in the somites (Fig. 1B), the intensity of nuclearstaining decreases in the craniocaudal direction, with little ifany xGCNF in the posterior presomitic mesoderm. Thewhole epidermis contains xGCNF, but staining is veryweak in the prospective tail. Posteriorly, expression ofxGCNF and xRARg, another nuclear receptor, is almostmutually exclusive at stage 24 (compare Fig. 1B,C). ThexGCNF predominates in the epidermis, the endoderm, andthe differentiated mesoderm, whereas xRARg is mainly

found in the posterior part of the neural tube and in theundifferentiated mesoderm of the posterior wall (Gont et

Fig. 1. xGCNF protein is predominantly expressed during tailbud stages.(A) Developmental Western blot. A polypeptide of the expected Mr of50 × 103 (arrow) is detected by a polyclonal mouse serum between themidneurula and early tadpole stages. A weak signal was detected at thisposition at late gastrula, stage 12.5, after long exposure (data not shown).A second signal results from a nonspecific reaction of the polyclonalantiserum. (B,C) xGCNF and xRARg nuclear proteins are differentiallyexpressed with respect to the anteroposterior axis. Slightly oblique para-sagittal section through a tailbud embryo at stage 24 after wholemountdouble immuno-staining with mouse anti xGCNF(B) and rabbit antixRARg (C), as described in Experimental Procedures. Counterstainingof all nuclei with DAPI is shown in (D). epi, epidermis; en, endoderm;fb, forebrain; e, eye anlage; so, somites; no, notochord; nt, neural tube; pw,posterior wall. Anterior is to the left. Scale bar, 200mm.

138 R. David et al. / Mechanisms of Development 79 (1998) 137–152

al., 1993). In the anterior, nuclei of the ventral headmesenchyme contain both receptors. Interestingly, bothreceptors are absent from most nuclei of the eye anlagenat this stage (Fig. 1B,C). At late tailbud, stage 30, xRARg

persists mainly in the head mesenchyme, gill arches, and tipof the tail (Ellinger-Ziegelbauer and Dreyer, 1991, 1993).The xGCNF antigen persists in the epidermis, somites, andendoderm and in the central nervous system its expressionbecomes restricted to specific ventrolateral cell nuclei of themidbrain and hindbrain area and to part of the spinal cord (tobe published elsewhere). Our immunohistological analysissubstantiates our previous observations by in situ hybridiza-tion on the spatiotemporally regulated pattern of xGCNFexpression during embryogenesis (Joos et al., 1996).Although the murine homologue of xGCNF has been pri-marily associated with the differentiation of neuronal cellsin culture (Bauer et al., 1997; Heinzer et al., 1998) and inembryos (Bauer et al., 1997; Su¨sens et al., 1997), our resultsdemonstrate the transient presence of xGCNF protein indifferentiating cells of all germ layers (Fig. 1B).

2.2. Ectopic expression of myc-tagged xGCNF proteins inembryos of X. laevis

The specific expression pattern of xGCNF during embry-ogenesis suggests that organogenesis may require a definedlevel of functional xGCNF, and therefore ectopic expressionof the protein might interfere with organogenesis. Overex-pression of the full-length protein should reveal a gain offunction phenotype, whereas expression of a dominantnegative protein should lead to a loss of function of theendogenous GCNF. We generated a nonfunctional GCNFby deleting the N-terminus, including part of the DNA bind-ing domain. Experimental evidence from several labora-tories suggests that efficient binding of GCNF to its targetrequires homodimer formation (Chen et al., 1994; Hirose etal., 1995; Joos et al., 1996; Borgmeyer, 1997; Bauer et al.,1997; Yan et al., 1997) and there is no evidence for hetero-dimer formation with the retinoid X receptor RXR (Borg-meyer, 1997; Bauer et al., 1997; and our unpublishedobservations), therefore a non-DNA-binding GCNF mutantis likely to act as a dominant negative by heterodimer for-mation with the intact GCNF, provided that the dimerizationinterface is preserved.

To test this hypothesis, full-length and N-terminally trun-cated polypeptides translatedin vitro were tested for bindingto DNA containing the response element DR0, which is adirect repeat of the motif AGGTCA. As expected, only thefull-length form of the xGCNF bound to DR0 (Fig. 2A, lane1), whereas the N-terminal truncation completely inhibitedbinding to DR0 (lane 5). When increasing amounts of thetruncated form were mixed with a constant amount of full-length xGCNF, less DR0 was bound (Fig. 2A, lanes 2–4).No complexes migrating faster than the xGCNF homodimerwere observed, indicating that heterodimers do not bindefficiently to the response element. We therefore expected

that an excess of the truncated xGCNFDN would impair thefunction of endogenous xGCNF in the embryo and wouldtherefore function as a dominant negative (dn) GCNF invivo.

For ectopic expression studies, the full-length and the 5′truncated xGCNF cDNA were cloned in frame behind acDNA encoding six myc-epitopes in the pCS2+ MTexpression vector as depicted in Fig. 2B. For transientexpression of fl and dnGCNF, mRNA encoding the fusionproteins shown in Fig. 2B was transcribed in vitro. Weroutinely injected 120 pg of mRNA per embryo into oneblastomere at the two-cell stage. To adjust the amount ofmRNA required for the synthesis of an adaquate amount ofprotein in the embryo, we performed Western blots afterinjection of the mRNAs. The monoclonal antibody (mAB)9E10 directed against the myc-tag exclusively recognizedthe polypeptides translated from the injected mRNAs (upperpanels of Fig. 3), whereas the xGCNF antiserum allowed thedirect comparison of the amounts of endogenous and exo-

Fig. 2. (A) Dominant negative effect of GCNFDN over flGCNF tested byEMSA using labeled DR0 as probe. The truncated GCNF does not bind toDR0 (lane 5). Increasing amounts of dnGCNF impair binding of flGCNF,most probably by formation of a heterodimer that does not bind efficiently(lanes 1–4). (B) Scheme showing the proteins encoded by injected mRNA.The open reading frame of xGCNF is shown as an open bar and the DNAbinding domain and the conserved regions I and II of domain E are indi-cated. Numbering of amino acids above the bar is as published (Joos et al.,1996). xGCNF and xGCNFDN cDNA was cloned into the vectorpcs2+ MT (Rupp et al., 1994). mRNA encoding myc-tagged proteinwas injected to allow distinction between endogenous and exogenousGCNF protein.

139R. David et al. / Mechanisms of Development 79 (1998) 137–152

genous xGCNF protein (middle panels of Fig. 3). The mABb7-1A9, directed against nucleoplasmin (Zimmer et al.,1988), provided a loading control (lower panels of Fig. 3).Protein from uninjected control embryos and from embryosinjected with mRNA encoding the fl or dnGCNF was ana-lyzed at the midgastrula, stage 11, neurula, stage 18, andtailbud, stage 23. After mRNA injection at the 2-cell stage,myc-tagged proteins were detected in significant amounts atstage 11, prior to accumulation of the endogenous xGCNF.Myc-tagged proteins were present until at least the tailbudstages in amounts comparable to the level of endogenousxGCNF (Fig. 3). About 120 pg of exogenous mRNA wererequired to obtain about equal amounts of endogenous andexogenous proteins per embryo. This corresponds to aseveralfold excess of the exogenous protein near the siteof injection, because the injected mRNA does not diffusethroughout the embryo (see below).

In order to determine the maximally tolerable level ofexogenous mRNA per embryo, 30–240 pg of mRNAencoding dn or flGCNF were injected at the two-cellstage. Although early development appeared unaffected,the percentage of embryos showing normal developmentbeyond neurulation decreased significantly with increasingdose of mRNA injected, and during tailbud stages mostembryos became extremely stunted, although their mainbody axis could still be identified. In most subsequentexperiments 120 pg of mRNA was injected into a singleblastomere at the two-cell stage, leaving one hemisphereas control. For phenotypic analysis we concentrated onembryos with a discernible main body axis and specimensseverely damaged by exogastrulation or spina bifida werescored as not evaluated in Table 1.

2.3. Overexpression of flGCNF perturbs tail and somitedevelopment

Injection of mRNA encoding flGCNF into one blasto-

mere at the two-cell-stage resulted in embryos with ashort main body axis at late tailbud stage 30. At later stages,multiple and crippled tail tips were found in a percentage ofinjected embryos that was dependent on the batch of eggs,whereas heads usually developed normally (Fig. 4 and datanot shown). In eight independent experiments compiled inTable 1, a total of 934 embryos surviving to the end ofneurulation were evaluated. The embryos are listed accord-ing to their most characteristic externally visible defects. Inall of the experiments malformations at the posterior endwere most frequently seen after overexpression of theflGCNF. Of a total of 331 evaluated embryos injectedwith flGCNF mRNA, 45% had significant defects in poster-ior axis formation. Between individual experiments, thispercentage varied between 31 and 58%. As compiled inTable 1, this phenotype was more rarely observed afterexpression of dnGCNF (12% of 265 specimens) or in con-trol embryos (5% of 338 specimens).

Aside from tail defects, disturbed patterns of somitogen-esis were also found in 47% of all tadpoles after overexpres-sion of flGCNF (Table 1). Although these tadpoles haddifferentiated muscles and were able to swim, the somitepattern appeared to be significantly compressed in areas ofmaximal overexpression of the flGCNF. Different degreesof local irregularity of the somites are illustrated in Fig. 5B–D. The regular somite pattern of a control sibling at stage 40is shown in Fig. 5A. The arrow in Fig. 5B points to a slightirregularity in the pattern, whereas the specimens shown inFig. 5C,D have irregular somite boundaries and a group ofcompressed somites. In the example shown in Fig. 5D, myc-tagged flGCNF could still be traced in the endoderm in theposterior third of the tadpole at stage 40, just beneath thearea of maximal compression of the somites. A disturbanceof somite formation, although on average at least as fre-quently observed as the posterior defects, is however, notexclusively associated with the overexpression of flGCNFbut was also observed after expression of the dnGCNF (see

Fig. 3. Expression of xGCNF proteins in embryos after injection of mRNA at the two-cell stage. Embryos were injected with 120 pg of mRNA encodingdnGCNF or with 40 pg flGCNF mRNA into one blastomere at the two-cell stage as indicated at the top. Control embryos were injected with water. Injectedembryos were analyzed by Western blotting at gastrula, stage 11, neurula, stage 18, or tailbud, stage 23, as indicated at the bottom. Upper panels: mycepitopetag of protein translated from exogenous mRNA detected by mAB 9E10. Middle panels: Exogenous and endogenous polypeptides detected by the antiserumdirected against xGCNF. Lower panel: Loading control, endogenous nucleoplasmin is detected by mAB b7-1A9 (Zimmer et al., 1988).


below). On the other hand, somitogenic defects wereobserved in only 5% of 338 control embryos (Table 1).

Doubling the dose of mRNA per embryo increased thepercentage of severely damaged embryos that were not eval-uated, but increased the percentage of specific defects onlyslightly. This effect is seen in Table 1, where the experi-ments with injection of two times 120 pg of mRNA intoboth blastomeres are marked with an asterisk. Posteriordefects after overexpression of flGCNF were seen in 57%of these embryos, as compared to 42% after injection of 120pg and the percentage of somitogenic defects increased from54 to 61%. In conclusion, regular somitogenesis apparentlyrequires a normal level of GCNF and tail development is

sensitive to ectopic overexpression of functional GCNF. Incontrast, head formation can proceed normally in the pre-sence of a moderate excess of flGCNF.

2.4. Expression of a dnGCNF affects development of headand neural tube

Whereas posterior defects prevailed after overexpressionof the flGCNF, expression of the dnGCNF resulted in fre-quent abnormalities in head development. This wasobserved in seven independent experiments with injectionof mRNA encoding dnGCNF into one or both blastomeresat the two-cell stage (Table 1). Of 265 embryos scored,

Table 1

Effects after mRNA injection at the two-cell stage

Experiment mRNA Embryosinjected

Survivors > stage 20 Not evaluated Anterior defects Posterior defects Somitogenic defects

n % n % n % n % n %

I$ GFP 30 24 100 8 33 1 4 0 0 0 0Dom. neg. 34 26 100 2 8 18 69 0 0 11 42Full length 30 18 100 4 22 3 17 8 44 6 33

II $ GFP 75 60 100 31 52 0 0 0 0 0 0Dom. neg. 90 75 100 32 43 37 49 8 11 20 27Full length 75 62 100 25 40 5 8 27 44 34 55

III $ GFP 60 41 100 13 32 0 0 0 0 0 0Dom. neg. 60 43 100 5 12 23 53 6 14 12 28Full length 70 47 100 21 45 6 13 17 36 17 36

IV$ GFP 42 22 100 4 18 1 5 0 0 0 0Dom. neg. 50 39 100 16 41 14 36 1 3 10 26Full length 50 39 100 23 59 4 10 12 31 31 79

V bGal. 80 75 100 1 1 9 12 3 4 ne neDom. neg. 80 41 100 2 5 25 61 7 3 ne neFull length 80 46 100 2 4 18 39 22 39 ne ne

VI bGal. 70 58 100 0 0 0 0 3 5 3 5Dom. neg. – – – – – – – – – – –Full length 70 63 100 0 0 3 5 31 49 35 55

VII* GFP 50 30 100 7 23 5 17 6 20 8 27Dom. neg. 50 19 100 8 42 11 58 5 26 00 58Full length 50 24 100 10 42 7 29 14 58 14 58

VIII* GFP 50 28 100 5 18 3 12 4 14 7 25Dom. neg. 50 22 100 8 36 13 59 4 18 12 54Full length 50 32 100 7 22 11 34 18 56 20 63

Total Control 457 338 100 69 20 19 6 16 5 18 5Dom. neg 414 265 100 73 28 141 53 31 12 76 29Full length 475 331 100 92 28 57 17 149 45 157 47

Embryos that developed beyond end of neurulation (stage 20) were evaluated at stage 23 in experiment V, and at stage 30 in all other experiments. Embryosshowing severe malformations of the exogastrula and spine bifida types, were scored as not evaluated. Anterior defects comprized narrow or short heads withoften one eye rudimentary or not detected from the outside (Figs. 6C and 8A). Somitogenic defects are illustrated in Fig. 5. Embryos with ‘posterior’ defectswere usually shorter than normal and had round (Fig. 6B), crippled, or multiple tail tips (Fig. 4). Experiments with double injection of 120 pg mRNA intoboth blastomeres are marked with an asterisk. Localization of the myc-tagged protein 2 days after injection, when control siblings had reached stage30 to 33,was possible in more than 90% of the embryos tested. In the experiments marked with a $, mRNA encoding GFP was co-injected with the xGCNF-specificmRNA. This control allowed us to locate the area of maximal expression by inspection of the living embryos under UV light. In this set of experiments, GFPwas detected in more than 90% of all injected embryos between late gastrula and at least stage 30.


about 53% had narrow or short heads, often in combinationwith rudimentary or even missing cement glands, eyes, or

ear cups. In contrast, anterior defects were observed in 17%of 331 embryos after overexpression of the flGCNF andonly 6% of 338 embryos after expression of GFP or b-galactosidase had abnormal heads (Table 1).

In order to correlate the observed defects with the site ofinjection, 30 pg of mRNA was injected into one marginal orvegetal blastomere at the 32-cell stage. The results of thisseries of experiments were similar to those shown in Table1, albeit with weaker effects due to the lower dose perembryo (data not shown). Subsequently, areas of maximalexpression were identified by wholemount immunostainingof the specimens with mAB 9E10 that detects the myc-tagged proteins (Fig. 6). Detection of the myc-tagged pro-teins allowed a retrospective identification of the lineage ofthe injected cells. Fig. 6 shows tailbud stage 30 embryosafter anti-myc-tag whole mount immunostaining. The con-trol embryos show that a comparatively high amount of MT-GFP is compatible with normal development (Fig. 6A). Fig.6B shows two examples of only moderately damagedembryos after ectopic expression of MT-flGCNF through-out the dorsal axis. Both heads appear normal, although theMT-flGCNF protein is found around the eyes. Both tailbudsare short and round. A ventral edema is seen in the upper

Fig. 4. Posterior defects after overexpression of full-length GCNF. Anormal embryo at stage 40 is shown at the upper left. The others aresiblings with a short main body axis, edema, and defects in tail develop-ment after injection of 120 pg of mRNA encoding flGCNF into one blas-tomere at the two-cell stage. Scale bar, 100mm.

Fig. 5. Overexpression of full-length GCNF disturbs somitogenesis. After injection of mRNA encoding flGCNF at the 2-cell stage, embryos were fixed,immunostained for expression of myc-tagged GCNF and inspected under interference contrast at stage 40. (A) Normal somite pattern; (B) Arrow marksminor disturbance; (C,D) Arrows point to a zone of irregular segmentation and compressed somites. Even at this advanced developmental stage, the area ofmaximal overexpression of myc-tagged GCNF can be recognized near the disturbed zone in (D). Scale bar, 200mm.


embryo at a site of abundant antigen. Fig. 6C shows twoembryos with predominantly anterior defects after expres-sion of dnGCNF along the main body axis. In addition, theupper specimen is significantly stunted. After wholemountstaining for the myc-tagged proteins, serial sections werecut and counterstained with toluidine blue for the histologi-cal analysis of the defects (Fig. 7). Most individuals thatexpressed dnGCNF but had no overt spina bifida had lesionsin the neural tube where the dnGCNF was localized (Fig.7A) and in about 30% of these cases the neural tube was notperfectly closed (Fig. 7D). This phenotype was histologi-cally analyzed in specimens from four independent experi-

ments after expression of dnGCNF (data not shown).Lesions in the neural tube were rarely observed after over-expression of flGCNF (Fig. 7B,C). The specimen shown inFig. 7B has MT-flGCNF expressed in the roof plate, yet theneural tube is closed. In Fig. 7C a disorganized somite isseen at the site of expression of MT-flGCNF. Single cellscontaining high amounts of myc-tagged dnGCNF orflGCNF often appeared undifferentiated or necrotic or notproperly integrated into the surrounding differentiated tissue(e.g. in Fig. 7D in the upper left neural tube). Furthermore,in embryos expressing dnGCNF a clearcut separation of theanlagen of different organs in the head and anterior trunkregion was frequently lacking (right parts of Fig. 7A,D) andthe notochord appeared to be fused to the adjacent somitesin many instances (data not shown). The presence of MT-GFP did not affect the differentiation of the neural tube (Fig.7E). The sites of myc-tagged protein expression were con-firmed on adjacent sections that were not stained with tolui-dine blue (data not shown).

To further scrutinize the specificity of the effects ofdnGCNF expression, we performed rescue experimentsshowing that frequency and strength of the typical defectsoccuring after injection of mRNA encoding dnGCNFwere reduced after coinjection of an equal amount ofmRNA encoding the flGCNF (Table 2). In experiments1 and 2, mRNA was injected into a single blastomere atthe two-cell stage and in experiment 3 both blastomereswere injected. In three independent experiments, additionof functional GCNF to embryos expressing dnGCNFincreased the percentage of survivors. Importantly, theaverage percentage of normal looking embryos increasedfrom 28% to 58% and the average percentage of embryoswith defective or missing eyes and cement glands wasreduced from 57 to 36%. Furthermore, the lesions wereless severe after coinjection of both mRNAs (Fig. 8).Representative specimens evaluated after injection ofdnGCNF mRNA into both blastomeres are shown inFig. 8A. These embryos have no externally visible eyeanlagen and the cement glands are missing or extremelycryptic. For comparison, two anomalous embryosobserved after coinjection of mRNA encoding dnGCNFand flGCNF are shown in Fig. 8B. These embryos appearsignificantly closer to normal phenotypes, although theeye anlagen are reduced in size as compared to the GFPexpressing controls shown in Fig. 8C. Importantly, rescueby flGCNF prevented the most extreme phenotypes witheyes and cement glands completely missing.

2.5. Markers of early mesoderm differentiation are notaffected by dnGCNF

Amazingly, the specific defects observed after injectionof mRNA at the two-cell stage became visible after theembryos had appeared normal up to neurula stages, withno indication of an abnormal gastrulation or neurulation.No higher incidence of exogastrulation than in the controls

Fig. 6. Immunodetection of exogenous myc-tagged proteins after injectionof mRNA. At the 32-cell stage, a single blastomere was injected with 30pg of mRNA encoding MT-GFP (A), MT-flGCNF (B), or MT-dnGCNF(C). Embryos were fixed at late tailbud stage 30 and stained as whole-mounts with mAB 9E10 for 24 h to reveal the myc-tagged proteins (browncolour) as described in Section 4. Mild phenotypes after expression offlGCNF are shown in (B), with short round tails. Strong phenotypesafter expression of dnGCNF along the whole main body axis, with pre-dominant failures in head differentiation, are shown in (C).


injected with mRNAs encoding b-galactosidase or GFP wasobserved and embryos with severe exogastrulation andspina bifida were not evaluated.

Although the endogenous xGCNF was not expressedbefore the end of gastrulation (Fig. 1A) and therefore unli-kely to be involved in the early events of mesoderm induc-tion, the exogenous GCNF proteins were found prior to theendogenous receptor (Fig. 3) and might therefore interferewith early events of embryogenesis.

To investigate whether markers that are differentiallyexpressed during early gastrula stages are affected, weinjected albino embryos into both blastomeres at the two-cell stage with two times 120 pg of mRNA encoding dn orflGCNF plus 30 or 120 pg of GFP mRNA as an in vivotracer. Embryos were fixed and subjected to whole mountin situ hybridization at gastrula stages. All specimens

expressed chordin mRNA at the dorsal lip at stages 10–10.5, irrespective of whether mRNA encoding dnGCNF,flGCNF, or GFP was injected (data not shown). This findingprecludes that premature expression of dnGCNF could ven-tralize the embryo, since after ventralization of embryos byUV-light expression of chordin mRNA is suppressed (Sasaiet al., 1996).

Similarly, the mesodermal marker brachyury (xbra;Smith et al., 1991) was expressed in its normal ringshapeddomain at stage 10.5–11 after ectopic expression ofdnGCNF (Fig. 9C,E) or flGCNF (Fig. 9F). An irregularpattern of xbra expression was observed in specimenswith abnormal gastrulation that normally did not surviveuntil neurula stage. Importantly, subsequent localizationof the exogenous protein by an anti-myc-tag antibodyshowed that xbra was transcribed at the sites of maximal

Fig. 7. Defective development of the neural tube after expression of dnGCNF. Embryos injected with 30 pg of mRNA into a dorsal marginal blastomere at the32-cell stage were fixed at stage 30 and immunostained for myc-tagged proteins as described in Fig. 6. Embryos with significant expression of myc-taggedproteins but without overt spina bifida were embedded in Technovit and serial transverse sections were analyzed for the exogenous protein by the brownperoxidase label. Failures in differentiation (A) or closure of the neural tube (D) were frequently found were MT-dnGCNF was produced (A,D). Afteroverexpression of flGCNF (B,C), the neural tube was usually closed, even when it contained labeled cells, as seen in the roof plate in (B). Note disturbedsomitogenesis on the injected side in (C). Differentiation of the neural tube remained unaffected after ectopic expression of MT-GFP (E). Alternating sectionswere briefly stained with toluidine blue to facilitate histological analysis (A–E) or remained unstained for most sensitive detection of the myc-tagged proteins(not shown). Only the dorsal part of each section is shown.


protein expression of dnGCNF or flGCNF (Fig. 9E,F).Furthermore, after expression of dnGCNF, the pattern ofmyoDb (Rupp and Weintraub, 1991) mRNA was not sig-nificantly affected at midgastrula stages (Fig. 9A). At laterstages, however, the transcriptional pattern of myoD wasaffected in embryos with significantly disturbed organogen-esis after ectopic expression of fl or dnGCNF (data notshown). These results are in agreement with the notion,that dnGCNF interferes with the normal function of theendogenous GCNF, which is not yet present in significantamount at early gastrula.

2.6. Molecular marker gene expression after interferencewith xGCNF function

RT-PCR analysis of selected molecular markers was usedto corroborate the existence of distinct phenotypes afterectopic expression of either dn or flGCNF. Because of thespatiotemporal expression pattern of the endogenousxGCNF and because of the absence of specific lesions up

to the end of neurulation, we investigated genes that areexpressed after gastrulation in a distinct pattern with respectto the anteroposterior axis, or are restricted to specific tis-sues during organogenesis. After exogenous expression ofeither full length or dnGCNF, RNA was prepared from totaland half embryos at stages 23 and 27, and from anterior,middle, and posterior thirds of early tadpoles at stage 30.Embryos injected with mRNA encodingb-galactosidase orGFP were used as controls. Of the markers used, xRARg2

and HoxB9 have a differential expression pattern withrespect to the anteroposterior axis in normal embryos. ThemRNA of HoxB9, previously called XlHbox6, is foundposteriorly (Sharpe et al., 1987) and xRARg2 mRNA isfound primarily in the anterior and posterior regions of theembryo, with a peak of expression during midneurulastages, coincident in time with the maximal expression ofendogenous xGCNF mRNA (Ellinger-Ziegelbauer andDreyer, 1991; Pfeffer and De Robertis, 1994; Joos et al.,1996). As seen in lanes 4–6 of Fig. 10, the results of the RT-PCR of the controls reflect the anteroposterior distribution

Table 2

Rescue of the dnGCNF phenotype by flGCNF

DnGCNF+ GFP dnGCNF+ flGCNF GFP

n % n % n %

Experiment 1Embryos injected 50 – 50 – 50 –Survivors 22 100 35 100 36 100Normal 3 14 18 51 28 78Anterior defects 17 77 15 43 6 17Posterior defects 8 36 6 17 2 6Spina bifida 3 14 8 23 4 11Not evaluated 1 5 1 3 2 6

Experiment 2Embryos injected 80 – 80 – 70 –Survivors 44 100 54 100 43 100normal 19 43 38 70 38 88Anterior defects 16 36 13 24 2 5Posterior defects 6 14 5 9 1 2Spina bifida 2 5 3 6 0 0Not evaluated 2 5 1 2 2 5

Experiment 3Embryos injected 60 – 60 – 60 –Survivors 23 100 36 100 49 100Normal 3 13 16 44 38 78Anterior defects 18 78 17 47 8 16Posterior defects 1 4 1 3 2 4Spina bifida 0 0 1 3 0 0Not evaluated 2 9 0 0 0 0

Embryos were injected at the two-cell stage into a single blastomere (experiments 1 and 2), or into both blastomeres (experiment 3) with mRNAs encodingdnGCNF and GFP, dnGCNF and flGCNF, or GFP alone. The total amount of mRNA injected per embryo was 200 pg in experiments 1 and 2 and 120 pg inexperiment 3. Embryos with lethal cytolysis or massive exogastrulation were removed. In three different experiments, the percentage of embryos thatsurvived beyond neurulation was lowest in the batches expressing dnGCNF (38 to 55%). After coexpression of dnGCNF and flGCNF, the percentage ofsurvivors was increased to between 60 and 70%. As illustrated in Fig. 8, the anterior defects were qualitatively less severe in the rescued embryos. Inalmostall of the embryos with ‘anterior defects’, eyes were too small or one or both eye anlagen were missing. In the batches injected with dnGCNF mRNA alone,about one third of the anteriorly defective specimens had no cement gland. This phenotype was very rarely found after rescue by flGCNF Posterior defectswere significantly less frequent than anterior defects and are not listed in this table.


of xRARg2 and of HoxB9 mRNA as it was previously foundby in situ hybridization.

Products resulting from the RT-PCR analysis were quan-

tified by means of a phosphorimager and the intensity of theEF1a band was used as an internal standard in order tocompare experimental and control embryos. After exogen-

Fig. 8. Typical lesions caused by dnGCNF are rescued by fl GCNF. Embryos were injected into both blastomeres at the two-cell stage with 30 pg of dnGCNFmRNA plus 30 pg of GFP mRNA (A), or 30 pg of dnGCNF mRNA plus 30 pg of flGCNF mRNA (B), or with 60 pg of GFP mRNA (C). The figure illustratestypical specimens of experiment 3 listed in Table 2, excluding the specimens with spina bifida or those scored as not evaluated. Anterior is to the right.

Fig. 9. Early expression of mesodermal markers is unaffected by dnGCNF or flGCNF. Albino embryos were injected in both blastomeres at the two-cell stagewith two times 120 pg of mRNA encoding dnGCNF (A,C,E), flGCNF (F). In addition, 30 pg of GFP mRNA per blastomere was injected to allow monitoringof protein expression in living embryos. Injected embryos were selected for significant expression of GFP at early gastrula, fixed at stage 10.5 to 11, andsubjected to wholemount in situ hybridization as described, using a digoxigenin-labeled antisense RNA specific of MyoDb (A,B) or xbra (C–F). Noninjectedcontrols are shown in (B) and (D). After in situ hybridization, the specimens shown in E and F were stained with mAB 9E10 to localize the areas of maximalexpression of the myc-tagged dnGCNF (E) and flGCNF (F) proteins.


ous expression of dnGCNF, we observed no significantchange in the amount of HoxB9 transcript, whereas thelevel of xRARg2 transcript was reduced anteriorly toabout 50% (Fig. 10A, lanes 1–4). In three independentexperiments after injection of 120 pg of dnGCNF mRNAinto a single blastomere, the relative amount of xRARg2

mRNA in the anterior third varied between 39 and 78% ofthat in the control. After injection of both blastomeres with120 pg each, this amount was 27% that of the control. Thereduction of the amount of xRARg2 transcript afterdnGCNF mRNA injection was substantiated by in situhybridization of embryos at stage 30 (data not shown).The xRARg2 mRNA located posteriorly was apparentlyunchanged as observed by both RT-PCR and by in situhybridization (Fig. 10A and data not shown). Similarly, areduction in the relative amounts of xRARg2 mRNA toabout 60% of the controls was seen in anterior but not pos-terior halves of embryos analyzed at stage 23 and stage 27after expression of dnGCNF (data not shown).

Overexpression of flGCNF had no significant effect onxRARg2 or HoxB9 (lanes 4–9 of Fig. 10A,B). Similarly, noapparent changes in the amount of xRARg1 mRNA wereinduced in our RNA injection experiments (data not shown).Furthermore, RT-PCR analysis was performed to comparethe abundance of mRNAs specific for slug (Mayor et al.,1995), snail (Essex et al., 1993), activin receptor II (Math-ews et al., 1992) after ectopic dn and flGCNF expression.Any changes observed in the amounts of these markers ascompared to the controls were subtle and usually less than afactor of two (data not shown).

Because our histological analysis suggested that loss ofGCNF function by exogenous expression of dnGCNF

resulted in fusion or lack of separation of several anlagenduring organogenesis, we included fibronection in our ana-lysis. Transcripts encoding fibronectin are expressed ubiqui-tously in embryos throughout the tailbud stages (Lee et al.,1984). After overexpression of flGCNF, the relative amountof fibronectin mRNA was increased, on average, by a factorof about two in 5 independent experiments (Fig. 10A, lanes4–9, and data not shown). This increase was not consis-tently seen in a particular part of the embryo at stage 30and it was not observed at earlier stages of development(data not shown).

Wholemount immunofluorescent staining of injectedembryos for fibronectin highlighted the defects observedin organogenesis (Fig. 11). In sections, ectopic boundariescontaining fibronectin were found in the notochord andsomites after expression of dnGCNF (Fig. 11B), and evenmore frequently after overexpression of flGCNF (Fig. 11C).The specimen shown in Fig. 11C shows an apparent dupli-cation of the notochord after overexpression of flGCNF. Inanother specimen sectioned horizontally, somitogenesis wasseverely affected in the part of the tail which overexpressedflGCNF (Fig. 11D).

In summary, these examples corroborate our observationsof the distinct phenotypes produced by expression of the fland the dnGCNF. Because the changes induced by injectedfl or dnGCNF transcripts on the amounts of xRARg2 and offibronectin did not exceed a factor of two, they may repre-sent an indirect consequence of interference with GCNFfunction.

3. Discussion

3.1. Exogenous expression of dn and flGCNF producedistinct phenotypes

Investigation of the normal pattern of xGCNF expressionin embryos revealed a widespread, although transient, pre-sence of the nuclear protein in all three germ layers through-out neurulation and early tailbud stages (Fig. 1B and datanot shown). Therefore, interference with GCNF function inembryos was likely to result in pleiotropic effects on celldifferentiation.

Embryos after exogenous expression of dn and flGCNFlooked normal until the end of neurulation and the transcrip-tional pattern of the early mesodermal markers chordin,xbra, and myoD was essentially normal after injection offl or dnGCNF mRNA (Fig. 9 and data not shown), indicat-ing that xGCNF is not involved in the early processes ofgastrulation and neurulation. Our results show that func-tional xGCNF is required for the differentiation of organsderived from the dorsoanterior ectoderm and mesoderm,since dnGCNF primarily disturbed differentiation of theneural tube, the head and sense organs, whereas the effectson differentiation of the tail were minor. Conversely, anoverdose of functional GCNF affected primarily the differ-

Fig. 10. Multiplex RT-PCR analysis of transcripts after exogenous expres-sion of fl and dnGCNF. Embryos were injected into one blastomere at thetwo cell stage with mRNA encoding either dnGCNF (lanes 1–3), GFP(lanes 4–6) or flGCNF (lanes 7–9). Anterior, middle and posterior thirdswere cut from tailbud embryos at stage 30, RNA was extracted and sub-jected to RT-PCR analysis as detailed in Section 4. (A) cDNA representingfibronectin, xRARg2, and EF1a was amplified simultaneously and quanti-fied by means of a phosphorimager and the EF1a band was used as aninternal standard. (B) RT-PCR for simultaneous quantification of HoxB9and EF1a cDNA.


entiation of the tail, consistent with the absence of xGCNFfrom the tailbud in normal embryos at early stages (Fig. 1Band Joos et al., 1996). The prevalence of head defects afterexpression of dnGCNF suggests that the toxicity ofdnGCNF is due to interference with the endogenousxGCNF, because dnGCNF may be predicted to induce aloss-of-function phenotype since efficient binding ofGCNF to its target requires homodimer formation (Chenet al., 1994; Hirose et al., 1995; Joos et al., 1996; Borg-meyer, 1997). On the other hand, massive overexpressionof any nuclear receptor is likely to cause pleiotropic, but lessspecific effects by trapping of cofactors required for a num-ber of different nuclear hormone receptors and potentiallyeven for other transcription factors (see Ogryzko et al.,1996, and references therein). Furthermore, massive expres-sion of a full-length or N-terminally truncated receptormight likewise trap a putative ligand and deplete the cellof corepressors or coactivators that bind to the intact E-domain. Since the phenotypes observed after expressionof dnGCNF and flGCNF were distinct after injection ofmoderate doses of mRNA, we suppose that they are char-acteristic of loss or gain of xGCNF function. Most impor-tantly, the observation that co-expression of flGCNFameliorated the development of dnGCNF expressingembryos demonstrates that dnGCNF specifically interfereswith the normal function of the endogenous GCNF (Fig. 8

and Table 2). As expected, complete rescue was notobtained, most probably because of the stochastic natureof dimer formation.

3.2. A potential role of xGCNF in somitogenesis

Defects in the somite pattern were observed after over-expression of flGCNF and after expression of dnGCNF(Figs 5, and 11; Table 1), indicating that the normal somi-togenesis depends on a correct timing and dosage of func-tional xGCNF in the paraxial mesoderm. Fig. 1B shows ananteroposterior gradient in the amount of xGCNF antigen inthe somites, with the lowest level of xGCNF in the posteriorsomitogenic mesoderm. Overexpression of flGCNF is likelyto result in the premature occurrence of xGCNF in cellnuclei of the presomitic mesoderm and thereby interferewith pattern formation. On the other hand, the dnGCNFmay interfere with the function of the xGCNF present inthe nuclei of somites that have already formed.

A wave of elongation of a zone in the presomitic meso-derm precedes somitogenesis and the somites further elon-gate after their formation (Elsdale and Davidson, 1983).Overexpression of GCNF may affect either or both ofthese elongation processes, and thereby induce abnormallyshort somites in the area of maximal overexpression (Fig.5D). This phenotype is reminiscent of the locally disturbed

Fig. 11. Fibronectin in extracellular matrix of tadpoles after injection of mRNA encoding dn and flGCNF. (A) Dorsal part of a transverse section of a stage 30embryo after expression of GFP as control. (B) Dorsal part of a transverse section of a stage 30 embryo after expression of dnGCNF. (C) Transverse sectionof sibling after overexpression of flGCNF reveals disrupted organization of the somites (so) and an apparent duplication of the notochord (no) in thisspecimen. (D) posterior part of a horizontal section of an embryo after overexpression of flGCNF. This embryo is bent laterally and the tail somites aredisorganized in the injected hemisphere. Scale bar, 300mm.


somite pattern induced by a heat shock prior to somitogen-esis (Cooke, 1978). The ample evidence, both at the struc-tural and molecular level, for the existence of a prepattern inthe somitogenic mesoderm of different vertebrate specieshas recently been reviewed (Gossler and Hrabe de Angelis,1997; Yamaguchi, 1997). Interestingly, segmentation butnot myogenesis inX. laeviswas also disturbed by functionalinterference with the Notch signaling pathway. This wasachieved by expression of antimorphic forms of either thesuppressor of Hairless (X-Su(H)), or of the X-Delta-2. Simi-lar to our observations, not only loss of function of theXenopusX-Delta-2, but also gain of function by overexpres-sion of the same protein resulted in segmentation defects(Jen et al., 1997).

3.3. Potential targets of xGCNF function

As a first approach to screen for potential targets ofxGCNF, we have analyzed several marker genes, whoseexpression is spatiotemporally regulated during embryogen-esis ofX. laevis, by means of RT-PCR. Due to the antero-posterior gradient of expression of the xGCNF mRNA itself,we dissected the embryos before mRNA extraction in orderto examine anterior and posterior anlagen separately.Indeed, dnGCNF interferes with the expression ofxRARg2 mRNA selectively in the anterior third of the tail-bud embryo. On the basis of the normal expression patternof xRARg2 and xGCNF in early tadpole embryos we sup-pose that xGCNF interferes with the expression of xRARg2

in those cells that normally express both receptors, particu-larly cells in the head mesenchyme (Fig. 1B) and not in cellsof the tailtip that do not express xGCNF in significantamounts. Anteriorly, xRARg2 is supposed to play a role inthe differentiation of the head neural crest cells and the headendomesoderm (Ellinger-Ziegelbauer and Dreyer, 1991,1993). It is likely that xRARg2 mediates the effects of exo-genous retinoids on the differentiation of the brain and senseorgans (Durston et al., 1989; Dreyer and Ellinger-Ziegel-bauer, 1996).

Retinoic acid-induced cell differentiation causes a transi-ent increase in GCNF transcription in PCC7-Mz1 (Bauer etal., 1997) and in P19 murine embryonic carcinoma cells(Heinzer et al., 1998). However, in embryos ofX. laevis,we observed no significant change in the expression ofxGCNF mRNA or protein after application of a dose ofretinoic acid sufficient to induce anterior defects (data notshown).

Putative GCNF response elements are found in the pro-moter regions of the murine genes for protamine 2 (Yan etal., 1997), butyrophilin (gb U67065), activin receptor II (gbM93400), the rat hepatocyte nuclear factor 1 gene (gbX67649), and the human genes for the platelet-derivedgrowth factor, PDGF, and for oxytocin (Su¨sens et al.,1997; Heinzer et al., 1998). We cannot yet determinewhether xRARg2 is a direct target of xGCNF, because thestructure of its promoter is not known. A promoter sequence

of oneX. laevisfibronectin gene is known (gb Y13284) anddoes not contain a DR0 motif. Therefore, the expression offibronectin could be indirectly regulated by GCNF. Theapparent excess of extracellular matrix material and theincrease in fibronectin mRNA observed after a gain ofxGCNF function indicates the regulation of cell to matrixinteractions as a potential function of GCNF in organogen-esis. The identification of direct GCNF target genes willprobably help to clarify the detailed mechanism leading tothe phenotypes observed.

3.4. xGCNF has a non-redundant function inorganogenesis

Our results on organogenic defects caused by loss ofGCNF function are in agreement with some observationsin GCNF knock-out mice. Complete loss of mGCNF func-tion leads to lethality between day 10.5 and 11.5 of embryo-nic development. Knock-out mice display defects inorganogenesis, including failures in closure of the neuraltube (Austin Cooney, personal communication).

In X. laevis, severe spina bifida type failures that dis-played after injection of dn or flGCNF mRNA to a variabledegree, dependent on the dose of RNA injected and on thebatch of eggs, were not evaluated in detail (Table 1). How-ever, failures in closure of the neural tube were frequentlyobserved on transverse sections of dnGCNF expressing spe-cimens without overt spina bifida (Fig. 7 and data notshown). We therefore suppose that failures in neural tubedifferentiation including necrotic areas containing cellsexpressing dnGCNF resulted from a specific loss of functionof the endogenous xGCNF (Fig. 7A,D). Undifferentiated ornecrotic cells with high levels of exogenous GCNF proteinthat had failed to integrate in tissue were also frequentlyfound in edema (data not shown).

Injection of mRNA encoding dnGCNF allowed us to gen-erate a spectrum of phenotypes, that are able to survivebecause the loss of function was only partial and was tran-sient. Preliminary observations on transgenic frogs indicatethat permanent expression of dnGCNF leads to lethality inembryos (M.S. Schwab and C. Dreyer, unpublished data).Embryonic lethality in the mouse and the distinct pheno-types ofX. laevisembryos after interference with xGCNFfunction suggest that GCNF is a nonredundant prerequisitefor the normal sequence of anteroposterior pattern forma-tion during organogenesis.

4. Experimental procedures

4.1. Plasmids and mRNA injection

PCR-fragments coding for amino-acids 1-435 (full lengthGCNF) and 55-435 (dominant negative GCNF) were gen-erated from the clone pBSKxGCNF (Joos et al., 1996) usingthe forward primers 5′ttgaattcGATGGACACATGGG-


AAG3′ and 5′ttgaattcGAACTGCGTTATGTCGC3′, res-pectively. The reverse gene-specific primer was5′ctgctctagaTGATCCAAGTAGCCTC3′. The resultingproducts were inserted into theEcoRI and XbaI sites ofthe expression vector pCS2+ MT (Rupp et al., 1994).These templates for mRNA synthesis are termed pCS2+MTxGCNF and pCS2+ MTxGCNFDN and encode pro-teins containing six N-terminal myc-epitopes plus the full-length or N-terminally truncated xGCNF.

A cDNA fragment coding for the enhanced green fluor-escent protein was amplified via PCR using the plasmidpEGFP-C1 (Clontech) as a template and the following pri-mers: forward, 5′cggaattcCATGGTGAGCAAGGGC-GAG3′, reverse, 5′tgaggatatc TTACTTGTACAGCTC-GTCCATGCC3′. The EcoRI/EcoRV digested PCR frag-ment was cloned into theEcoRI/SnaBI site of the pCS2+expression vector resulting in pCS2+ MTGFP.

The plasmids were linearized withNotI and 5′-cappedmRNA was synthesized with SP6 RNA polymerase usinga Message-Machine kit (Ambion). RNA (120 pg) diluted inH2O to a volume of 4 nl was microinjected into one or bothblastomeres at the 2-cell-stage, whereas blastomeres ofembryos at the 32-cell-stage were injected with 30 pg ofRNA diluted in 1 nl. The incidence of partial exogastrula-tion and spina bifida varied from one batch of eggs toanother and was significantly less frequent after injectionof water instead of mRNA (data not shown). As a control wetherefore injected a portion of each batch of embryos withthe same amount of mRNA encoding either myc-taggedenhanced green fluorescent protein (GFP) orb-galactosi-dase (b-gal). Embryos were kept in 1× MBSH until 2 to5h after injection and then in 0.1× MBSH until fixation.Staging of control embryos was performed according toNieuwkoop and Faber (1967). In some experiments, spinabifida phenotypes were considerably more severe afterinjection of mRNA encoding dn or flGCNF than after con-trol mRNA (data not shown).

4.2. Electrophoretic mobility shift assays (EMSA)

EMSA were performed as described in Joos et al. (1996).Templates for coupled mRNA and protein synthesis in vitrowere pSG5-xGCNF (1–1465) for full length xGCNF andpSG5-xGCNF (52–1465) for its N-terminally truncatedform.

4.3. Western blotting

A polyclonal mouse serum raised against a his-taggedfusion protein containing aa 69 to 435 of xGCNF (essen-tially domains D, E and F) was affinity-purified as describedpreviously (Joos et al., 1996) and antibodies directed againstthe six his-tags were removed by application of polyhisti-dine sepharose (Sigma). Protein extractions from embryosand Western blotting was carried out as described bySchwab and Dreyer (1997).

4.4. Immunostaining

For wholemount immunostaining, embryos were fixedin 4% paraformaldehyde in MEMFA buffer for 3 h atRT and stored in methanol at−20°C until further proces-sing. Before rehydration, embryos were transferred toDent’s fixative (20% DMSO in methanol) and kept at−20°C overnight. Wild type embryos designated forindirect immunoperoxidase staining were, in addition,bleached in 67% methanol, 10% H2O2, pH 8 underneon light. Embryos were gradually rehydrated in PBS,blocked for 1 h at RT in 20% bovine serum in PBS andsubsequently incubated with the primary antibody in 20%bovine serum in PBS for 5 to 24h at RT (anti myc-tagmAB 9E10 and rabbit anti-fibronectin). Peroxidase-coupled secondary antibodies were visualized by incubat-ing the embryos in 0.3 mg/ml DAB, 0.06% H2O2 in PBS.After examination of the embryos under the dissectionmicroscope, 50mm vibratome sections were prepared andexamined with Nomarski optics as described (Hollemannet al., 1996). Alternatively, embryos were embedded inTechnovit (Kultzer) after peroxidase staining and alter-nating 5 mm thin sections were stained with 0.05% tolui-dine blue for 1 min or left unstained to facilitatedetection of the peroxidase label. For double stainingwith mouse anti xGCNF and rabbit anti RARg (Ellin-ger-Ziegelbauer and Dreyer, 1993), affinity-purified anti-sera were diluted in 20% bovine serum in PBScontaining 0.05% Tween20 and 2.5% DMSO. Incubationwas extended to at least 40 h at RT, to ensure completepenetration. After extensive washing overnight, embryoswere incubated with the secondary antibody (Cy3-goatanti mouse IgG and DTAF-goat anti rabbit IgG for indir-ect immunofluorescence and peroxidase coupled goat antimouse IgG for peroxidase staining) for 5 or 24 h at RTand again washed overnight. For detection of the immu-nofluorescent label, embryos were embedded in Techno-vit (Kulzer, Germany), sectioned and counterstained withDAPI as described by Schwab and Dreyer (1997). Sec-ondary antibodies were from Jackson Immuno Research,West Grove, USA.

4.5. In situ hybridization

Albino embryos were injected with two times 120 pgdn or flGCNF mRNA plus 30 or 120 pg of GFP mRNAinto each blastomere at the two-cell-stage. At early gas-trula stage 10, embryos were selected for significant GFPexpression under UV light and fixed between stage 10.25and 11. Whole mount in situ hybridization was performedessentially as described (Joos et al., 1996), using digox-igenin-labeled antisense RNA specific of chordin, xbra,and myoDb. After photography, specimens were immu-nostained with mAB 9E10 for 24 h followed by perox-idase-coupled goat anti mouse IgG to detect myc-taggedproteins.


4.6. Microscopy and photography

Fluorescence-labeled Technovit-sections were analyzedunder epifluorescence illumination with a Zeiss Axioplanmicroscope (Zeiss, Oberkochen, FRG) and photographedwith Fuji Provia 400 slide film, or with a Sony 3CCDvideo camera using Analysis software (Soft Imaging Sys-tems, Munster, Germany). Whole mount peroxidase stainedembryos were cleared in 33.3% benzyl alcohol in benzoicacid and photographed at a Zeiss Axiovert microscope withNomarski optics or at a Zeiss Stemi SV 11. All color trans-parencies were digitized. Images were processed usingAdobe Photoshop and Aldus Freehand.

4.7. RNA preparation and RT-PCR analysis

Total RNA was isolated from whole embryos at stage 23,from anterior and posterior halves of embryos at stages 23and 27, and from anterior, middle and posterior thirds oftadpoles at stage 30, using TRIZOL reagent (Gibco, BRL).Analyzing thirds of embryos rather than whole embryosshould allow us to monitor more subtle changes thatmight escape detection, if mRNA from whole embryoswere subjected to RT-PCR. Material from five or six indi-viduals was pooled for each RNA preparation. QuantitativeRT-PCR was performed as described in Niehrs et al. (1994).The following primers were used: xRARg2: 5′CAGAGCC-CACACTGCTGC3′ and 5′GTCTGGTAGCACCACTTC-C3′; fibronectin: 5′TGACAATGCTCGTAGCTC3′ and 5′-GGTCAATAGGATGGCAAGAG3′; HoxB9: 5′TACTTA-CGGGCTTGGCTGG3′; and 5′GCGTGTAACCAGTTG-GCTG3′; EF1a: 5′CAGATTGGTGCTGGATATGC3′ and5′ACTGCCTTGATGACTCCTAG3′. The annealing tem-perature was 57°C for all primer pairs, and the number ofcycles resulting in an increase of the amount of the radio-active PCR-product by a factor of two per cycle was opti-mized for each primer pair (data not shown). For themultipex PCR reactions shown in Fig. 10, 24 cycles wereperformed. The product lengths were 552 bp for xRARg2,782 bp for fibronectin, 216 bp for HoxB9, and 279 bp forEF1a. The EF1a-specific cDNA was simultaneously ampli-fied in the same tube as the cDNA under investigation. Theradioactive PCR products were separated on 7.5% polyacry-lamide gels and were quantified using a phosphorimager(Molecular Dynamics). The intensity of the PCR-productunder investigation was routinely normalized to the internalEF1a-standard.

Acknowledgements

We wish to thank Drs. Sarah Crampton and HerbertSteinbeisser for many helpful suggestions on the manu-script, Brigitte Glaser, Metta Riebesell and Anne Schohlfor expert technical assistance, and Ulrike Goßweiler forhelp with preparing the figures. We also thank Dr. Heidrun

Ellinger-Ziegelbauer for anti RAR antisera, Drs. HerbertSteinbeisser and Ralph Rupp for anti-sense RNA probes,and Dr. Austin Cooney for communicating unpublishedresults.

References

Bauer, U.M., Hirsch, S.S., Reinhardt, S., Pauly, T., Maus, A., Wang, F.,Heiermann, R., Rentrop, M., Maelicke, A., 1997. Neuronal cell nuclearfactor: a nuclear receptor possibly involved in the control of neurogen-esis and neuronal differentiation. Eur. J. Biochem. 249, 826–837.

Borgmeyer, U., 1997. Dimeric binding of the mouse germ cell nuclearfactor. Eur. J. Biochem. 244, 120–127.

Chen, F., Cooney, A.J., Wang, Y., Law, S.W., O’Malley, B.W., 1994.Cloning of a novel orphan receptor (GCNF) expressed during germcell development. Mol. Endocrinol. 8, 1434–1444.

Cooke, J., 1978. Somite abnormalities caused by short heat shocks to pre-neurula stages ofXenopus laevis. J. Embryol. Exp. Morphol. 45, 283–294.

Dreyer, C., Ellinger-Ziegelbauer, H., 1996. Retinoic acid receptors andnuclear orphan receptors in the development ofXenopus laevis. Int. J.Dev. Biol. 40, 255–262.

Durston, A.J., Timmermanns, J.P., Hage, W.J., Hendriks, H.F., DeVries,N.J., Heidefeld, M., Nieuwkoop, P.D., 1989. Retinoic acid causes ananterioposterior transformation in the developing central nervoussystem. Nature 340, 140–144.

Ellinger-Ziegelbauer, H., Dreyer, C., 1991. A retinoic acid receptorexpressed in the early development ofXenopus laevis. Genes Dev. 5,94–104.

Ellinger-Ziegelbauer, H., Dreyer, C., 1993. The pattern of retinoic acidreceptor gamma (RARg) expression in normal development ofXenopuslaevisand after manipulation of the main body axis. Mech. Dev. 41, 33–46.

Elsdale, T., Davidson, D., 1983. Somitogenesis in amphibia. IV. Thedynamics of tail development. J. Embryol. Exp. Morphol. 76, 157–176.

Essex, L.J., Mayor, R., Sargent, M.G., 1993. Expression ofXenopussnailin mesoderm and prospective neural fold ectoderm. Dev. Dynam. 198,108–122.

Gont, L.K., Steinbeisser, H., Blumberg, B., de Robertis, E.M., 1993. Tailformation as a continuation of gastrulation: the multiple cell populationsof theXenopustailbud derive from the late blastopore lip. Development119, 991–1004.

Gossler, A., Hrabe de Angelis, M., 1997. Somitogenesis. Curr. Topics Dev.Biol. 38, 225–287.

Heinzer, C., Su¨sens, U., Schmitz, T.P., Borgmeyer, U., 1998. Retinoidsinduce differential expression and DNA binding of the mouse Germ CellNuclear Factor in P19 embryonal carcinoma cells. Biol. Chem. 379, inpress.

Hirose, T., O’Brien, D.A., Jetten, A.M., 1995. RTR: a new member of thenuclear receptor superfamily that is highly expressed in murine testis.Gene 152, 247–251.

Hollemann, T., Schuh, R., Pieler, T., Stick, R., 1996.XenopusXsal-1, avertebrate homolog of the region specific homeotic gene spalt ofDrosophila. Mech. Dev. 55, 19–32.

Jen, W.C., Wettstein, D., Turner, D., Chitnis, A., Kintner, C., 1997. TheNotch ligand, X-Delta-2, mediates segmentation of the paraxial meso-derm inXenopusembryos. Development 124, 1169–1178.

Joos, T.O., David, R., Dreyer, C., 1996. xGCNF, a nuclear orphan receptoris expressed during neurulation inXenopus laevis. Mech. Dev. 60, 45–57.

Kapelle, M., Kratzschmar, J., Husemann, M., Schleuning, W.D., 1997.cDNA cloning of two closely related forms of human germ cell nuclearfactor (GCNF). Biochim. Biophys. Acta 1352, 13–17.

Katz, D., Niederberger, C., Slaughter, G.R., Cooney, A.J., 1997. Charac-


terization of germ cell-specific expression of the orphan nuclear recep-tor, germ cell nuclear factor. Endocrinology 138, 4364–4372.

Lee, G., Hynes, R.O., Kirschner, M., 1984. Temporal and spatial regula-tion of fibronectin in earlyXenopusdevelopment. Cell 36, 729–740.

Lei, W., Hirose, T., Zhang, L.X., Adachi, H., Spinella, M.J., Dmitrovsky,E., Jetten, A.M., 1997. Cloning of the human orphan receptor germ cellnuclear factor/retinoid receptor-related testis-associated receptor and itsdifferential regulation during embryonal carcinoma cell differentiation.J. Mol. Endocrinol. 18, 167–176.

Mathews, L.S., Vale, W.W., Kintner, C.R., 1992. Cloning of a second typeof activin receptor and functional characterization inXenopusembryos.Science 255, 1702–1705.

Mayor, R., Morgan, R., Sargent, M.G., 1995. Induction of the prospectiveneural crest ofXenopus. Development 121, 767–777.

Niehrs, C., Steinbeisser, H., De Robertis, E.M., 1994. Mesodermal pattern-ing by a gradient of the vertebrate homeobox gene goosecoid. Science263, 817–820.

Nieuwkoop, P.D., Faber, J., 1967. Normal table ofXenopus laevis. NorthHolland, Amsterdam.

Ogryzko, V.V., Schiltz, R.L., Russanova, V., Howard, B.H., Nakatani, Y.,1996. The transcriptional coactivators p300 and CBP are histoneacetyltransferases. Cell 87, 953–959.

Pfeffer, P.L., De Robertis, E.M., 1994. Regional specificity of RARgamma isoforms inXenopusdevelopment. Mech. Dev. 45, 147–153.

Rupp, R.A., Weintraub, H., 1991. Ubiquitous MyoD transcription at themidblastula transition precedes induction-dependent MyoD expressionin presumptive mesoderm ofX. laevis. Cell 65, 927–937.

Rupp, R.A.W., Snider, L., Weintraub, H., 1994.Xenopusembryos regulatethe nuclear localization of XMyoD. Genes Dev. 8, 1311–1323.

Sasai, Y., Lu, B., Picolo, S., De Robertis, E.M., 1996. Endoderm inductionby the organizer-secreted factors chordin and noggin inXenopusanimalcaps. EMBO J. 15, 4547–4555.

Schwab, M.S., Dreyer, C., 1997. Protein phosphorylation sites regulate thefunction of the bipartite NLS of nucleolin. Eur. J. Cell Biol. 73, 287–297.

Sharpe, C.R., Fritz, A., De Robertis, E.M., Gurdon, J.B., 1987. A homeo-box-containing marker of posterior neural differentiation shows theimportance of predetermination in neural induction. Cell 50, 749–758.

Smith, J.C., Price, B.M., Green, J.B., Weigel, D., Herrmann, B.G., 1991.Expression of aXenopushomolog of Brachyury (T) is an immediate-early response to mesoderm induction. Cell 67, 79–87.

Susens, U., Borgmeyer, U., 1996. Characterization of the human germ cellnuclear factor gene. Biochim. Biophys. Acta 1309, 179–182.

Susens, U., Aguiluz, J.B., Evans, R.M., Borgmeyer, U., 1997. The germcell nuclear factor mGCNF is expressed in the developing nervoussystem. Dev. Neurosci. 19, 410–420.

Yamaguchi, T.P., 1997. New insights into segmentation and patterningduring vertebrate somitogenesis. Curr. Opin. Genet. Dev. 7, 513–518.

Yan, Z.H., Medvedev, A., Hirose, T., Gotoh, H., Jetten, A.M., 1997.Characterization of the response element and DNA binding propertiesof the nuclear orphan receptor germ cell nuclear factor/retinoid recep-tor-related testis-associated receptor. J. Biol. Chem. 272, 10565–10572.

Zimmer, F.J., Dreyer, C., Hausen, P., 1988. The function of the nuclearenvelope in nuclear protein accumulation. J. Cell Biol. 106, 1435–1444.


Anteroposterior patterning and organogenesis of Xenopus laevis ...

Documents

Transcript of Anteroposterior patterning and organogenesis of Xenopus laevis ...