SEARCH

SEARCH BY CITATION

Keywords:

  • Xenopus;
  • TGF-β, activin, BVg1, Xbrachyury;
  • goosecoid;
  • chordin;
  • cycloheximide

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

In Xenopus, activin-like signals are able to induce and pattern mesoderm in a concentration-dependent manner. Previous experiments demonstrated that discrete gene expression patterns can be formed in animal cap explants as a response to graded activin signals. We analyzed the spatiotemporal appearance of goosecoid (gsc), chordin (chd), and Xbrachyury (Xbra) mRNAs in whole Xenopus embryos ectopically expressing activin or BVg1. To discriminate between direct transcriptional regulation and indirect, protein synthesis-dependent effects of ectopic signals, we combined overexpression studies and cycloheximide treatment. Our experiments revealed long-range signaling of activin/BVg1, but the expression patterns of gsc, chd, and Xbra in response to activin/BVg1 indicated that repressors are essential to establish the proper expression of these genes. Analysis of endogenous gsc, chd, and Xbra transcript distribution in embryos treated with cycloheximide supported this concept. We, therefore, conclude that inhibition is fundamental during early embryonic patterning. Developmental Dynamics 233:418–429, 2005. © 2005 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

During Xenopus development, transforming growth factor-beta (TGF-β) signaling molecules such as activin, Vg1, the nodal-related proteins and derriere are playing pivotal roles in pattern formation (Harland and Gerhart, 1997; Heasman, 1997; Green, 2002; Smith and White, 2003). A crucial property of these molecules is that they exert dose-dependent effects on cell fate and gene expression. Moderate amounts of activin, for example, induce the expression of the mesodermal T-Box gene Xbra, whereas high amounts repress Xbra and activate the organizer gene goosecoid (Green et al., 1992; Gurdon et al., 1994, 1996). Experiments with activin-loaded microbeads placed onto animal cap explants indicated the existence of an activin gradient that emanates from the bead and regulates gene expression in a spatially defined pattern (for reviews, see McDowell and Gurdon, 1999; Gurdon and Bourillot, 2001; Green, 2002). In this so-called activin/gsc/Xbra system, the dose-dependent effect of activin was even quantified: for a given cell that harbors approximately 5,000 activin receptors, 100 occupied receptors result in the transcription of Xbra, whereas 300 occupied receptors lead to the induction of goosecoid expression (Dyson and Gurdon, 1998).

These results suggest graded activities of activin-like growth factors regulating the expression of gsc and Xbra along the vegetal–animal axis in the intact early embryo. They do, however, not explain how the level of activin signaling is interpreted by the cells. Microbead experiments indicated that a short time (2 hr) after bead application the cells express both gsc and Xbra in the same region, whereas later on (5 hr after bead application), the expression domains of these two genes appear separated (Papin and Smith, 2000). This separation of the gsc and Xbra expression domains depends on protein synthesis, strongly suggesting the presence of newly synthesized Xbra inhibitors in regions of high activin signaling activity (Papin and Smith, 2000). Goosecoid can repress Xbra expression (Latinkic et al., 1997; Latinkic and Smith, 1999), but additional Xbra repressors have to be postulated, because overexpression of a dominant-negative VP16-gsc does not affect Xbra repression in animal cap explants treated with activin-loaded microbeads (Papin and Smith, 2000). These experiments indicate that a putative activin gradient would only roughly induce a “pattern” of Xbra and gsc, which has to be refined by additional factors that inactivate Xbra. Whether similar inhibitory inputs are necessary for shaping the expression domains of other genes such as gsc or chd, which both respond to high levels of activin, remained elusive.

Most of the previous experiments dealing with the properties of the activin/gsc/Xbra system were performed in dispersed cells, in conjugates of different explants, or in isolated animal caps and, therefore, outside of the systemic influence of the embryo. An alternative approach is the microinjection of mRNAs coding for activin-like signals into an animal cap cell of the 8- to 16-cell embryo. As a result, a circular field of ectopic bottle cells is formed in the outer animal epithelium, which is surrounded by concentric gene expression domains of gsc in the vicinity and Xbra in the periphery (Kurth and Hausen, 2000). In such embryos, the formation of ectopic and endogenous expression patterns can be directly compared. It is also possible to analyze the effects of overlapping ectopic signal gradients, which gives additional information about the properties of the proposed gradients. Furthermore, possible long distance effects of the ectopically expressed signal molecules can be monitored all over the embryo. In this study, we used activin or BVg1 as inducing activities. Despite the different regulation of activin and BVg1 (see Discussion section), both proteins signal through the same receptors (Schulte-Merker et al., 1994) and induce similar concentration-dependent patterns of bottle cell formation and gsc/Xbra expression (Kurth and Hausen, 2000). We analyzed (1) the interactions of two distinct signaling centers, and (2) the establishment of ectopic and endogenous expression patterns of chd, gsc, and Xbra in normal and cycloheximide (CHX) -treated embryos between blastula and early gastrula stages.

Our results indicate that discrete mesodermal gene expression patterns in the embryo cannot be explained by direct responses of the cells to graded signaling activities of activin or BVg1. The data implicate that maternal and zygotic inhibitors are essential for the formation of correct gene expression patterns in the early gastrula. Therefore, we propose repression as a major mechanism during early pattern formation.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Endogenous and Experimentally Induced Patterns of Bottle Cell Formation and Gene Expression Suggest the Activity of Gradients of Activin-Like TGF-βs

At early gastrula stages, the expression patterns of mesodermal genes such as Xbrachyury (Xbra), chordin (chd), and goosecoid (gsc) are characterized by specific spatial relationships. Xbra marks the mesodermal mantle in the marginal zone, and its expression domain leaves a small gap to the region where bottle cells form (Fig. 1A). On the dorsalmost side, this gap represents the organizer region harboring, among others, transcripts of chd (Fig. 1A) and gsc (data not shown).

thumbnail image

Figure 1. A–E: Expression profiles of mesodermal genes in the vicinity of endogenous and experimentally induced bottle cells. A: Expression of Xenopus brachyury (Xbra, blue) and chordin (chd, magenta) in early gastrulae. chd transcripts indicate the organizer region, which fits into the gap between the dorsal bottle cells (arrow) and the Xbra-positive mesodermal ring. BE: Microinjection of activin mRNA causes the formation of an ectopic bottle cell field in the animal cap and the expression of chd (B), Xbra (C), and gsc (E) in characteristic spatial relationships to the ectopic bottle cells (schematically illustrated in D). Note the “bridge” of gsc-positive cells between the endogenous organizer and the ectopic gsc domain in E. F–H: Shifts of expression domains in regions of overlapping gradients of activin. F: Schematic drawing illustrating the theoretically postulated shifts of the domains I–III caused by different degrees of overlap of two signal gradients. A peripheral overlap leads to the formation of an additional domain II (1), a more extensive overlap to the expansion of domain I (2), and to the fusion of two bottle cell fields in the outer epithelium (3). G: Double-injections of activin mRNA into animal blastomeres of stage 6–6.5 embryos were performed at different distances. G1,2: Two complete Xbra-rings are formed. In the center of the animal cap, the two rings are fused. G3,4: Less distance between the injection sites causes the formation of a uniform Xbra-negative region encompassing both bottle cell fields. This region (corresponding now to region I) is surrounded by a single Xbra “ring.” The two bottle cell fields become fused under these conditions. H: Single injections at decreasing distances to the marginal zone (1). H2: At a distance, the ectopic Xbra domain forms a complete ring. The periphery of this ring (region III) fuses with the periphery of the endogenous Xbra ring (arrowheads) to form a Xbra-positive bridge. H3,4: More pronounced overlaps of endogenous and ectopic signal gradients cause partial (3) or complete (4) reduction of the endogenous Xbra expression, eventually causing an Omega-like Xbra pattern (4, compare with E). Arrowheads indicate the endogenous expression sites, and asterisks indicate the ectopic bottle cell fields, which sometimes can also be recognized by the accumulated brown pigment (C,G2,G4). I–M: Cell movements in embryos overexpressing activin in the animal cap. I: Activin mRNA was coinjected with fluorescein-dextran-amine (FDA), and animal caps were explanted at stage 8. KM: They were cultured either alone (K) or in combination with an uninjected cap (L,M, the border between the two caps is roughly indicated by the dashed yellow line). At stage 10.5, they were fixed, embedded in plastic, and sectioned. FDA-positive epithelial cells undergo apical constriction (indicated by the curved arrows in K and M), and the stained cells underneath form a coherent mass. Arrowheads in K indicate some FDA-negative bottle cells (identified by their accumulated pigment) directly adjacent to the FDA-positive cell mass. M: In conjugates, only the injected cap forms bottle cells. N–P: Similar activin-induced distribution patterns of labeled cells appear in undissected embryos. N: Tangential section through an ectopic bottle cell field. Arrows indicate the bottle cell field at the center of the image; the asterisk indicates a region of constricted but FDA-negative cells. O,P: Cross-sections through bottle cells at the center (O) or at the periphery (P) of an ectopic bottle cell field. Note the FDA-negative bottle cells in P where cell borders were visualized with an antibody against β-catenin. Scale bars = 100 μm in K,M, 50 μm in N–P.

Download figure to PowerPoint

Local overexpression of activin causes the formation of an ectopic field of bottle cells with adjacent concentric domains of gene expression (Kurth and Hausen, 2000; Fig. 1B–E). The organizer genes chd and gsc are expressed at the center, in regions of high activin signaling activity (Fig. 1B,D,E), whereas Xbra is expressed at the periphery, in regions of moderate activin signaling (Fig. 1C,D). The endogenous and ectopic gsc domains in Figure 1E are connected to each other by a “bridge” of gsc-expressing cells, suggesting that the two signaling centers have combined their peripheral activities to induce the high activin responsive gsc gene there. This pattern may rely on the activities of gradients emanating from both the injection site and the dorsal-vegetal cells.

To address this effect in more detail, additional injection experiments were performed. We intended to create overlaps of activin gradients emanating from ectopic and endogenous signaling centers (Fig. 1F–H). In a first series of experiments, embryos were injected twice in the animal region of stage 6–6.5 embryos with decreasing distances between the injection sites (Fig. 1G). Alternatively, activin mRNA was injected at various distances from the marginal zone (Fig. 1H). Dependent on the degree of overlap of endogenous and ectopic signaling centers, different effects on the expression of mesodermal genes would be expected. We used Xbra expression as a read-out, which represents domain II in the schematic diagrams depicted in Figure 1D,F. Cells in domains III do not receive sufficient signal molecules to induce Xbra. If, however, the signal intensity is raised by an overlap of two signaling centers, ectopic Xbra expression could be induced (Fig. 1F1). Such an additive effect can indeed be observed in our experiments (Fig. 1H2). If two signaling centers overlap more extensively, the signal intensity would eventually reach levels incompatible with Xbra induction (Fig. 1F2, domain I). Such a reduction of Xbra-expression in regions of intense overlap is depicted in Figures 1G4 and 1H3,4. In the outer epithelium, an intense overlap would lead to the fusion of the bottle cell fields (Fig. 1F3, see Fig. 1G4 for initial bottle cell field fusion). Similar shifts of gene expression or bottle cell formation in overlapping gradients were also observed after local overexpression of BVg1 or Xnr1 (data not shown).

Bottle cell formation can be incorporated into the activin-gsc-Xbra system as an epithelial response to high levels of activin (Kurth and Hausen, 2000). To correlate apical constriction with activin production in our assay, the lineage tracer fluorescein-dextran-amine (FDA) was coinjected together with activin mRNA into an animal cell of 8- to 16-cell embryos (Fig. 1I–P). At stage 10.5 the distribution of FDA-labeled cells was assayed in isolated single caps (Fig. 1I,K) and in conjugates of labeled and unlabeled caps (Fig. 1L,M). It became evident that apical constriction is not completely restricted to labeled epithelial cells. Unlabeled bottle cells in the vicinity of the FDA-positive cells (Fig. 1K) indicate that amounts of activin sufficient for bottle cell induction may have diffused over several cell diameters (arrowheads). Similar results were obtained from undissected embryos after injection of activin mRNA and FDA (Fig. 1N–P). In conjugates of injected and uninjected caps, however, bottle cell formation is not induced in the noninjected cap. Presumably the distance prevents the appropriate signal intensity to be established in the distant epithelial cells (Fig. 1M).

CHX Treatment of Whole Embryos and Characterization of the Resulting Phenotypes at Gastrula Stages

The expression patterns presented above seem to be consistent with the concept of activin-like signals forming gradients in the embryo. However, similar patterns may be caused by a relay mechanism. Furthermore, it is not clear how the proposed gradients are interpreted by the cells to create discrete gene expression domains.

For a further analysis of these issues, we used CHX to investigate the spatiotemporal expression profiles in situ under conditions of reduced protein synthesis. The experiments were performed in embryos overexpressing activin or BVg1 in the animal half. With this approach, the difference between direct and indirect effects of TGF-β signals on gene expression should become clear, and it should be possible to discriminate between relay and direct long-range signaling, because a relay based on the induction of secondary effectors would be sensitive to CHX.

We transferred stage 8 blastulae before the onset of zygotic transcription for 30 to 180 min to a culture medium containing 10 μg/ml CHX, a concentration typically leading to a 90–95% drop in protein synthesis, as deduced from several previous studies (Cascio and Gurdon, 1987; Sokol, 1994; Papin and Smith, 2000; Clements and Woodland, 2003). After this treatment, embryos were placed back in normal culture medium (Fig. 2A). We first questioned the effectiveness of the method on zygotic protein production and analyzed the expression of the zygotic cell adhesion protein E–cadherin (Choi and Gumbiner, 1989; Angres et al., 1991) in comparison to the maternally supplied β-catenin (Schneider et al., 1993, 1996) at stage 11.5. In normal embryos, E-cadherin is clearly detectable at this stage (Fig. 2B). CHX treatment severely interferes with the expression of this protein after 90 min of exposure, and only reduced levels of E-cadherin are detectable after 30 or 60 min of CHX exposure. The amount of the maternal protein β-catenin is moderately reduced after 30, 60, or 90 min of CHX exposure and further decreases after longer treatments. It is, however, still detectable even after 3 hr of CHX treatment (Fig. 2B). At stage 13, E-cadherin is present in CHX-embryos treated for 90 min but is only weakly expressed after 2 hr of CHX treatment (Fig. 2C). In both cases, the expression level is much lower than in control embryos, indicating that protein synthesis has not reached normal levels at this stage. These protein expression data have been corroborated by double immunostainings of E-cadherin and β-catenin (data not shown). At early neurula stages (st. 15), the staining intensities of E-cadherin and β-catenin have reached normal levels in CHX-treated embryos (data not shown). Taken together, the delay of E-cadherin production between 3 and 5 hr is consistent with the previous notion that protein synthesis is restored in animal caps 5 hr after CHX withdrawal (Cascio and Gurdon, 1987). Furthermore, CHX embryos displayed no gastrulation movements until controls reach mid-gastrulation (several hours after the CHX pulse, data not shown). Later on, the CHX embryos gastrulate and enter neurulation, although some aspects of development appear abnormal (data not shown, see also Fig. 4). All subsequent experiments were done with 90-min CHX treatments, because shorter exposures were not effective enough and longer ones often led to cell death.

thumbnail image

Figure 2. Cycloheximide (CHX) treatment of whole embryos. A: At stage 8, normal and activin/BVg1-overexpressing embryos were transferred to culture medium containing CHX for different periods of time. They were transferred back to normal culture medium and scored for protein expression and phenotypes at gastrula stages. TGF-β, transforming growth factor-beta. B: Western blot of zygotic E-cadherin (E-cad, top) and maternal β-catenin (β-cat, bottom) at stage 11.5. Embryos treated with CHX for approximately 90 min or longer are devoid of E-cadherin, whereas β-catenin can still be detected after 180 min of CHX exposure. co, control. C: Simultaneous antibody stainings of E-cadherin and β-catenin in normal and CHX embryos at stages 11.5 and 13. When control siblings had reached stage 13, considerable E-cadherin expression can be detected after 90 min CHX, whereas only weak expression is detected after treatment for 120 min. D–G: Phenotypes of gastrulae overexpressing BVg1 and treated with CHX. D: BVg1-overexpressing embryo at stage 10.5, vegetal view. Arrows indicate dorsal bottle cells. E: Same embryo, animal view. Ectopic bottle cells are indicated by arrowheads. F: BVg1-injected and CHX-treated embryo, vegetal view; no indication of endogenous lip formation. G: Same embryo, animal view; no ectopic bottle cells are visible. The cells of the CHX embryo are much larger than those of the control due to the CHX-typical cell cycle arrest. H,I: Animal–vegetal sections through BVg1/fluorescein-dextran-amine (FDA)-injected (H) and BVg1/FDA-injected and CHX-treated (I) embryos; overlays of phase contrast and fluorescent images. H: The embryo displays the typical distribution of labeled cells, presumably caused by apical constriction of the outer epithelial cells. The deeper cells form a coherent mass, which somehow joins this epithelial movement (indicated by arrows, compare with Fig. 1). I: After CHX treatment, no apical constriction movement is visible and the embryo has a blastula-like phenotype. The green bars indicate the borders of BVg1 producing cells; the dashed arrows and the blue bars indicate the minimum range of signal diffusion as deduced from the in situ data disclosed in Figure 3. an, animal; bc, blastocoel; veg, vegetal.

Download figure to PowerPoint

thumbnail image

Figure 4. Phenotypes at the tadpole stage. AE: Control (A), and after 30 (B), 90 (C), 120 (D), and 180 (E) min of cycloheximide (chx) treatment. C,D,E: The CHX-treated embryos are viable but lack axial especially anterior structures, indicating a patterning defect.

Download figure to PowerPoint

thumbnail image

Figure 3. Emergence of ectopic activin or BVg1-induced expression patterns of chd, gsc, and Xbra in the absence or presence of cycloheximide (CHX). A–I: mRNAs were injected at the eight-cell stage (activin, A–C; BVg1, D–I), and the embryos were subjected to in situ analysis of chd (A,B), gsc (D,E), and Xbra (G,H); the early gastrula expression patterns are schematically illustrated in (C,F,I) (lateral views). A,D,G: At stage (st) 9, the patterns look similar for all three genes. B,C,E,F,H,I: The typical patterns disclosing chd and gsc transcripts at or underneath the bottle cell field and Xbra transcripts in a ring at the periphery appear at early gastrula. KU: Embryos injected at the eight-cell stage were subsequently treated for 90 min with CHX and subjected to in situ analysis of chd (K,L), gsc (N–P), and Xbra (R–T), and their early gastrula patterns were schematically summarized (M,Q,U; lateral views; dark blue, strong expression, light blue, weaker expression). K,N,R: At stage 9, the expression patterns are similar to each other (K,N,R) as well as to the patterns in non–CHX-treated embryos (compare with A,D,G). From early gastrulation onward, expansion of all expression domains is observed. L: Chd is strongly expressed in a large dorsal sector and slightly weaker in the animal cap and parts of the vegetal region (dorsal [do], animal [an], and vegetal [veg] views of the same embryo). O,P: Ectopic and endogenous gsc domains expand (O) and eventually become fused at stage 10.5 (P, animal view). P: The gsc expression additionally expands irregularly to the vegetal half (vegetal view of the same embryo). S,T: The Xbra expression domains do also expand (S, animal view), become fused, and eventually cover the whole animal half (T, animal view); the vegetal half is devoid of Xbra transcripts (T, vegetal view of the same embryo). Q,U: In contrast to the embryo in L (chd), the dorsal sides of the embryos in P (gsc) and T (Xbra) cannot be identified with certainty (? in Q and U). Arrowheads indicate endogenous expression domains and, in the case of gsc and chd, the dorsal side. Asterisks mark the ectopic bottle cell fields in B,C,E,F,H,I and ectopic gene expression sites in A,D,G,K–U. ven, ventral.

Download figure to PowerPoint

We next analyzed the effects of CHX treatment on the morphology of embryos injected with BVg1 mRNA. Here, the situation was different from that of the uninjected embryo. High amounts of signal molecules can be produced hours before mid-blastula transition (MBT) as a result of the mRNA injection. A typical BVg1-induced ectopic bottle cell field in the animal cap is shown in Figure 2E (see endogenous lip of the same embryo in Fig. 2D). CHX embryos injected with BVg1 mRNA display neither ectopic nor endogenous lip formation (Fig. 2F,G), indicating a complete block of BVg1-induced apical constriction. Figure 2F,G also illustrates the arrest of cell proliferation as another characteristic aspect of CHX embryos (see, e.g., Clements and Woodland, 2003). Therefore, the cells remain much larger than in controls.

Cell movements during gastrulation were further assayed by coinjecting FDA and BVg1. In this case, similar distribution patterns of injected cells were observed as in activin/FDA-injected embryos (Fig. 2H, compare with Fig. 1K,M). Embryos additionally treated with CHX, however, show a blastula-like phenotype and no condensed field of labeled cells (Fig. 2I), indicating a block of all BVg1-induced cell movements.

In summary, these experiments indicate that the whole-mount CHX treatment is effective. They further demonstrate that the effect is not lethal and that it is reversible after transferring embryos back to normal culture medium (see also Fig. 4).

CHX Treatment Causes Superinduction of chd, gsc, and Xbra in Response to Activin or BVg1

How do the expression patterns of mesodermal genes such as chd, gsc, and Xbra form in such morphogenetically inactive embryos? For solving this problem, we analyzed the spatiotemporal expression patterns of these genes in embryos injected with activin or BVg1 mRNAs with and without CHX treatment (Fig. 3). In stage 9 blastulae (without CHX), the ectopic expression domains of all three genes are homogenously stained spots in the animal half (Fig. 3A,D,G). At this time point, Xbra expression is not inhibited in the area of high BVg1 signaling activity. Two hours later, however, at early gastrulation, the typical ectopic expression patterns with gsc and chd at the center and Xbra at the periphery became established (Fig. 3B,C,E,F,H,I). CHX-treated and injected embryos display similar spot-like expression patterns at stage 9 (Fig. 3K,N,R). At gastrula stages, however, dramatic differences appear compared with the situation in embryos cultured without CHX. Chd transcripts can be found all over the embryo, with the exception of a small area in the vegetal hemisphere. The most conspicuous staining, however, is still confined to the prospective dorsal side (Fig. 3L,M). Ectopic and endogenous gsc expression domains both expand, become fused, and finally the animal half and marginal zone are strongly stained (Fig. 3O–Q). Moreover, the expression expands in an irregular manner far into the vegetal half (Fig. 3P, right). The ectopic Xbra expression domain also increases dramatically and completely covers the animal half and marginal zone at stage 10.5 (Fig. 3S–U). Hence, at this time point chd, gsc and Xbra expression domains are completely overlapping in these regions and display no patterns that are consistent with graded responses to the ectopic signals. In contrast to chd and gsc transcripts, which are also synthesized in the vegetal half (Fig. 3L,P), Xbra is not expressed in that region (Fig. 3T), indicating differences in the regulation of these genes in the vegetal hemisphere (fast zygotic or maternal inhibition of Xbra, slow zygotic inhibition of chd, gsc).

In summary, the expression of gsc, chd, and Xbra is strongly induced in CHX embryos, and the expression domains are expanded dramatically, indicating long distance signaling (compare with Fig. 2I). Therefore, the main CHX effect is a lack of inhibition rather than of induction. Furthermore, inhibition of gene expression in the embryo is crucial not only for the proper activation of Xbra but also for the regulation of the organizer genes chd and gsc. Obviously, activin-like signals do not only induce a pattern of activation but also, maybe more importantly, a pattern of repression. Finally, we observe differences between the negative regulation of Xbra and gsc/chd.

Emergence of Expanded Endogenous Gene Expression Patterns in CHX-Treated Embryos

At tadpole stages, CHX embryos display phenotypes that indicate patterning defects. They show truncated axes and lack anterior structures after exposures to CHX lasting longer than 60 min (Fig. 4). Furthermore, histological analysis of control and CHX-treated gastrulae indicated that the gastrulation movements in CHX embryos are delayed and not properly coordinated (data not shown). The in situ analysis of embryos overexpressing activin or BVg1 implicates that a lack of represssion at early gastrula stages might be responsible for these morphogenetic defects that ultimately lead to headless embryos (see above). In another series of in situ hybridization experiments, therefore, we analyzed the expression patterns of chd, gsc, and Xbra in uninjected CHX-treated embryos from blastula to gastrula. Transcripts of these genes can be first detected on the dorsal side of the embryo around stage 9 (Fig. 5A,D,G). In contrast to the other two genes chd displays a large dorsal expression domain (Fig. 5A), which very likely depends on early β-catenin signaling and marks the blastula chordin- and noggin-expressing region (BCNE; Wesseley et al., 2001; Kuroda et al., 2004). At early gastrulation (st. 10–10+), chd, gsc and Xbra transcripts are distributed in the known characteristic patterns (Fig. 5B,C,E,F,H,I). Initially, the expression domains of these genes overlap significantly and become separated at stage 10 (confirmed by double in situ hybridizations, data not shown; see also Artinger et al., 1997).

thumbnail image

Figure 5. Spatiotemporal expression of chd, gsc, and Xbra in early control and cycloheximide (CHX)-treated embryos. A: In control stage (st) 9 blastulae, chd is expressed in a large area on the prospective dorsal side. B: At the onset of gastrulation, the organizer region is stained. D,E: Early gsc staining is detected in the dorsal vegetal–marginal region (D) and later in the organizer (E). G,H: Xbra transcripts also appear first in the dorsal marginal zone (G) and later mark the mesodermal ring (H). C,F,I: These early gastrula patterns are schematically illustrated in C,F, and I, respectively (lateral views). K,O,R: In CHX-treated embryos, the early chd, gsc, and Xbra expression domains are similar to the control. L,M: At stage 10+, however, the chd pattern resembles an slightly expanded stage 9 pattern. L: Note for example the chd-positive vegetal cells. M: The lateral–ventral regions of the animal cap remain chd-negative. P: In stage 10.5 embryos, gsc is strongly expressed in the dorsal marginal zone, weaker staining can be seen in lateroventral regions (left, curved arrows; vegetal view) and in the animal cap (right; animal view). S,T: At the same time, Xbra is expressed dorsally in the same region. It appear slightly expanded animalward when compared with the blastula pattern, but strong signal is still confined to the dorsal side of the embryo. S: Weak signal can be seen in lateral and in animal regions. S,T: Vegetal and ventral areas seem to be devoid of Xbra transcripts. N,Q,U: The CHX-“gastrula” patterns are schematically illustrated in N, Q, and U, respectively (lateral views; dark blue, strong expression, light blue, weaker expression). Arrowheads indicate the endogenous expression sites, the arrows in B,E,H indicate the dorsal lip. an, animal; do, dorsal; lat, lateral; veg, vegetal; ven, ventral.

Download figure to PowerPoint

In CHX embryos, the initial gene expression domains resemble those in control embryos (Fig. 5K,O,R). In many embryos, the stage 9 Xbra expression is rather weak but never absent. This finding could reflect either the activity of a maternal or a pre–MBT-activated TGF-β component, or it could indicate some residual protein synthesis sufficient for initial induction of nodal expression by VegT (Kofron et al., 1999). At stage 10–10.5, all three genes show expanded expression domains. Chd transcripts form an even larger dorsal field and can also be found in vegetal cells (Fig. 5L,N). Animal cap cells outside this domain do not express chd (Fig. 5M,N). Gsc expression expands to lateroventral regions in the marginal zone and to the animal hemisphere, the strongest signal intensity being found on the dorsal side (Fig. 5P,Q). Strong Xbra expression remains confined to the dorsal side, but it expands to the animal half of the embryo where weaker staining can be observed (Fig. 5S,U). The lateral marginal expression site, however, is weak, and the ventral expression is missing (Fig. 5T,U), indicating that zygotic activating factors are needed for the lateroventral expansion of Xbra expression. Later, at midgastrula stages, the gsc and Xbra patterns are similar to the early gastrula patterns in normal embryos, but both appear slightly expanded to the animal hemisphere (data not shown). It appears that, after an initial expansion of gsc and Xbra expression (e.g., to the animal hemisphere), which is probably caused by a lack of inhibitory factors, the patterns are later reduced roughly to the normal situation. Taken together, the data support the idea that inhibition is a major component of the early developmental program.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Propagation of Ectopic Activin/BVg1 Signals in the Embryo

The idea that morphogen gradients regulate major events during embryogenesis is highly attractive for developmental biologists (Wolpert, 1969). Among the embryonic processes that are thought to be dependent on such morphogen activities are the development of the wing imaginal disc in Drosophila or the mesoderm formation in Xenopus (for a recent review, see Gurdon and Bourillot, 2001). The Xenopus animal cap explant in combination with growth factor-loaded microbead implantation turned out to be a powerful tool for studying the basic mechanisms of gradient formation and interpretation. An alternative approach is the microinjection of mRNAs coding for activin-like signals into single animal cap cells and in situ analysis of responsive genes in the intact embryo (Kurth and Hausen, 2000). Although protein gradients created by mRNA microinjections may differ from those emanating from microbeads, the microinjection approach results in gene expression patterns similar to those obtained in microbead experiments (Kurth and Hausen, 2000; this report). With a series of elegant experiments in zebrafish that combined such single cell injections with cell transplantations and the use of mutants, Chen and Schier (2001) recently demonstrated that the zebrafish nodal-related protein squint behaves as a bona fide morphogen.

Our in situ hybridization analysis after single and double injections of activin mRNA revealed patterns of Xbra expression that could well be interpreted as a dose-dependent response of the cells to overlapping activin gradients emanating concentrically from the injection sites (Fig. 1). Dependent on the degree of overlap, concentration-dependent shifts of Xbra expression domains were observed, a property theoretically postulated for a classic morphogen (Gurdon and Bourillot, 2001). Similar patterns, however, could be achieved by a relay mechanism. In fact, we observed that overexpression of the transcription factor VegT, which acts cell autonomously, induced Xbra patterns highly reminiscent to those induced by activin, probably through the induction of nodals and derriere (Kofron et al., 1999; Onuma et al., 2002) (unpublished observation). This scenario would represent a classic relay mechanism. On the basis of these experiments, it cannot be excluded that activin and BVg1 work in a similar manner, and indeed, it has been shown that some TGF-βs such as TGF-β1, Xnr2, and Cyclops exert only short-range effects (Reilly and Melton, 1996; Jones et al., 1996; Chen and Schier, 2001).

A relay mechanism would depend on protein synthesis. Local overexpression of activin/BVg1 in combination with CHX treatment revealed, however, long-distance induction of gene expression even in the absence of protein synthesis. This result indicates long-range effects without additional CHX-sensitive relay mechanisms at least for activin and BVg1. Lineage tracing further demonstrated that, after microinjection of BVg-1 followed by CHX treatment, only a part of the animal cap contained growth factor producing cells, when control siblings had reached midgastrulation (see Fig. 2I). BVg1-injected CHX embryos of this age, however, displayed strong Xbra staining all over the animal cap and the marginal zone, and the expression zones of gsc and chd, two genes that were activated only at higher doses, expanded even further into the vegetal hemisphere, leaving only a small area free of gsc and chd transcripts. Therefore, substantial amounts of the exogenous signaling molecules may diffuse to distant regions in the embryo.

In our CHX experiments, chd expression appears to be a direct response to TGF-β signals, which is in contrast to previous observations in animal cap explants (Sasai et al., 1994). Although we cannot resolve this discrepancy, it should be mentioned that different assays were used: activin application to animal cap explants (Sasai et al., 1994) in contrast to microinjection of activin mRNA (this study).

Interpretation of Activin-Like Signal Gradients: The Role of Inhibitors

As discussed above, it seems plausible that activin and BVg1 are able to signal over long distances. The mechanisms, however, by which cells interpret the local amounts of signal molecules are still not well understood. On one hand, it appears clear that cells can measure a threefold difference in the levels of bound activin and react accordingly by differential gene expression (Dyson and Gurdon, 1998). On the other hand, it has been shown for the inhibition of Xbra in regions of high activin levels that the information “No. of occupied receptors” is transferred only in a protein synthesis-dependent manner (Papin and Smith, 2000; this study). Inhibition as a crucial mechanism for the regulation of Xbra in the embryo became evident from experiments with embryos transgenic for Xbra variants, which were mutated in the regulatory regions of their promotors. In such embryos, the resulting lack of specific negative regulation coincided with expression of Xbra at ectopic locations (Lerchner et al., 2000).

Our results show that CHX-sensitive inhibition is not restricted to Xbra but also affects gsc and chd, two organizer genes that are both induced by high levels of activin-like signals. In CHX embryos ectopically expressing activin or BVg1, gsc, chd, and Xbra are strongly induced and their ectopic expression domains fuse with the endogenous expression sites to form large, uniform and overlapping domains.

Several conclusions can be drawn from these observations. First, the coinduction of gsc, chd, and Xbra indicates that, above a critical level of signal intensity, all three genes become expressed simultaneously in a given cell and that no distinct gene expression domains form in such embryos. Second, CHX-sensitive repressors obviously act in the embryo to restrict the expression sites of gsc, chd, and Xbra. Part of this inhibitory influence may be explained by factors regulating the expression of TGF-β–responsive genes. This interpretation is supported by experiments showing that Goosecoid, Otx-2 (Pannese et al., 1995), Mix.1 (Rosa, 1989), and SIP1 (Verschueren et al., 1999) repress transcription of Xbra (Latinkic et al., 1997; Latinkic and Smith, 1999; Verschueren et al., 1999; Lerchner et al., 2000). Alternatively, the production and/or activity of the TGF-β signals themselves could be inhibited. Xnr1, activin, or BVg1 induce the production of their own competitors. Cerberus specifically inhibits Xnr1 (Agius et al., 2000), follistatin acts on activin (Schulte-Merker et al., 1994), and Lefty acts on BVg1/Xnr1 (Chen and Shen, 2004; Cheng et al., 2004). These factors could restrict the intensity and propagation of TGF-β signals in the embryo. Downstream signaling may be further regulated by inhibitory Smads and Smad ubiquitin ligases (Smurfs), which ultimately leads to degradation of Smad complexes or the TGF-β receptors (Kavsak et al., 2000; Lin et al., 2000). These inhibitory influences probably cooperate during ectopic pattern formation. On the basis of the in situ patterns in CHX embryos overexpressing activin/BVg1, we postulate that the TGF-β–dependent repressors that directly or indirectly down-regulate gsc, chd, and Xbra are missing under these conditions.

A lack of these repressors could also account for the expansion of early gene expression domains in uninjected CHX embryos. In addition, a different type of inhibition in the animal hemisphere might be deduced from the expression profiles in those embryos. Although BVg1 or activin is not active in the animal cap, we noted ectopic expression of gsc and Xbra there, which is, however, much weaker than the superinduction observed in microinjected embryos. This finding argues for the existence of TGF-β–independent zygotic repressors that down-regulate basic activities of gsc and Xbra, an interpretation consistent with the growth factor-independent induction of gsc and Xfkh-1 in CHX-treated animal cap explants (Dawid et al., 1993; Tadano et al., 1993; Sokol, 1994).

Finally, there are obvious differences in the regulation of the high-level activin-responsive genes chd and gsc, and the low-level responsive gene Xbra in the vegetal hemisphere of the embryo. Chd and gsc transcripts are found far in the vegetal half, whereas Xbra is not expressed there and its expression domain discloses a sharp border to the vegetal hemisphere. Different turnover rates of the various inhibitors could be responsible for these differences. Another possible explanation could be that a maternal inhibitor is limiting Xbra expression in the vegetal half, whereas newly synthesized zygotic factors are needed to down-regulate chd and gsc in that region. Figure 6 presents a hypothetical summary of some of the different inhibitory activities that might act along the animal–vegetal axis on the dorsal side of the embryo.

thumbnail image

Figure 6. Hypothetical model depicting possible inhibitory influences in different regions of the early Xenopus gastrula. The drawing provides a dorsal view of a stage 10 gastrula and focuses on the formation of distinct gastrula expression patterns (Xbra, chd, gsc) along the animal–vegetal axis. The dark gray area represents the organizer, the thick line indicates the dorsal bottle cells, and the hatched area in the marginal zone represents the Xbra-expressing mesoderm. Some of the hypothesized inhibitory activities are incorporated into the drawing: 1, zygotic inhibitors of basic expression of Xbra and gsc; 2, zygotic inhibitors of nodal signal propagation (e.g., Lefty); 3, zygotic inhibition of Xbra by Gsc; 4, zygotic inhibitors of the expansion of gsc and chd expression to the vegetal hemisphere; 5, zygotic and/or maternal (?) inhibitors of the expansion of Xbra expression to the vegetal hemisphere. For further explanations, see text.

Download figure to PowerPoint

Taken together, our observations implicate the existence of a wide variety of inhibitory activities in early pattern formation (TGF-β–dependent vs. TGF-β-independent, animal vs. vegetal, maternal vs. zygotic [?]). This inhibitory diversity implicates a complex network of inhibition comprising signal inhibitors (e.g., cerberus, follistatin, lefty, smurfs) as well as transcriptional repressors (e.g., gsc, Mix.1, SIP1). The methodology described in this report could serve as a valuable prerequisite for functional screens to unravel this inhibitory network.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Embryos, Microinjections, and Whole-Mount CHX Treatment

Adult Xenopus laevis were purchased from the African Xenopus Facility C.C. (South Africa). Embryos were obtained by in vitro fertilization as described previously (Fey and Hausen, 1990), cultured in 0.1 MBSH (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO3, 0.82 mM MgSO4, 0.41 mM CaCl2, 0.33 mM Ca(NO3)2, 10 mM HEPES (pH 7.4), 10 μg/ml streptomycin sulfate and penicillin), and staged according to Nieuwkoop and Faber (1967).

mRNA injections (5 nl per injection) were performed in 2/3 MBSH containing 4% Ficoll-400 (Sigma, Munich, Germany). Approximately 50–100 pg of activin mRNA or BVg1 mRNA were injected. After injection, the embryos were left in Ficoll solution for 2–3 hr and then transferred to 0.1 MBSH for further cultivation.

To inhibit protein synthesis at the onset of zygotic transcription (MBT), embryos were transferred at stage 8 to a medium containing 10 μg/ml CHX for a period of 90 min. At this concentration, protein synthesis is typically reduced to 5–10% (Cascio and Gurdon, 1987; Sokol, 1994; Papin and Smith, 2000; Clements and Woodland, 2003). After that step, the embryos were fixed either at stage 9 or cultivated further in 0.1 × MBSH and fixed at gastrula stages.

Expression Constructs and mRNA Synthesis

cDNA plasmids containing the full-length sequences of the following genes were used for overexpression experiments: activin (pBSK, restriction with EcoR1, transcription with T3-RNA-Polymerase; Thomsen et al., 1990), bvg1 (pSP64T3, restriction with EcoR1, transcription with SP6-RNA-Polymerase; Thomsen and Melton, 1993), and Xnr1 (pSP64T, restriction with SmaI, transcription with SP6-RNA-Polymerase; Jones et al., 1995). Capped sense mRNAs for microinjections were prepared with T3 or SP6 mMessage mMachine kits (Ambion) according to the manufacturer's instructions. The RNA was resuspended in sterile TE buffer (1 mM ethylenediaminetetraacetic acid, 10 mM Tris-HCl pH 8.0) and quantified by comparing fluorescence intensity to a marker with a known concentration on an ethidium bromide-stained agarose gel. Aliquots were stored at −70°C and diluted in sterile injection buffer (88 mM NaCl and 15 mM Tris-HCl, pH 7.5 in sterile water).

In Situ Hybridization

Whole-mount single and double in situ hybridizations were performed as described previously (Harland, 1991; Kurth and Hausen, 2000). Digoxigenin (DIG)-labeled and fluorescein-labeled antisense mRNAs were synthesized using T7 or SP6-RNA-Polymerases. The following plasmids were used as templates for the transcription of antisense riboprobes: goosecoid (H7 full-length clone in pBSK, restriction with EcoR1, transcription with T7-RNA-Polymerase; Blumberg et al., 1991; Cho et al., 1991), Xbra (pSP73, restriction with SalI, transcription with SP6-RNA-Polymerase; Smith et al., 1991), and chordin (Δ59, restriction with EcoR1, transcription with T7-RNA-Polymerase; Sasai et al., 1994).

Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis and Western Blotting

Protein extraction was performed as described previously (Müller et al., 1994). Proteins were separated in a 7% acrylamide gel and blotted onto a nitrocellulose membrane. Uniformity of loading was assessed by staining the blot with 0.2% Ponceau S (Sigma). Blotted proteins were stained with anti–E-cadherin (mouse monoclonal 10H3, Angres et al., 1991) and anti–β-catenin (rabbit polyclonal P14L, Schneider et al., 1993, 1996). Primary antibodies were detected by peroxidase coupled secondary goat anti-mouse or goat anti-rabbit antibodies (Dianova). The signal was detected on X-ray films using the Renaissance chemiluminescence detection system (NEN).

Lineage Tracing and Animal Cap Explants

To analyze movements and distribution of cells ectopically expressing activin or BVg1, 0.5% FDA (Molecular Probes) was coinjected together with the corresponding mRNAs into an animal blastomere of the 8- to 16-cell stage. Whole embryos were cultured until stage 10.5 and the distribution of the progeny of injected cells was analyzed in histological preparations. For the characterization of cell movements in animal caps, these were explanted from injected embryos at stage 8 and cultivated until stage 10.5. Some of the injected caps were conjugated with noninjected caps from control embryos (see Fig. 1L), and the conjugate was cultured until stage 10.5. Single caps or conjugates were finally fixed and subjected to histological analysis as described in the next section.

Histology and Immunostaining

Embryos, caps, and conjugates were fixed in MEMFA (4% formaldehyde, 0.1 M MOPS pH 7.4, 2 mM ethyleneglycoltetraacetic acid, 1 mM MgSO4) for 2 hr at room temperature or overnight at 4°C followed by post-fixation and facultative storage in ice-cold methanol. Specimens were washed in methanol at room temperature, infiltrated, embedded in glycolmethacrylate (Technovit 7100, Kulzer, Wehrsheim, Germany), and cut into 5-μm-thick serial sections for fluorescence microscopy. Immunofluorescent whole-mount labeling was performed as described previously (Kurth et al., 1999; Kurth, 2003) using the following primary antibodies: P14L (polyclonal rabbit antibody against β-catenin; Schneider et al., 1993, 1996), and a mouse monoclonal anti–E-cadherin antibody (10H3, Angres et al., 1991). As secondary antibodies, Cy3-coupled goat anti-rabbit IgGs (Dianova) and Alexa 488-coupled goat anti-mouse IgGs (Molecular Probes) were applied.

Microscopy, Image Acquisition, and Processing

Plastic sections were analyzed either with a Zeiss Axioplan microscope or an Olympus BH 2 microscope both equipped with epifluorescence optics. Selective filters for Cy3- and Alexa-fluorescent signals (AF Analysentechnik, Germany) were used. Histological and fluorescence data were acquired using a digital camera (Sony) and the Analysis program (Soft Imaging Systems). Alternatively, a Coolpix 4500 digital camera (Nikon) was used. Fluorescent and phase contrast images for lineage analysis were captured and processed using the Analysis (Soft Imaging Systems) or the SPOT and MetaView programs. In situ samples were analyzed with a Zeiss stereomicroscope, and images were acquired using the Coolpix 4500 digital Camera (Nikon). Picture processing and figure mounting were performed with the Photoshop (Adobe, version 5.0) and Freehand (Macromedia, version 8) programs.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Andrea Belkacemi and Christina Schmidt for excellent technical assistance, Francois Fagotto and Gerald Thomsen for the kind donation of constructs, and H.-H. Epperlein for valuable comments on the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES