SEARCH

SEARCH BY CITATION

Keywords:

  • Cactus;
  • Drosophila melanogaster;
  • Dorsal;
  • I-κB;
  • Rel proteins;
  • Xenopus laevis;
  • dorsoventral patterning

Abstract

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

In Drosophila, the Toll/Dorsal pathway triggers the nuclear entry of the Rel protein Dorsal, which controls dorsoventral patterning in early embryogenesis and plays an important role in innate immunity of the adult fly. In vertebrates, the homologous Toll/IL-1 receptor signaling pathway directs the nuclear localization of Rel/NF-κB complexes, which activate genes involved in proliferation, apoptosis, and immune response. Recently, first evidence has been reported for the activity of vertebrate Rel proteins and a Toll-like signaling pathway in the dorsoventral patterning process of Xenopus laevis embryos. Given the evolutionary divergence of the fly and frog model organisms, these findings raise the question, to what extent the effector functions of this pathway have been conserved? Here, we report the ability of two Xenopus Rel proteins to partially substitute for several, but not all, functions of the Dorsal protein in Drosophila embryos. Our results suggest the interaction between Rel proteins and their cytoplasmic inhibitors as an important interface of evolutionary adaptation. Developmental Dynamics 235:949–957, 2006. © 2006 Wiley-Liss, Inc.


INTRODUCTION

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

The NF-κB/Rel proteins are dimeric transcription factors that mediate the activity of a conserved signal transduction pathway, the Drosophila Toll/Dorsal and the vertebrate Toll/IL-1 receptor signaling cascade (Huguet et al., 1997; Anderson, 2000; Stathopoulos and Levine, 2005). Normally Rel proteins are retained in the cytoplasm by inhibitory proteins of the IκB family, which mask their nuclear localization signal (NLS). The ubiquitin-dependent degradation of the IκB proteins upon the activation of Toll/IL-1 receptors and subsequent intracellular signaling components finally releases the Rel complexes to enter the cell nucleus and regulate target genes. The Rel protein family was named after the v-rel oncogene, which had been isolated from the turkey retrovirus REV-T. All transcription factors of this family share a 300 amino acid domain, the Rel homology domain (RHD), which harbors the regions responsible for dimerization, DNA binding, nuclear localization and interaction with the IκB proteins. Rel proteins bind so-called κB DNA sites as homo- or heterodimers. Different dimers exhibit different affinities for distinct κB sites and are also preferentially bound by certain IκB inhibitors (Huguet et al., 1997; Gosh et al., 1998).

Several Rel proteins have been identified in various organisms (Huguet et al., 1997). Vertebrate Rel proteins are involved in the activation of a large number of genes that play a role in proliferation, stress, inflammation, and immune responses (Hayden and Gosh, 2004). In Drosophila, Rel proteins activate genes that are crucial both for response to infections and early development (Anderson, 2000; Stathopoulos and Levine, 2002). The dorsoventral (DV) polarity of the Drosophila embryo is determined by a graded nuclear localization of the Drosophila Rel protein Dorsal (Moussian and Roth, 2005). The highest concentration of nuclear Dorsal protein defines the future ventral side of the embryo. Dorsal is ubiquitously distributed in the syncytial embryo but is kept in the cytoplasm by its inhibitor Cactus, a homolog of the vertebrate IκB proteins. The degradation of Cactus and the resulting nuclear transport of Dorsal on the ventral side of the embryo is initiated by an extracellular cascade of proteases, which lead to the processing of the ligand Spätzle. The mature Spätzle ligand binds the transmembrane receptor Toll, which triggers through intracellular components the phosphorylation of Cactus. This phosphorylation event targets Cactus for ubiquitin-dependent degradation (Bergmann et al., 1996; Reach et al., 1996; Liu et al., 1997). The transcription factor Dorsal then represses on the ventral side target genes such as decapentaplegic (dpp) and zerknüllt (zen) required for dorsal ectoderm specification and activates in parallel target genes such as twist, snail, and short gastrulation (sog) required for mesoderm and neuroectoderm specification (Stathopoulos and Levine, 2002). The embryo is thereby subdivided into three different territories along the DV axis (presumptive mesoderm, neuroectoderm, and dorsal ectoderm). The Dorsal-dependent activation of sog further counteracts the activity of Dpp protein resulting in a Dpp activity gradient, which subdivides the dorsal ectoderm (for review, see Raftery and Sutherland, 2003).

In Drosophila, further proteins with a RHD were identified, Relish and the Dorsal-related immunity factor Dif (Ip et al., 1993; Dushay et al., 1996). Recently, the ability of Dif to substitute for Dorsal protein was reported (Stein et al., 1998). Dif, which is most closely related to Dorsal (49% identity, 64% similarity in the RHD), becomes activated upon infection and controls genes of the innate immune response of Drosophila (Meng et al., 1999; Rutschmann et al., 2000). Homologs of Dorsal were also found in other insects. In the beetle Tribolium castaneum, a homolog of Dorsal is also present at the time of the very early embryogenesis; however, it appears to have a less direct role in DV patterning (Chen et al., 2000).

Recent studies demonstrated that a vertebrate protein, the Xenopus protein Chordin, which is related to Drosophila Sog, inhibits the vertebrate Dpp homolog BMP4 in a comparable manner. Heterologous experiments with Xenopus and Drosophila embryos demonstrated that Sog and Chordin can substitute for each other. Sog and Chordin inhibit the transforming growth factor-β molecules Dpp and bone morphogenetic protein (BMP) 4 by direct binding (Sasai et al., 1995; Ferguson, 1996). Although these proteins exhibit evolutionary conserved activities, their expression domains within the embryos are inverted. In Drosophila, DPP defines the dorsal part of the embryo, whereas BMP4 activity is required for the establishment of the ventral side of the Xenopus embryo. These findings support the hypothesis that an inversion of the DV body axis occurred between arthropods and chordates during evolution (Sasai et al., 1995; Ferguson, 1996). An alternative hypothesis, proposed recently by Meinhardt (2002) suggests that vertebrates and arthropods diverted from a Hydra-like animal, but diverged by developing different strategies for midline formation. Distinguishing between these hypotheses will require a detailed understanding of the commonalities and differences in the use of apparently conserved regulatory cascades in the different phyla. The maternal signals that control the zygotic expressions of Sog/Chordin and Dpp/BMP4 have so far been thought to be completely different. Recent evidence, however, indicated that in Xenopus a Toll/IL-1 signaling pathway also contributes to early axial patterning (Armstrong et al., 1998; Prothmann et al., 2000). These similarities raise the general question to which extent signal transduction pathways and their target genes are conserved throughout evolution. The Drosophila Toll/Dorsal and its vertebrate counterpart, the Toll/IL-1 receptor signaling pathway, represent a well-characterized transcription control module, which is an excellent model to investigate the coupling of signal transduction pathways to conserved and/or diverged gene functions.

One approach to address this topic is to perform cross-species experiment to reveal the degree of coupling between transcriptional control and conserved signaling events. Previous work demonstrated that mammalian cis-regulatory elements can be recognized by the Drosophila Dorsal protein, indicating that enhancers can couple conserved signaling to divergent gene functions (Gonzalez-Crespo and Levine, 1994). Given the structural and functional parallels between the fly and vertebrate pathways, we set out to ask if vertebrate Rel proteins can substitute for the Dorsal protein functions in early Drosophila embryogenesis. Five different Rel proteins have been identified to date in Xenopus laevis, all of which are maternally expressed (Kao and Hopwood, 1991; Mishina et al., 1995; Suzuki et al., 1995; Tannahill and Wardle, 1995; Inukai et al., 1998; Suzuki et al., 1998; Yang et al., 1998). To investigate the ability of Xenopus Rel proteins to compensate for Dorsal functions, we performed microinjections of Xenopus Rel mRNAs into dl−/− embryos. We show that Xrel1 and Xrel2, but not XrelB, are able to rescue ventral structures in completely dorsalized embryos. They give rise to a lateralized phenotype, which exhibits one distinct value of the DV gradient of Dorsal protein activity around the entire embryo. In the dl−/− embryos, these Xenopus Rel proteins activate and repress known targets of Dorsal. Our observations, thus, demonstrate that Xrel1 and Xrel2 proteins can partially substitute for Dorsal in principal but fail to sense the spatial cue of the Spätzle/Toll signal input.

RESULTS

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

Phylogenetic Sequence Comparison of Rel Proteins

The Rel/NF-κB transcription factors can be subdivided into four subfamilies due to sequence comparison and phylogenetic distance trees (Huguet et al., 1997; Friedman and Hughes, 2002). The largest group encompasses RelA (p65), RelB/Irel, C-Rel/vRel, Dorsal, and Dif. The second group includes the NF-κB1(p50) and NF-κB2(p52) groups. The third group is represented solely by Drosophila Relish. The transcription factors of the fourth group, the NF-AT genes, also contain an RHD but are rather considered as an outgroup (Huguet et al., 1997; Gosh et al., 1998).

Here, we focus on the first group that also contains Tribolium Dorsal. None of the RHDs of the Xenopus Rels has a significant high similarity to the RHD of Drosophila Dorsal (see Fig. 1). It is not possible, therefore, to identify a Dorsal homolog based on overall sequence similarity in the RHD. Furthermore it has to be considered that Dorsal is active as a homodimer during embryogenesis, whereas vertebrate Rels of the first group rather form heterodimers within the group or with members of the second group (e.g., p65/p50; see Huguet et al., 1997; Gosh et al., 1998). Taking the results of the sequence analysis together, we decided to focus on experiments that might reveal a functional ortholog of the Dorsal protein.

thumbnail image

Figure 1. Phylogenetic comparison of insect (Drosophila melanogaster: Dm Dorsal and Dm Dif; Tribolium castaneum: Tc Dorsal) and vertebrate Rel proteins (Xenopus laevis: Xrel 1, 2, 3, and B; Homo sapiens: Hs Rel A, Hs C-Rel, and I-Rel [Hs Rel B]; Mus musculus: Mm Rel A, Mm C-Rel, and Mm Rel B). The sequences were aligned using Clustal method (MacVector), and the tree was constructed using the neighbor-joining algorithm (MacVector: best tree, uncorrected P, gaps distributed proportionally).

Download figure to PowerPoint

Xenopus Rel Proteins Have Ventralizing Activities in Drosophila Embryos

The Drosophila embryo is patterned along the DV axis by the nuclear gradient of Dorsal protein and, thereby, subdivided into ectoderm and mesoderm. During further development, tissues derived from ectodermal regions secrete the cuticle of the first-instar larva (see Fig. 2). To assess the activities of Xenopus Rel proteins, we started by injecting synthetic XRel1 mRNA into wild-type embryos. The resulting cuticles showed different degrees of ventralization as seen from the extension of ventral denticles toward the dorsal side of the embryo (data not shown). We then investigated whether Xrel1 mRNA could rescue the dorsal mutant phenotype, which results in a complete dorsalization of the larval cuticle (Fig. 2B). The mRNA of Drosophila dorsal served as control to evaluate the extent of rescue, which can be achieved in this experiment. dorsal mRNA injected into completely dorsalized dl−/− embryos clearly led to the formation of ventral structures. The phenotypes ranged from a weak rescue as indicated by partially dorsalized embryos to a completely restored wild-type pattern and even slightly ventralized embryos (Fig. 2C; Table 1). Injection of Xrel1 mRNA into completely dorsalized dl embryos also restored ventral structures (Fig. 2D) such as ventral denticles, Filzkörper, and antennal sense organ. The mRNA of Xrel2, which is closely related to Xrel3, resulted in a similar rescue of cuticular structures (Table 1, and data not shown). However, XrelB was not able to rescue, even at comparatively high doses (Table 1, and data not shown). Mammalian RelBs are known to require NF-κB1(p50) for heterodimerization (Gosh et al., 1998), suggesting that the observed inactivity of RelB might be explained by the lack of a suitable dimerization partner present in the fly.

thumbnail image

Figure 2. Injection of Xenopus Xrel1 mRNA into dl−/− embryos partially rescues ventral cuticular structures. A–D: Darkfield photographs of cuticle preparations. A: A differentiated wild-type cuticle with ventral structures such as denticles and Filzkörper. B: The cuticle of a dorsalized dlH/Df(2L)TW119 embryo. C: This phenotype can be fully rescued by injection of Drosophila Dorsal mRNA (0.25 μg/μl). D: A partially rescued cuticle of a dlH/Df(2L)TW119 embryo injected with Xenopus Xrel1 mRNA (0.25 μg/μl). The cuticle exhibits circumferential ventral denticles and Filzkörper-like structures.

Download figure to PowerPoint

Table 1. Injection of mRNAs (as Listed Above) Into dlH/Df(2L)TW119 Embryosa
mRNAμg/μlexp.CuticlesD0 [%]L3 [%]L2 [%]L1 [%]  Rescue [%]
  • a

    On average, 50 embryos were injected per single experiment. The number of differentiated cuticles varies from mRNA to mRNA and exhibits a concentration dependency. Classification of the strength of rescue of the completely dorsalized dlH/Df(2L)TW119 embryos after injection with the following: Xrel mRNAs, scoring of apolar rescue (Roth et al., 1991): D0, complete dorsalization; L3, antennal sense organ present; L2, Filzkörper +/- antennal sense organ; L1, ventral epidermis (denticles) +/- L2 features; Dorsal mRNA (Anderson et al., 1985): D0, complete dorsalization; D1, strong dorsalization (Filzkörper), no ventral epidermis; D2, weak dorsalization; ventral epidermis (no mesoderm); D3, as D2 + partial mesoderm; Wt, V4, weak ventralization.

Xrel10.255824761829  53
 0.1125721684  28
 0.05246781570  22
 0.01121100000  0
Xrel20.25282505025  75
 0.05274803314  20
 0.0113197300  3
XrelB1.25247100000  0
    D0 [%]D1 [%]D2 [%]D3 [%]Wt [%]V4 [%]rescue [%]
Dorsal0.75116006315013100
 0.2522832814448397
 0.111285025170092
 0.051153367000067

Notably, Xrel1- and -2-injected embryos exhibited ventral structures but no wild-type body shape. On the contrary, these embryos still had an elongated body characteristic for a lateralized phenotype in which mesoderm formation at the ventral side of the embryo is compromised (Anderson et al., 1985; Roth et al., 1991). Such embryos lack polarity and express a particular positional level of the DV pattern elements around the entire circumference. In summary, Xenopus Rel1 and Rel2 can promote the formation of ventral structures of the Drosophila embryo. However, they appear to lack the ability to induce a polar sequence of pattern elements along the DV axis.

Xrel1 RNA-Microinjections Lead to a Lateralized Gastrulation Phenotype

The analysis of the cuticles of Xrel mRNA-injected dl−/− embryos led us to the assumption that the rescue phenotype is apolar. To confirm this conclusion, we used the morphogenetic movements that occur during gastrulation as sensitive indicators for polarity of injected embryos. In wild-type embryos, the invagination of the mesoderm occurs through a furrow at the ventral side of the embryo (ventral furrow, Fig. 3A). When the germband extends, the posterior midgut primordium carrying the pole cells moves to the dorsal side of the embryo. In front of the extending germband the dorsal folds become visible. At the anterior of the embryo cephalic furrows form in lateral positions. The fully dorsalized dl−/− embryos do not exhibit any ventral or cephalic furrow and lack a polarized extension of the germband (Fig. 3B). The dorsal folds expand around the entire circumference of the embryo. In Xrel1 mRNA-injected dl−/− embryos we could not observe any movement resulting in the formation of the ventral furrow, which suggests the absence of mesoderm formation at the ventral side of the embryo (Fig. 3C). Nevertheless, these embryos obviously formed a symmetric cephalic furrow, indicating an expansion of lateral fates around the entire circumference. The pole cells remained at the posterior pole during germband extension, indicating a lack of polarity of the gastrulation movements. Together, these findings show that injection of Xrel1 mRNA into dl−/− embryos induces structures derived from lateral positions of the axis but does not induce mesoderm derived from ventralmost positions. Furthermore, we conclude that Xrel1 cannot polarize the Drosophila embryo, despite the rescue of some dorsovental pattern elements.

thumbnail image

Figure 3. Injection of Xenopus Xrel1 mRNA into dl−/− embryos leads to a lateralized gastrulation phenotype. Living embryos were photographed with anterior to the left and dorsal up. A: The wild-type embryo shows a normally polarized gastrulation with dorsal folds, asymmetric cephalic furrow, and anterodorsal movement of the posterior plate (pole cells). B: The dlH/Df(2L)TW119 embryo exhibits an apolar gastrulation indicated by circumferential dorsal folds and no movement of the posterior plate. C: The injection of Xenopus Xrel1 (0.25 μg/μl) in a dlH/Df(2L)TW119 embryo causes a lateralized gastrulation phenotype that is characterized by the presence of a symmetric cephalic furrow (Roth et al., 1991). The posterior plate shows no movement, as indicated by the position of the pole cells.

Download figure to PowerPoint

Xenopus Xrel Proteins Can Repress dpp and Activate Known Target Genes of Dorsal

Dorsal activates or represses in a concentration-dependent manner zygotic target genes responsible for the formation of distinct cell types along the DV axis. These target genes can be grouped into three classes according to their response toward different levels of nuclear Dorsal (Stathopoulos and Levine, 2004). Type I target genes are activated by peak levels of Dorsal and, therefore, can only be expressed in the ventralmost 20% of the embryonic circumference. To this group belong twist (twi) and snail (sna), which are required for mesoderm specification and for the early morphogenetic events of ventral furrow formation. Type II targets are activated by intermediate levels of Dorsal and are repressed by sna. They are, therefore, expressed in ventrolateral stripes abutting the presumptive mesoderm. A typical type II target gene is rhomboid (rho), which contributes to the patterning of the neuroectoderm. Finally, the type III target genes sense even low amounts of nuclear Dorsal protein. Some are activated by Dorsal, like short gastrulation (sog), which is expressed in broad later domains. Others, however, are repressed by Dorsal, which restricts their expression to the dorsal 40% of the embryonic circumference. The most important member of the latter group is decapentaplegic (dpp), which is required for the specification of the dorsal (non-neurogenic) ectoderm and amnioserosa. We wondered, which of these genes may also be regulated by the heterologous Xenopus Rel1 and Rel 2 proteins?

In dl−/− embryos, dpp is expressed around the entire circumference of the embryo (Fig. 4). The injection of Xrel1 or Xrel2 mRNA into dl- embryos caused a repression of dpp expression in the central area of the embryo. The fully dorsalized dl−/− embryos are devoid of sog expression (Fig. 4). Upon injection with either Xrel1 or Xrel2 mRNAs, we could observe a strong activation of sog expression around the entire circumference of the embryo. We note that both the repression of dpp and the activation of sog appeared to be weaker after Xrel2 mRNA injections than after Xrel1 injections.

thumbnail image

Figure 4. Injections of Xenopus Xrels mRNA into dl−/− embryos alter the expression of zygotic target genes of Dorsal protein. Whole-mount RNA in situ hybridizations and immunological staining reactions were carried out with probes for the indicated RNA probes, or Twi antibody, as indicated to the left. Embryos at the cellular blastoderm stage were oriented with anterior to the left and dorsal side up. Columns from left to right: Wild-type embryos, dorsalized dlH/Df(2L)TW119 embryos, Xenopus Xrel1 mRNA (0.25 μg/μl) -injected dlH/Df(2L)TW119 embryos and Xenopus Xrel2 mRNA (0.05 μg/μl) -injected dlH/Df(2L)TW119 embryos. Probes/antibody: first row: dpp. Dorsal expression in wild-type embryos and overall expression in dlH/Df(2L)TW119 embryos. Xrel1- and -2-injected dlH/Df(2L)TW119 embryos show significantly reduced expression of dpp in the central region of the embryo where the injection occurred. Second row: sog. Ventral expression in wild-type embryos and no expression in dlH/Df(2L)TW119 embryos. Xrel1- and -2-injected dlH/Df(2L)TW119 embryos exhibit high expression of sog. Third row: Twist. Ventral expression of Twist protein in wild-type embryos and no expression in dlH/Df(2L)TW119 embryos. In Xrel1- and -2-injected dlH/Df(2L)TW119 embryos, Twist protein can be detected around the dorsoventral (DV) circumference in a distinct area near the injection site where the highest concentrations the Xrel mRNAs were likely to be. Fourth row: sna. Ventral expression in wild-type embryos, no expression in dlH/Df(2L)TW119 and Xrel1-injected dlH/Df(2L)TW119 embryos. Xrel2-injected dlH/Df(2L)TW119 embryos show faint sna expression close to the injection site. Fifth row: rho. Rhomboid exhibits its complex expression pattern in wild-type embryos by showing an anterior “head” stripe, a dpp-dependent dorsally located and a dorsal-dependent ventral expression domain. In dlH/Df(2L)TW119 embryos, the ventral expression domain disappears and the two dorsal ones spread circumferentially around the DV axis of the embryo. In Xrel1- and -2-injected dlH/Df(2L)TW119 embryos, the dpp-dependent dorsal expression of rho is almost ablated, but its ventral expression domain is not restored.

Download figure to PowerPoint

Embryos lacking Dorsal activity do not show any expression of twi and no ventral expression of sna. The injection of Xrel1 or Xrel2 mRNA into dl−/− embryos led to an activation of twi expression and the production of Twist protein (Fig. 4). In this case, Xrel2 mRNA appeared to be more potent than Xrel1 mRNA. This difference between both Xrels was further corroborated by the exclusive expression of sna in Xrel2 mRNA-injected embryos, reminiscent of the combined requirement of Dorsal (as a Rel protein) and Twist for sna activation (Ip et al., 1992). The co-expression of twi and sna in circumferential stripes after Xrel2 mRNA injection should lead to a ring of mesoderm invagination. However, the expression of twi and sna was only transient, and the expression of both genes had faded away at the onset of gastrulation. This is the likely explanation for the absence of a circumferential ventral furrow and the lack of mesoderm, despite the initial activation of twi and sna transcription.

The last gene we analyzed was rho, which codes for a transmembrane protein. rho has a complex expression pattern. In wild-type embryos, it is expressed in three stripes of which the ventrolateral ones are activated by intermediate levels of Dorsal. The dorsal stripe and apparently the “head” expression domains depend on sog (Bier et al., 1990). While the ventrolateral expression domains can be seen already during syncytial blastoderm stages, the dorsal stripe appears shortly before gastrulation. Its expression levels become segmentally modulated at the beginning of gastrulation (Bier et al., 1990). In dl−/− embryos shortly before gastrulation, rho is evenly expressed around the entire circumference of the DV axis as a result of the expansion of its dorsal-most stripe (Fig. 4). After injection of dl−/− embryos with the Xrel mRNAs, no early rho expression was detected, demonstrating that its early ventrolateral expression domain was absent. Later embryos with incipient morphogenetic movements showed circumferential stripes of rho expression interrupted by a gap lacking rho close to the side of injection. These rho stripes correspond to the dorsal rho expression domain, as inferred form their late appearance and segmental modulation. Further, the occurrence of gaps in the center of the embryo is likely to reflect the suppression of dpp in these regions, because the dorsal rho domain requires Dpp signaling.

In summary, the experiments show that Xrel1 and Xrel2 are able to activate target genes that respond to high (type I) and very low levels of Dorsal (type III), however, they fail to activate the type II target gene rho, which responds to intermediate levels. These expression patterns are consistent with the cuticle (Fig. 2) and gastrulation (Fig. 3) phenotypes in that they also lack polarity along the DV axis.

DISCUSSION

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

We have shown in this study that Xenopus Rel proteins can partially substitute for functions of the Dorsal protein in early Drosophila embryogenesis. Xrel1 and Xrel2 induce the formation of ventrolateral (neurectodermal) cell fates. However, they fail to induce ventralmost (mesodermal) cell fates. This phenotype can be explained by the repression of dpp and the activation of the Dpp inhibitor sog by Xrel1 and Xrel2. In Drosophila, the neuroectoderm forms as a default state in the absence of Dpp signaling. Indeed, the phenotypes observed after injection of Xrel1 and Xrel2 into dl−/− mutant embryos are identical to those produced by dl dpp double mutants (Wharton et al., 1993). Thus, we conclude that the main regulatory consequence of Xrel1 and Xrel2 injections is the repression of Dpp activity, which results from the additive effects of dpp repression and sog activation. The activation of two major mesodermal genes (twi and sna) by Xrel2 has apparently no phenotypic consequences, because the gastrulation movements lack any signs of mesoderm invagination. This finding is likely to be due to the fact that twi, the major mesodermal determinant, is only transiently expressed.

Despite their ability to induce the formation of neurectodermal cell fates, Xrel1 and Xrel2 are not able to induce DV polarity in dl−/− mutant embryos. Cuticles and gastrulation movements demonstrate that the same cell fates are present around the entire embryonic circumference, and the expression pattern of the different dorsal target genes are symmetric with regard to the DV axis. Thus, Xrel1 and Xrel2 are unable to respond to the asymmetric input of Spätzle/Toll signaling from the ventral side.

Earlier experiments have investigated the question, whether Dorsal protein can be substituted by other insect Rel/NF-κB family members. Thus, it was shown that Dif, which is normally involved in the innate immune response to pathogens, can partially rescue the Dorsal mutant phenotype. The same applies to the Dorsal protein from Tribolium, a represent of the coleopterans, which diverged from the dipterans approximately 250 million years ago. In rescue experiments with Dif and Tribolium Dorsal, the embryos showed correct DV polarity, suggesting that these Rel/NF-κB proteins were able to interact with Drosophila Cactus (see Chen et al., 2000). This interaction would keep them sequestered in the cytoplasm of the early embryo and render their nuclear translocation sensitive to the polarity signal, which emerges at the ventral side—only if the Toll receptor is activated does their nuclear import occur. In the case of Tribolium Dorsal, it was shown that the protein forms a nuclear gradient in Drosophila embryos, which is indistinguishable from that of the endogenous Dorsal gradient (Chen et al., 2000). Nevertheless, Tribolium Dorsal was both a weaker activator and also a weaker repressor of the Drosophila target genes than the endogenous Dorsal protein. Only injection experiments with high amounts of mRNA led to mesoderm formation, demonstrating that, in principle, overexpressed Tribolium Dorsal can induce stable activation of twi and sna. In contrast, Dif activated twi and sna transcription only transiently and induced rho only weakly in a narrow ventral domain. Notably, no ventral furrow formation was observed (Stein et al., 1998). With regard to the induction of target genes, this situation resembles that after Xrel2 injection, where mesoderm formation does not occur despite the activation of sna and twi. The difference between Dif and Dorsal was explained by the failure of Dif to interact with the basic helix–loop–helix (bHLH) protein Twist (Stein et al., 1998). Indeed, the interaction of Dorsal with bHLH proteins has been shown to be required for strong and permanent activation of target genes in the mesoderm (Ip et al., 1992).

Based on the apparent sequence divergence, it is likely that both Xrel1 and Xrel2 proteins are unable to interact with Drosophila Twist and, therefore, cannot induce permanent expression of twi. This finding would contribute to their failure to induce the ventrolateral domain of rho expression. However, why are the two Xrels able to control the expression of type III genes, which respond to very low Dorsal levels, while they are unable to induce the type II target gene rho, which responds to intermediate Dorsal levels? Recent systematic studies of the cis-regulatory regions of Dorsal target genes (Stathopoulos and Levine, 2004; Papatsenko and Levine, 2005) show that type III enhancers are mainly characterized through clusters of high affinity Dorsal binding sites, whereas type II enhancers share a complex structure in which Dorsal binding sites are in close proximity to binding motifs of three different transcription factors: Twist, Suppressor of Hairless, and an unknown factor (potentially Dorsal-interacting protein 3). Thus, type II enhancers show a stronger dependence on the combination of Dorsal with other transcription factors, while type III enhancers mainly depend on the quality of individual Dorsal binding sites. Accordingly, type III enhancers should be less dependent on specific protein–protein interactions that could have changed during evolution. This might explain, why sog and sog but not rho are controlled by Xrel1 and Xrel2.

Our observations provide a basis for future experiments, aimed at improving the compensatory capacity of Xenopus Rel proteins in dl−/− flies, to reveal evolutionary adaptations of the REL/I-κB system in the two species. The failure of Xrel1 and Xrel2 to induce DV polarity may be caused by their inability to bind to Drosophila Cactus. Attempts to compare the intracellular localization of myc-tagged XRel/Dorsal protein variants in Drosophila embryos have failed so far, apparently due to interference of the tag with Rel protein properties (data not shown). Alternatively, they may not serve as targets of phosphorylation by the Toll signaling cascade. Drier and colleagues (Drier et al., 2000) have shown that not only the degradation of Cactus but also the phosphorylation of Dorsal is required for Dorsal nuclear import. It is already known that Rel proteins normally bear specificity for particular IκB inhibitors (see Huguet et al., 1997). By coinjection of mRNAs for Xenopus Rels and IκBs, one could test whether heterologous IκBs can be target of the Toll signaling cascade in Drosophila (Fernandez et al., 2001). In addition, cooperative protein interactions between exogenous XRel proteins and endogenous Drosophila bHLH proteins such as Twi may be insufficient, significantly restricting XREL-dependent target gene activation in Drosophila. This hypothesis could be investigated by injecting XRel/DmDorsal protein chimeras.

Given the evidence for an involvement of a Toll-like signaling activity in Xenopus embryogenesis (Armstrong et al., 1998; Prothmann et al., 2000), we have set out to ask if Xenopus Rel proteins are capable to substitute for functions of Drosophila Dorsal. Specifically, the Xenopus Rel1 and Rel2 proteins showed their ability to activate or repress genes such as sog and dpp, both genes of great evolutionary interest (see De Robertis and Sasai, 1996; Ferguson, 1996). Recently, we described the identification and activity of a component of the vertebrate Toll/IL-1 signaling pathway the MyD88 in Xenopus laevis (Prothmann et al., 2000). The MyD88 protein is an important adapter mediating the ligand-triggered signal from Toll/IL-1 receptors to the IκB degradation machinery, which finally leads to the release of Rel/NF-κB protein complexes into the nucleus. Loss-of-function experiments with dominant-negative MyD88 protein mutants revealed that a still unknown Toll/IL-1 signaling pathway contributes to the regulation of specific DV regulatory genes (Prothmann et al., 2000). Indeed, signaling through this MyD88-sensitive pathway is read out into NF-κB–dependent gene activation (Armstrong, Fagotto, Prothmann and Rupp, to be published elsewhere). Our findings, described here, provide further support for Rel proteins as components of an ancient d/v-patterning system. The analysis of the Xrel phenotypes in dl−/− flies suggests that fine-tuned protein interactions, downstream of conserved signaling pathways, contribute decisively to the species-specific interpretation of the signals.

EXPERIMENTAL PROCEDURES

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

Fly Strains and Maintenance

Flies were grown and eggs were collected under standard conditions (Wieschaus and Nüsslein-Volhard, 1986). The wild-type strain was Oregon R. dl−/− mutants were produced by crossing Df(2L)TW119 and In(2L)dlH (Steward and Nüsslein-Volhard, 1986).

RNA Microinjections

The coding regions of Xenopus Rel1 (GenBank accession no. M60785), Rel2 (accession no. Z49252), RelB (accession no. D63332), and Drosophila Dorsal (accession no. P15330) were amplified by polymerase chain reaction and subcloned into pCS2+ (Rupp et al., 1994) to generate SP6 transcripts for mRNA microinjections. For in vitro synthesis of capped RNA transcripts, plasmids were linearized with Asp718 (Roche, Mannheim). The transcripts were diluted as indicated and injected into the central region of approximately 1-hr old embryos at 18°C (preblastoderm stage).

Documentation of Cuticular Rescue and Gastrulation Phenotypes

Cuticle preparation of embryos was done according to Wieschaus and Nüsslein-Volhard (Wieschaus and Nüsslein-Volhard, 1986). After clearing, cuticles were photographed using darkfield optics. Injected embryos were scored for the presence of antennal sense organ (ASO), Filzkörper, and ventral denticles (Anderson et al., 1985; Roth et al., 1991). Gastrulation of living embryos was documented under brightfield illumination after submersion in Voltalef 10s oil.

Immunohistochemistry

RNA in situ hybridization and immunohistochemical stainings of embryos were carried out according to Roth et al. (1989) and Tautz and Pfeifle (1989). For immunological detection of Twist protein, a polyclonal anti-Twi antiserum (1:1,000 dil.; Roth et al., 1989) was used, together with biotinylated horseradish peroxidase–avidin complexes bound to biotinylated secondary antibody (Vector Laboratories, Avidin/Biotin ABC system).

Acknowledgements

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

We thank Dr. Britta Linder for comments on the manuscript. R.R. was supported by a grant from the DFG.

REFERENCES

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