A co-dependent requirement of xBcl9 and Pygopus for embryonic body axis development in Xenopus



The Wnt/β-catenin transcriptional activation complex requires the adapter protein Pygopus (Pygo), which links the basal transcription machinery to β-catenin, by its association with legless (Lgs)/ B-cell lymphoma-9 (Bcl9). Pygo was shown to be required for development in vertebrates, but the role of Lgs/Bcl9 is unknown. We identified an amphibian orthologue of Lgs/Bcl9, XBcl9, which interacted biochemically with Xβ-catenin and XPygo2. The body axis promoting ability of Xβ-catenin was diminished when residues required for its interaction with XBcl9 were mutated. In blastula embryos, XBcl9 was transiently preferentially expressed in nuclei of dorsoanterior cells and ectopically expressed XBcl9 required XPygo2 to localize to nuclei. Furthermore, while neither XBcl9 nor XPygo2 alone affected development when ectopically expressed, both were required to induce supernumerary axis and dorsal gene activation. Like XPygo2, depletion of maternal XBcl9 alone caused dorsal defects. These results indicated an essential role of the Pygo-Bcl9 duet in vertebrate body axis formation. Developmental Dynamics 239:271–283, 2010. © 2009 Wiley-Liss, Inc.


The unfertilized radially symmetric Xenopus egg consists of a pigmented animal hemisphere and a yolk-laden vegetal hemisphere. The egg cortex is a thin outer layer of egg cytoplasm associated with the plasma membrane. Breaking of radial symmetry is precipitated by the fertilizing sperm, which causes the cortex to rotate with respect to the inner egg cytoplasmic mantle (Vincent et al., 1986; Vincent and Gerhart, 1987). The rotational shift displaces cortical components originally concentrated at the vegetal pole to a location opposite the sperm entry site (Holowacz and Elinson, 1995; Marikawa et al., 1997; Rowning et al., 1997; Marikawa and Elinson, 1999; Weaver and Kimelman, 2004; Tao et al., 2005). These components activate canonical Wnt signaling in cells so that they differentiate into dorsal, anterior structures to produce an elongated body axis (Marikawa et al., 1997; Rowning et al., 1997; Marikawa and Elinson, 1999; Tao et al., 2005). In the absence of the rotation, or when the Wnt pathway is inhibited, the embryo fails to develop a body plan (Vincent and Gerhart, 1987; Tao et al., 2005).

Canonical Wnt signaling activates an elaborate network of transducers that stabilize soluble/cytoplasmic β-catenin, which then translocates to the nucleus. In the nucleus, β-catenin is assembled into a transcriptional activation complex that promotes expression of genes required for axis development (Cadigan, 2008; Huang and He, 2008). The essential components of the transcriptional activation complex includes the adapter protein Pygopus (Pygo; Belenkaya et al., 2002; Kramps et al., 2002; Thompson et al., 2002; Hoffmans and Basler, 2004; Hoffmans et al., 2005). Pygo links the basal transcription machinery to β-catenin (Carrera et al., 2008), by its association with legless (Lgs).

Structurally, Pygo is composed of an N-terminal NHD motif with transactivating potential connected to a C-terminal plant homeodomain (PHD), also known as a leukemia associated protein (LAP) domain, by a proline-rich linker region (Kramps et al., 2002; Thompson et al., 2002). In flies, Lgs is a 1464 amino acid protein with no commonly conserved structural motifs. However, the Lgs N-terminus does contain three stretches of ∼30 amino acids, referred to as homology domains (HD), and are numbered sequentially from HD1 to HD3 (Kramps et al., 2002). The HD1 interacts with the PHD of Pygo and the HD2 with armadillo repeat 1 of fly β-catenin (Arm; Kramps et al., 2002; Thompson et al., 2002; Hoffmans and Basler, 2004; Townsley et al., 2004; Sampietro et al., 2006). Both lgs and Pygo were considered to be essential for the fly wingless (wg/Wnt) pathway because mutant alleles of both genes phenocopied wg and arm mutants (Kramps et al., 2002; Thompson, 2004). Furthermore, development of lgs mutants could not be rescued by constitutively active Arm protein, indicating that it was critically dependent on Lgs for activity (Kramps et al., 2002).

A variety of studies have demonstrated that Pygo is necessary for vertebrate development. In Xenopus, antisense depletion of maternal Pygo resulted in inhibition of body axis development similar to that caused by β-catenin depletion (Heasman et al., 2000; Belenkaya et al., 2002) while zygotic depletion of Pygo using antisense morpholinos caused defects in head and eye development (Lake and Kao, 2003). Surprisingly, restoration of expression of a variety of head-specific genes in these Pygo-deficient embryos was achieved by expression of Pygo proteins that lacked the ability to participate in Wnt-dependent transcription (Lake and Kao, 2003). In mouse, targeted deletion of mpygo2 resulted in a variety of developmental defects, most of which were related to canonical Wnt signaling, but also for non-Wnt function, such as lens development (Song et al., 2007; Li et al., 2007; Jonckheere et al., 2008). These studies suggested that Pygo has functions required for development in vertebrates that are broader than those dependent on, or dedicated to canonical Wnt signaling.

Because of its apparent multifunctional roles, an important question that we ask is: What regulates the participation in vertebrate development of Pygo in canonical Wnt cellular functions? In this study, we have addressed this question by examining legless/Bcl9, the partnering component to Pygo in Xenopus development. We report the identification and biochemical characterization of the Xenopus orthologue of Bcl9 (XBcl9), its embryonic expression patterns and its dorsalizing activity in early embryos. In particular, our data indicate that XBcl9 protein accumulates initially in the dorsal side of the marginal zone of early blastula stage embryos. We provide evidence of a co-dependency of XBcl9 and XPygo2 on nuclear localization, which is required to promote dorsal gene expression and axis formation. Furthermore, we demonstrate the requirement of maternal XBcl9 for body axis formation. These data provide supporting evidence that XPygo-Bcl9 is an essential component for axis formation during Xenopus development.


Identification of XBcl9, the Xenopus Orthologue of Lgs/Bcl-9

A candidate Xenopus orthologue of lgs/Bcl9 with accession number BC070813 (Klein et al., 2002), was identified by ‘Blast’ing (www.pubmed.com) known fly, mouse, and human Lgs/Bcl9 cDNA and protein sequences against all Genbank Xenopus laevis sequences. BC070813 has two tandemly arranged opening reading frames (ORFs): a 2,586-bp N-terminal ORF and a 1,893-bp C-terminal ORF. Conceptual translation of the N-terminal ORF predicted a protein with significant homology (66%) to the N-terminal half of fly, mouse, and human Lgs/Bcl9. The second ORF predicted a protein containing sequences that had significant homology (78%) with vertebrate Bcl9 sequences only.

These two out-of-frame ORFs are separated by a single nucleotide, which may likely have been introduced during cloning of BC070813, but this does not strictly rule out the possibility of BC070813 representing an expressed pseudoallele of a Xenopus lgs/Bcl9 orthologue. We used, therefore, the polymerase chain reaction (PCR) to amplify a single cDNA clone, named XBC01, with an uninterrupted ORF spanning both BC070813 ORFs from cDNA of Nieuwkoop and Faber (1994; N&F) stage 1 embryos that when translated in vitro, produced a single protein of approximately 160–170 kDa (Fig. 2A).

The XBC01 cDNA encoded a predicted 1473 amino acid protein similar to Lgs/Bcl9 (theoretical mass ∼156 kDa), with 65 amino acid residues at the N-terminus not found in other Lgs/Bcl9 sequences. The predicted protein has the six domains conserved in the Bcl9 family (HD1 to HD6; Katoh and Katoh, 2003). HD1–3 are present in the N-terminal half (Fig. 1A) and HD4–6 in the C-terminal half (data not shown). Phylogenetic analysis (ClustalW2) of the protein sequences of all known and predicted Bcl9 and Bcl9-2 proteins suggests that XBC01 encodes either XBcl9 or XBcl9-2 (Fig. 1B). Based on protein sequence, however, XBC01 predicts a protein more closely related to Bcl9 than Bcl9-2. It is ∼70–71% identical to murine and human Bcl9 as compared to Bcl9-2, which has ∼35% identity with the predicted XBC01 protein but is only ∼23% identical to Drosophila Lgs (Fig. 1B), where only HD1–3 is conserved. Because it is most similar to the mammalian Bcl9 proteins, we have named the gene encoding this predicted protein, XBcl9 (NCBI accession no. GQ891057).

Figure 1.

Identification of the Xenopus Bcl9 orthologue. A: Alignment of the evolutionarily conserved homology domains (HD1–3) of XBcl9 with other selected Lgs/Bcl9 family members. Identical residues are shaded. B: Phylogram based on the amino acid sequences of all known and predicted Bcl9 and Bcl9-2 proteins available from NCBI with accession numbers for each protein included in parentheses. C: The percentage identity of full-length XBcl9 amino acid sequences with selected orthologous Bcl9 proteins.

Biochemical Characterization

XBcl9 encodes a protein with biochemical characteristics indicative of its role as a Lgs/Bcl9 orthologue. In Drosophila, Lgs/Bcl9 interacts directly with Pygo2 by means of the HD1 domain and β-catenin by means of the HD2 domain to mediate transcriptional activation of Wnt reporter constructs and target genes (Parker et al., 2002, 2008; Thompson et al., 2002). Consistent with these previous observations, in vitro GST-Pulldown assays demonstrated that XBcl9 interacts with Xβ-catenin and both isoforms of XPygo2. The in vitro GST-XBcl9/Xβ-catenin interaction was specific but weaker than the interaction between GST-XBcl9 and XPygo2α/β (Fig. 2B).

Figure 2.

Identification and biochemical characterization of XBcl9 protein by GST pulldown analysis. A: In vitro translated proteins from BC070813, XBcl9 (HD1–3; i.e., the N-terminal ORF from BC070813) and full-length embryonically derived XBcl9 cDNAs. B: XBcl9 (HD1–3) interacts with both isoforms of XPygopus2 (XPygo2) and Xβ-catenin. C: Residues D162 and D164 of Xβ-catenin mediate its interaction with XBcl9, in vitro. XAxin and XTcf3 were used as known binding protein controls to show specificity of the point mutations. D: GST-pulldown demonstrating residues H417 and R418 of XBcl9 specifically disrupt its interaction with Xβ-catenin without affecting its ability to bind XPygo2, in vitro. NB. Bottom panel of Coomassie-stained GST proteins shows equal amounts of proteins were used for each pulldown.

Mutation of the aspartate residues to alanines at positions 162 or 164 within armadillo repeat 1 of β-catenin, was sufficient to reduce the transcriptional activity of β-catenin (Hoffmans and Basler, 2004). Subsequent analysis of the crystal structure of the HD2–β-catenin-Tcf4 complex confirmed that Bcl9 interacts with the so-called “acidic knob” centered at D162 and D164 of β-catenin while residues H417 and R418 of Bcl9 were important for its interaction with β-catenin (Sampietro et al., 2006). We generated two mutant forms of Xβ-catenin, one in which D164 was altered to an alanine (Xβ-cateninD164A) and one in which both D162 and D164 were altered to alanines (Xβ-cateninD162A,D164A; Fig. 2C). Similarly, H417 and R418 of the HD2 of XBcl9 were both mutated (Fig. 2D) to alanines to generate XBcl9H417A,R418A. These mutant proteins were then tested for their ability to interact in Xenopus embryos.

Mutation of the above-mentioned residues in Xβ-catenin significantly reduced but did not completely eliminate its ability to interact with XBcl9, in vitro (Fig. 2C), as compared to wild-type Xβ-catenin. Both, however, were able to interact with XAxin or XTcf3 (Fig. 2C). Likewise, XBcl9H417A,R418A interacted strongly with GST-XPygo 2β but not with GST-Xβ-catenin (Fig. 2D). These data indicated that the predicted protein interactions of XBcl9, and the critical residues associated with those interactions, are conserved between Xenopus and other species.

Xβ-cateninD162A,D164A Inefficiently Rescues Xβ-Catenin Depleted embryos

The interaction between Xβ-catenin and XBcl9 is also required to promote axis formation. The dorsoanterior index (DAI) scale was used to quantitatively measure dorsal axis formation: DAI = 5 represents normal phenotype; DAI > 5 is dorsalized; DAI < 5 is ventralized (Kao and Elinson, 1988). As previously shown (Heasman et al., 2000), antisense morpholinos complementary to Xβ-catenin caused complete elimination of the body axis when injected into two-celled embryos resulting in an average DAI score of 0.83 (n = 71; Table 1). As expected, wild-type Xβ-catenin mRNA rescued axis development in a dose-dependent manner raising the average DAI to 1.79 at 200 pg and 2.82 at 400 pg (Fig. 3D,F,H; Table 1). Xβ-cateninD162A,D164A mRNA, however, was less effective at rescuing axis development (Fig. 3C,E; Table 1) raising the average DAI to 1.27 at 200 pg and 1.12 at 400 pg.

Figure 3.

Mutations required for Xβ-catenin to interact with XBcl9 reduce its ability to rescue Xβ-catenin–depleted embryos. A: Uninjected control embryos. B: A total of 15 ng of Xβ-catenin morpholino (MO) was injected into the dorsal-vegetal region of two-cell stage embryos to inhibit dorsal axis formation. C–H: Increasing amounts of either wild-type (WT) Xβ-catenin (D,F,H) mRNA or Xβ-cateninD162A,D164A mutant mRNA (C,E,G) were co-injected with Xβ-catenin MO to rescue morphant phenotypes.

Table 1. Xβ-CateninD162A,D164A Inefficiently Rescues Body Plan Formation as Compared to Wild-Type Xβ-Catenin
SampleDAInAve DAIMedian DAI
Uninjected  21841 (DAI 8)884.985
xβ-cat MO436139  710.830
xβ-cat MO +200pg D162A, D164A113341 221.270.5
xβ-cat MO +400pg D162A, D164A3351683 651.120
xβ-cat MO +800pg D162A, D164A5210221 3432.954
xβ-cat MO +200pg WT1418104 1381.792
xβ-cat MO +400pg WT5515101010552.823
xβ-cat MO +800pg WT21101117 8493.314

At 800 pg, both the wild-type and mutant Xβ-catenin mRNAs rescued development to an equivalent amount suggesting that XBcl9 can still weakly interact with Xβ-cateninD62A,D164A as indicated in Figure 2C. Alternatively, mutant Xβ-catenin may act in an alternate manner by displacing endogenous wild-type Xβ-catenin from the plasma membrane similar to “localization” mutants described previously in Xenopus (Miller and Moon, 1997) and as suggested for a Lgs binding defective armadillo (armS10-D164A) mutant of Drosophila (Hoffmans and Basler, 2004).

XBcl9 mRNA and Protein Expression

The embryonic expression of XBcl9 is consistent with a potential role for this gene in axial patterning. Northern blot analysis of total RNA extracted from N&F staged embryos detected a single, ∼5.4-kb band at all stages corresponding to the XBcl9 transcript. The highest levels of XBcl9 mRNA were present during cleavage and blastula stages (N&F stages 1–8). By gastrulation (Stage 10), XBcl9 mRNA decreased to very low but detectable levels that persisted throughout embryonic development (Fig. 4A). These XBcl9 levels mirrored the temporal expression pattern of XPygo2β (Lake and Kao, 2003). Western blotting using affinity purified polyclonal rabbit antiserum to detect XBcl9 protein extracted from staged embryos indicated that, in early development, XBcl9 was expressed in a reciprocal pattern to that of its mRNA. Protein expression was at low levels throughout cleavage and blastula stages but plateaued at St.10 until early tail bud stages, after which it became undetectable (Fig. 4B).

Figure 4.

Expression of XBcl9 mRNA and protein is developmentally regulated. A: Northern blot analysis for XBcl9 mRNA. The 28s and 18s ribosomal RNAs are indicated. Total RNA was extracted from embryos staged according to Nieuwkoop and Faber (N&F, 1994) and normalized to Histone 4 (H4) levels. B: Western blot analysis of XBcl9 protein from total protein extracts of staged embryos. Total protein was standardized using β-tubulin as the loading control. C: Western blot analysis of XBcl9 protein distribution along the animal–vegetal axis in stage 9 embryos. Total protein levels were normalized to β-tubulin as a loading control.

Subcellular XBcl9 Expression and XBcl9 Levels Correlate With Dorsal Development

The marked increase in XBcl9 expression in embryos between St.8 and St.10 was mainly attributed to accumulation of protein in their marginal zones observed at St.9 (Fig. 4C). Because Bcl9 is a predominantly nuclear protein required for Wnt target gene transcription (Thompson et al., 2002; Thompson, 2004; Townsley et al., 2004; Krieghoff et al., 2006), and both known XBcl9 binding proteins, Xβ-catenin and XPygo2, are required for dorsal axis formation (Heasman et al., 2000; Belenkaya et al., 2002), we next determined whether or not the subcellular expression of XBcl9 correlated with embryonic polarity.

Embryos from stages 7, 8, 8.5, and 9.5 were fixed and bisected into dorsal and ventral halves. Embryos were tilted orthogonally to the animal–vegetal axis within 30–40 min after fertilization until they completed the first cell division, to ensure the position of the plane of bilateral symmetry (Black and Gerhart, 1985). Immediately after tilting, the surface of the equatorial zone was marked with Nile Blue crystals so that the presumptive dorsal region could be identified at later stages. Embryo halves were then processed for immunofluorescence using XBcl9 antiserum and visualized by scanning confocal microscopy (Fig. 5A–C). At stage 7, XBcl9 was detected diffusely in both dorsal and ventral sections with no localization to nuclei (Fig. 5A). By stage 8.5, XBcl9 was present at the periphery of both dorsal and ventral cells but in the nuclei of dorsal marginal and vegetal cells only (Fig. 5B(ii, iv, vi, and viii)). Between stages 8 and 8.5, 24–30% of dorsal nuclei expressed XBcl9 compared with 2–5% of ventral nuclei (Fig. 5D). This dorsal–ventral asymmetry in nuclear expression of XBcl9 remained until late stage 9 when XBcl9 was expressed at cell boundaries and in nuclei throughout the entire embryo (Fig. 5C(ii–iv and vi–viii)), coincident with the overall elevated levels of XBcl9 protein (Fig. 4C). This dorsal–ventral nuclear staining pattern suggests an association of XBcl9 with dorsal cell fate during early development.

Figure 5.

XBcl9 subcellular localization and levels correlate with dorsal development. A–C: Stage 7, 8.5, and 9.5 embryos bisected into dorsal and ventral halves were immunofluorescently stained for endogenous XBcl9 protein (green) using polyclonal antiserum. A(ii, iv), B(ii, iv), and C(ii, vi) are preimmune serum negative controls for XBcl9 staining. C (iii, iv, vii, and viii) are magnified images of nuclei expressing XBcl9. DAPI (4′-6-diamidino-2-phenylindole) staining specifically identifies nuclei. D: Quantitative representation of the fraction of nuclei expressing XBcl9 in 8–12 dorsal (blue) and ventral (red) embryo halves. Data from prestage 8 (stage 7) embryos were not included, because we found no evidence of nuclear staining at this stage. E: XBcl9, Xβ-catenin, and XGSK3β levels in D2O dorsalized or ultraviolet ventralized embryos. Total protein was standardized to β-tubulin levels. DAI, dorsoanterior index; n = sample size. Scale bar = 150 μm in A,B,C (i–ii and v–vi), 200 μm in C (iii–iv and vii–viii).

The activities of Wnt pathway components, Xβ-catenin, Dishevelled, and GBP positively correlate with dorsal tissue identity in normal embryos and those that have been hyperdorsalized with deuterium oxide (D2O; Scharf et al., 1989; Rowning et al., 1997; Miller et al., 1999; Weaver et al., 2003). Correspondingly, ultraviolet (UV) -irradiated, ventralized embryos lack dorsal–ventral polarity in the expression of these dorsal markers and levels of GSK3β, a Wnt pathway inhibitor, negatively correlates with dorsal identity (Dominguez and Green, 2000). Based on the observation that XBcl9 is preferentially expressed in dorsal nuclei, we determined if XBcl9 levels correlated with dorsalization of the Xenopus embryo. Embryos were treated with D2O to hyperdorsalize them, and UV-irradiated to ventralize them. Cohorts of 10 embryos were taken from each treatment for western blot analysis and the remaining ones were left to develop until late tail bud stages for DAI scoring. In D2O-treated embryos, XBcl9 and Xβ-catenin levels increased with DAI but, surprisingly, no difference was detected in total GSK3β levels in embryos with average DAI ∼6.4–6.7 (Fig. 5E). The lack of detectable reduction in total GSK3β levels compared with previous findings may be explained by the higher degree of hyperdorsalization (DAI ≥7) previously reported (Dominguez and Green, 2000). In UV ventralized embryos, there was no consistently detectable change in either XBcl9 or Xβ-catenin levels, nor in GSK3β levels, as expected (Dominguez and Green, 2000). These results confirm that XBcl9 protein levels positively correlate with dorsal but not ventral cell fate.

Ectopic Expression of XBcl9 With XPygo in Embryos Induces Secondary Axis Development

Consistent with studies in Drosophila embryos in which Lgs and Pygo are required together in a complex to activate transcription from Wnt target genes (Kramps et al., 2002; Parker et al., 2002; Thompson et al., 2002), only when XBcl9 and Xpygo2 were ectopically co-expressed, could they promote axis formation in Xenopus. Embryos injected ventrally (either vegetally or marginally) with either XBcl9 or XPygo2 synthetic mRNA developed normally (Fig. 6C,D). The lack of any marked phenotype by overexpression of XPygo2 is consistent with previous findings (Lake and Kao, 2003). However, co-injection of XBcl9 and XPygo2 mRNAs into the ventral-vegetal region of two-cell stage embryos frequently caused axis duplication (52/86, 60%; Fig. 6E,G).

Figure 6.

Dorsalization of Xenopus embryos by co-expression of XBcl9 and XPygo2β. A: Uninjected control embryos. B: Enhanced green fluorescent protein (EGFP) mRNA (1 ng) injected control. C: EGFP-XBcl9 mRNA (4 ng) injected embryos. D: XPygo2β mRNA (2 ng) injected embryos. E,F: Co-injection of EGFP-XBcl9 (4 ng) and XPygo2β (2 ng; blue arrows indicate duplicated axes; purple arrow indicates large endodermal mass). G: Frequency of embryos exhibiting 2° axis duplication or enlarged endoderm. Sample size is bracketed above bars indicating number of scored embryos. H: Quantitative polymerase chain reaction (qPCR) analysis of dorsal, ventral, and endodermal markers in stage 10 embryos that were uninjected (i), injected with EGFP mRNA (ii), EGFP-XBcl9 mRNA (iii), XPygo2β mRNA (iv), or EGFP-XBcl9 plus XPygo2β mRNA(v). Relative fold change (Y-axis) as compared to the uninjected controls was calculated using the 2−ΔΔCt method (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008).

The quantitative PCR (qPCR) analysis of marker genes confirmed the up-regulation of many Wnt/dorsal target genes such as siamois (∼3- to 3.5-fold increase), chordin (∼2.5- to 3-fold increase), and goosecoid (∼2.5- to 3-fold increase) as well as a modest reduction in the ventral marker Bmp4 (∼0.4-fold decrease; Fig. 6H). In addition, co-expression of XBcl9 and XPygo2β induced or expanded the amount of dorsal endoderm in 14 of 86 (16%) embryos (Fig. 6F,G) as suggested by the large increase in the Wnt/β-catenin targets Xnr5 (∼6.5-fold increase) and Xnr6 (∼9-fold increase; Fig. 6H). No change was detected in other endodermal markers such as Xnr1, Xnr2, and Endodermin (Data not shown), although a modest increase (∼0.5-fold) in Xsox17α and (∼0.75-fold) mix2 was detected (Fig. 6H). These results indicated that both XBcl9 and Xpygo2 cooperatively promote dorsal gene expression.

XPygo2 Is Required for XBcl9 Nuclear Expression and Enhances Xβ-Catenin Binding

The formation of a secondary axis by the co-injection of XBcl9 and XPygo2 suggested that these two proteins interact with each other, in vivo. Furthermore, in 293T cells, Lgs/Bcl9 was shown to be a nuclear protein only when co-expressed with Pygo despite the presence of a putative nuclear localization sequence (KRRK: XBcl9 amino acids 542–545; Townsley et al., 2004). Consistent with this previous observation, when expressed alone in stage 9 animal pole explants, enhanced green fluorescent protein (EGFP) –XBcl9 accumulated in randomly dispersed aggregates throughout the animal cap cells but was never detected in nuclei (Fig 7A, top row). However, when co-injected with MT-XPygo2, it was found in nuclei (Fig. 7A, bottom row) indicating that XBcl9 nuclear localization is dependent on the presence of Pygo.

Figure 7.

XPygo2β localizes XBcl9 to nuclei and enhances its interaction with Xβ-catenin. A: Enhanced green fluorescent protein (EGFP) -XBcl9 (2 ng) was injected alone or in combination with MT-XPygo2β (1 ng) into animal caps. Animal caps were explanted at stage 8 and cultured until stage 9. XBcl9 localization was detected by EGFP fluorescence; MT-XPygo2β was detected by immunostaining for the Myc tag and visualized by confocal microscopy. Nuclei were identified by counterstaining with DAPI (4′-6-diamidino-2-phenylindole). B: Co-immunoprecipitation of mRFP-Xβ-catenin and/or MT-XPygo2β with EGFP or EGFP-XBcl9. Equal amounts of input protein per co-immunoprecipitation was confirmed by Western blot of total protein extracts and further standardized by β-tubulin as a loading control. Scale bar = 150 μm.

Our above results indicating that XBcl9 activity is augmented by Xpygo2 in embryos was substantiated by in embryo protein interaction assays. Combinations of RNAs encoding EGFP-XBcl9, MT-XPygo2β, and mRFP-Xβ-catenin were injected into two-cell stage embryos which were allowed to develop until stage 9 when total protein was harvested from 40–50 whole embryos for co-immunoprecipitation experiments. As demonstrated in vitro (Fig. 2B), XBcl9 interacted very strongly with XPygo2β but weakly with Xβ-catenin in vivo (Fig. 7B). When co-expressed with XPygo2β, XBcl9 interacted much more strongly with Xβ-catenin (Fig. 7B), indicating that XPygo2β enhances or facilitates their interaction in embryos.

Maternal XBcl9 Is Required for Axis Formation

While the above experiments indicated that misappropriate expression of XPygo-Bcl9 could promote axis formation, they do not directly address a requirement for XBcl9 in development, as was previously demonstrated for maternal XPygo2 mRNA (Belenkaya et al., 2002). Therefore, we next determined if maternally expressed XBcl9 was required for axis formation. For this analysis, we depleted maternal XBcl9 transcripts using a phosphorothioated antisense oligonucleotide (oligo) from oocytes that were subsequently matured in vitro and processed for embryonic development using the host-transfer technique (Mir and Heasman, 2008). XBcl9 mRNA was reduced by approximately 70% and XBcl9 protein was reduced to nearly undetectable levels, when as much as 4 ng of the antisense oligo was injected into fully grown oocytes as compared to uninjected, control oocytes (Fig. 8A,B).

Figure 8.

XBcl9 is required for dorsal axis formation. A: Quantitative polymerase chain reaction (qPCR) analysis of XBcl9 mRNA levels in 1 ng, 2 ng, and 4 ng of antisense oligo injected, progesterone-matured stage VI oocytes relative to uninjected (UI) controls. B: Western blot analysis of XBcl9 protein levels corresponding to (A). Protein levels were standardized using b-tubulin as a loading control. C: Phenotype analysis of embryos with reduced XBcl9 mRNA and protein levels from host-transfer. Phenotypes were scored at stage 24. D: qPCR analysis of molecular markers at stage 9 and stage 10 in XBcl9 depleted (XBcl9) embryos. The XBcl9 embryos illustrated were injected with 4 ng of antisense oligos.

When fertilized using the host-transfer technique, XBcl9 depleted (XBcl9) oocytes developed severe axis abnormalities. Injection of 1 ng and 2 ng of the antisense oligo resulted in loss of dorsal development in 3/26 (12%) and 6/38 (16%) embryos, respectively, when scored at tail bud stages (Fig. 8C). At 4 ng, dorsal defects were present in 16/31 (52%) embryos (Fig. 8C). The control cohort all developed normally (0/43, 0%; Fig. 8C). The qPCR analysis of Wnt/dorsal molecular markers revealed significant reductions in the levels of siamois, Xnr3, Chordin, and Goosecoid at stage 9 and stage10, in XBcl9 depleted embryos (Fig. 8D). There were no observable changes in the levels of the ventral markers Bmp4 and XWnt8 or the endodermal marker Xsox17α (Fig. 8D).

To confirm specificity of the XBcl9 antisense oligo mediated dorsal deficiency, manually defolliculated oocytes were coinjected with 2 ng of XBcl9-specific antisense oligo along with XBcl0 mRNA and cultured for 48 hr at 18°C. While 4 ng of oligo caused a greater frequency of dorsal defects, we used 2 ng to limit the total amount of exogenous nucleic acids being introduced into the oocytes. Thus, before progesterone-induced in vitro maturation, XBcl9 oocytes were injected with either 500 pg of XBcl9 mRNA alone or co-injected with 250 pg of XPygo2 mRNA because injection of both mRNAs were necessary for ectopic axis formation (as shown in Fig. 6E,F).

Uninjected control embryos developed normally in 64/69 (93%) cases. Loss of dorsal development was observed in 20/87 (23%) embryos derived from oocytes injected with 2 ng of XBcl9 antisense oligos. Injection of XBcl9 rescuing mRNA alone into XBcl9 oocytes was sufficient to restore normal axis formation in 31/37 (84%) and cause maternally derived dorsal-anteriorization, seen as an expanded dorsal endoderm region in 5/37 (14%) embryos (Fig. 9). Co-injection of XBcl9 and XPygo2 rescuing mRNAs into XBcl9 oocytes, caused excessive dorsal-anteriorization in 43/43 (100%) embryos (Fig. 9). These results indicated that maternal XBcl9 is required for normal axis development.

Figure 9.

XBcl9 mRNA restores development to axial defective, XBcl9 embryos. A: Embryos depleted of maternal XBcl9 mRNA by injection of 2 ng of XBcl9 antisense oligonucleotides (αs Bcl9, n = 87) were co-injected with 500 pg of XBcl9 mRNA (αs Bcl9+Bcl9 RNA, n = 37) or additionally, with 250 pg of XPygo2 mRNA (+Pygo2 RNA, n = 43). Data shown are for embryos generated from three females (1–3). Oocytes from female 3 were injected with antisense oligos and co-injected with XBcl9 and Xpygo2 mRNA. B: Examples of maternal XBcl9-depleted embryos (upper right panel) co-injected with XBcl9 (lower left panel) and XBcl9 plus XPygo2 (lower right panel) mRNA. Examples of embryos with dorsal defects are indicated with green arrowheads and with expanded dorsal endoderm (dorso-anteriorization) with red arrowheads.


Specification of the dorsal body axis during Xenopus development depends on tightly regulated temporal and spatial activation of the Wnt signaling pathway. Because many of the intracellular components of the Wnt pathway are present maternally (Heasman, 2006a, b), normal development occurs only when these molecules are properly distributed in the developing embryo. Polarization of the embryo into dorsal and ventral regions is, therefore, initiated by the cortical rotation that results in the stabilization and nuclear accumulation of β-catenin in dorsally fated cells (Schneider et al., 1996; Larabell et al., 1997). These events occur very early in embryogenesis and appear to ‘prime’ cells to express dorsal target genes several hours later. Hence, the identification of Wnt pathway components that can mediate the β-catenin-TCF transcription complex is a necessary step toward understanding the dynamics of the complex and how it operates during embryonic development.

Because legless/Bcl9 has been described as a core component of Wnt signaling during Drosophila embryogenesis as well as important for transcription of synthetic Wnt reporter constructs in mammalian cell lines (Kramps et al., 2002; Parker et al., 2002; Thompson et al., 2002; Townsley et al., 2004; Stadeli and Basler, 2005), we hypothesized that Bcl9 is a necessary transcriptional regulating protein required for Wnt-mediated processes during Xenopus development. Our results indicated that neither Bcl9, nor Pygo are dispensable for TCF/β-catenin-mediated function. Both proteins strongly interact with each other and both are required to localize to nuclei, activate dorsal gene targets and promote axis formation. Thus, we propose that while Pygo and possibly Bcl9 may have non-Wnt associated functions, their interaction is essential, at least for canonical Wnt-associated axis formation events.

Temporal gene expression patterns are usually considered indicative of when a gene is required during embryogenesis. Our results confirming XBcl9 as a component that interacts with β-catenin and Pygo in Xenopus supports the role for an early requirement for the XPygo/Bcl9 complex in dorsal axis formation. Nuclear accumulation of maternal β-catenin by the 16- to 32-cell stage (Larabell et al., 1997) specifies dorsal competency in those cells and is required to specifically activate genes that define the dorsal axis. On the other hand, XBcl9 protein was not detected in dorsal nuclei until midblastula, even though there were relatively high mRNA levels present maternally. The expression of dorsal target genes such as Xnr3 (Smith et al., 1995), siamois, Xnr5, and Xnr6 (Lemaire et al., 1995; Takahashi et al., 2000) that follows the appearance of XBcl9 protein in dorsal nuclei, therefore, suggests that a determinative step of dorsal cell fate is dependent on the timing of XBcl9 nuclear entry thus completing the link between β-catenin, Pygo2, and the basal transcriptional machinery (Fig. 10).

Figure 10.

Schematic representation of canonical Wnt signaling dependency on XBcl9-Pygo for spatiotemporal regulated dorsal gene expression. A: Radially symmetric Xenopus embryo. Dorsal determinants (e.g., maternal Wnt proteins) are associated with vegetal cortex. B,C: Cortical rotation results in the asymmetric expression of Wnt11 (B, blue) that locally stabilizes β-catenin (C, red) by 8- to 16-cell stage. D: XBcl9 (green) localizes to dorsal nuclei at stage 8–8.5, concomitant with the timing of dorsal gene activation. (D, dorsal; V, ventral).

Recently, a novel transactivation region between HD4 and HD5 of Bcl9 was identified. This region augmented β-catenin stimulated transcription from DNA reporter constructs specifically in lymphoid (B and T) cell lines and was largely independent of Pygo2 (Sustmann et al., 2008). We did not find any evidence of such intrinsic activity of XBcl9, because overexpression of XBcl9 did not alter axis development. It is possible, however, that this Pygo-independent function simply is not required during early embryogenesis, but may be important for later development.

The confirmed requirement of the XBcl9-Pygo duet for canonical Wnt signaling in Xenopus may provide new insight into β-catenin mediated transcription in mammalian development. Studies in Drosophila have clearly identified a co-dependency for Lgs/Bcl9 and Pygo proteins for Wnt signaling (Kramps et al., 2002; Thompson et al., 2002). In addition to completing the link between Pygo and β-catenin-LEF/TCF bound to target promoters, analysis of the crystal structure of HD1 domain of Bcl9 and the PHD region of Pygo demonstrated that the HD1 enhanced the affinity of Pygo's PHD for methylated histones, implicating its role in chromatin remodeling (Fiedler et al., 2008). Pygo-Bcl9 might function, therefore, at multiple levels to regulate the transcription of Wnt target genes. Because Pygo2−/− null mice develop very mild congenital defects (microopthalmia) compared with mice with defects in other canonical Wnt pathway components (Song et al., 2007), these observations might suggest that Pygo has an evolutionarily diminished Wnt pathway function in mice. It is possible, however, that Bcl9 and Pygo are broadly co-dependent for canonical Wnt function in vertebrates. Bcl9−/−/Pygo2−/− developmental models might, therefore, need to be exploited to distinguish the canonical-Wnt and non-Wnt pathway requirements for Pygo in mouse development.


In Silico Identification of XBcl9

A Xenopus laevis I.M.A.G.E. clone, accession number BC070813, with homology to mammalian Bcl9 protein sequences was identified by in silico analysis (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

Plasmids and RNA Synthesis

XBcl9 coding region cDNA was PCR amplified from stage 1 Xenopus cDNA, and cloned into pCS2+ ClaI and XhoI restriction sites (forward primer: 5′-CG ATCGATATGCTGGAG CTGCAGGAGG-3′; reverse primer: 5′-CGCTCGAGGCA GCTTAAAACATCATGTTCCCG-3′; annealing temperature 60°C). All XBcl9 primers were based on the BC070813 sequence (restriction sites are underlined).

pCS2+/EGFP and pCS2+/mRFP expression plasmids were constructed by PCR amplification of coding regions (lacking stop codons) and inserted after restriction digestion into BamHI/ClaI sites in pCS2+. pEGFP-C2 and pcDNA/mRFP (Addgene) were used as templates for mRFP and EGFP DNA. EGFP and/or mRFP tagged constructs were then made by PCR amplification of XBcl9 or Xβ-catenin coding regions from embryonic cDNA and inserted after restriction digestion, downstream of the fluorescent tags (ClaI/XhoI restriction sites). The pCS2+/XTcf3 expression plasmid was constructed by PCR amplification of XTcf3 coding region from embryoniccDNA and after restriction digestion, inserted into pCS2+ (XhoI/XbaI restrictions sites). pCS2+/XPygo2α/β was previously described (Lake and Kao, 2003). pCS2+/XPygo2β was used as template to PCR clone the XPygo2β coding region into pCS3+MT (EcoRI/XhoI restriction sites) containing an N-terminal Myc tag. GST tagged XBcl9 (HD1–3), XPygo2β and Xβ-catenin coding regions were also PCR cloned, restriction digested, and inserted into pGEX4T-1: XBcl9(HD1–3) and XPygo2β using EcoRI/Xho1 sites and Xβ-catenin using BamHI/EcoRI sites. XBcl9H417A,R418A and Xβ-cateninD164A or Xβ-cateninD162A,D164A point mutations were made by Quick Change Site-Directed Mutagenesis (Stratagene). XBcl9H417A,R418A primers (5-CTTTCT CAAGAGCAACTG-GAGGCCGCAGAA CGCTCTTTGCA AACCCT-3 and the reverse compliment), Xβ-cateninD164A primers (5-CTGCTTAATGACGAGGC CCAGGT TGTAGTTAAC-3 and the reverse compliment) and Xβ-cateninD162A,D164A primers (5-CTGCTT AATGACGAGGCCCAGGTTGTAGTT AAC-3 and the reverse complement) were all polyacrylamide gel electrophoresis purified (IDT DNA). pCS2+MT-XAxin was a kind gift from Peter Klein and pYX-Asc/mBcl9-2 was purchased from Open Biosystems.

In vitro transcription of all RNAs used in this study was performed using Sp6 mMessage mMachine RNA synthesis kit, as per manufacturer's protocol (Ambion).

Embryo Collection and Manipulation

Wild-type embryos were obtained from female Xenopus laevis using standard techniques as described (Kao and Lockwood, 1996; Lake et al., 2001), maintained in 5% Normal Amphibian Medium (NAM) and staged according to Nieuwkoop and Faber (N&F; 1994). Embryos were tilted and immobilized in wells of an agarose-coated Petri dish in 4% Ficoll in 50% NAM until completion of first cell division to establish the position of the dorsal–ventral axis. Animal caps were excised from blastula stage embryos as described (Kennedy et al., 2007), cultured for 1 hr in 50% NAM and fixed in MEMFA for 30 min at room temperature and dehydrated and stored in 100% methanol at −20°C until needed. Dorsoanterior development was semiquantitatively measured using the DAI scale (Kao and Elinson, 1988). Both the median and average DAI are reported to better represent the central tendency because the average may not properly reflect the data if the distribution of the data values are skewed.


Maternal XBcl9 mRNA was depleted from stage VI oocytes as described (Mir and Heasman, 2008) using phosphorothioated antisense oligonucleotides (IDT DNA). XBcl9 antisense: 5′ G*G*G*TCACGATACAGCAGTGCTC* A*T*C 3′ (asterisk indicates phosphorothioate bonds). Antisense injected oocytes were cultured for 2 days before progesterone-induced maturation.

RNA Analysis: Northern Blotting and qPCR

Total RNA was extracted from staged embryos using the Nucleospin RNA II Kit (Clontech Laboratories, Inc.) and analyzed by northern blotting essentially as described (Kao and Hopwood, 1991). Probes were generated by PCR amplification of a fragment of Histone 4 (Lake and Kao, 2003) or XBcl9 coding region (485–1019 bp). XBcl9 primers: forward 5′ ATGCATTCCAGTAACCCCAAAGTG ′3; reverse 5′ AGTAGAGA AGACATACACCACTT′3). PCR products were gel purified (Ultrafree-DA, Millipore Corp.) used as templates for incorporation of 32P-α-ATP into random primed probes (Prime-a-Gene, Promega). Radiolabeled probes were hybridized at 58–60°C and visualized by autoradiography.

qPCR analysis on total RNA extracted from stage 10 embryos injected with XBcl9 and/or XPygo2β was performed as described (Kennedy et al., 2007). Isolation of total RNA from oocytes and embryos from the host transfer experiment was performed as described (Tao et al., 2005). First-strand cDNA synthesis was performed using M-MLV reverse transcriptase as per manufacturers protocol (Invitrogen). Synthetic cDNA was diluted 1:10 and 2 μl was used per qPCR reaction. Relative gene expression was analyzed by SYBR Green (Applied Biosystems) incorporation detected using the ABI Prism 7000 Detection System (Applied Biosystems). Results were analyzed using the comparative Ct method (i.e., 2−ΔΔCt; Livak and Schmittgen, 2001; Schmittgen and Livak, 2008). Relative levels were normalized to ornithine decarboxylase (ODC) levels. All qPCR primer sequences are listed in Sp2.

Co-immunoprecipitation and Western Blotting

Co-immunoprecipitation (Co-IP) was performed by injecting 40–50 embryos with in vitro synthesized RNA at the two-cell stage. Total protein was extracted by homogenization in 1XTm (10 mM Tris [pH 7.5], 1% Triton X-100, 10 mM ethylenediamine tetraacetic acid [EDTA], 0.02% NaN3) with protease inhibitors, and incubated on ice, 30 min. Homogenate was then passed through a 1-cc syringe 8–10 times and centrifuged at ∼14,000 rpm at 4°C, 10 min to pellet cellular debris. An equal amount of total protein lysate (as determined by Bradford Assay, Bio-Rad) was used for each IP. Final volumes were equalized with 1XTm. Anti-GFP (Abcam; dilution 1:250) was used for IP at 4°C rotating overnight. Protein A beads (GE Healthcare) were prewashed with IP buffer, and 30 μl of 50% bead slurry was added to each IP for 1 hr at 4°C, with rotation. Beads were washed 4 times with 1XTm and twice with 1× PBS (phosphate buffered saline; pH 7.5). Co-IP'd proteins were analyzed by Western blotting as described below. Antibodies used and dilutions were as follows: anti-Myc (dilution 1:2,000; 9E10, Developmental Studies Hybridoma Bank), anti-GFP (dilution 1:3,000; Abcam), anti-RFP (1:1,000), and anti–β-tubulin (dilution 1:3,000; Developmental Studies Hybridoma Bank).

Bcl9 antiserum was generated by injection of GST fused to a human Bcl9 fragment (amino acids 697–1294) into New Zealand White rabbits. Cross-reacting antiserum was affinity purified by passing whole antiserum 5 times over GST protein covalently linked to Glutathione Sepharose 4B beads (GE Healthcare) packed into protein purification columns (Econo-Column, Bio-Rad) using a peristaltic pump (2232 Microperplex S, LKB Bromma). Flow-through was then passed 5 times through a second column packed with beads covalently cross-linked to GST-XBcl9 using peristalsis. XBcl9 specific antibodies were eluted from the beads using Glycine-HCl (pH 2.0). The pH was neutralized using 2 M Tris, pH 8. Buffer exchange using 1XTBS (25 mM Tris, pH 7.5; 150 mM NaCl) was then performed by successive dilution-concentration cycles using Amicon Ultra-15 tubes (Millipore). Ten percent glycerol was added to the final concentrated XBcl9 antibody fraction. XBcl9 Membranes were probed using affinity purified XBcl9 antiserum (dilution 1:750), β-tubulin antibody (Developmental Studies Hybridoma Bank; dilution 1:3,000), β-catenin (monoclonal) antibody (Santa Cruz; 1:1,000 dilution) and GSK3β (Cell Signal; 1:500 dilution). Horseradish peroxidase-conjugated secondary antibodies (Amersham) were detected using the ECL-Plus and ECL (enhanced chemiluminescence; GE Healthcare) Western detection systems, respectively.

GST Pulldowns

GST-pulldowns were performed as per manufacturer's protocol. GST-fusion protein was expressed in BL 21 RP competent cells, extracted, and purified using Glutathione Sepharose 4B beads (GE Healthcare). Approximately 1 μg of total GST fusion protein was used in each reaction. Approximately 2 μl of XPygo2α/β, 5 μl of Xβ-catenin, 3 μl of XTcf3, 1 μl of XAxin, and 1 of μl XBcl9 (wild-type) or XBcl9H417A,R418A in vitro translated proteins, made in cell-free rabbit reticulolysate (Promega) as per manufacturer's protocol, were used per GST-pulldown. Gels were stained in Coomassie for 1 hr at and destained overnight (74%H2O, 6% glacial acetic acid, 20% methanol) at room temperature.

Immunofluorescent Staining and Imaging

Stored embryos/animal caps fixed in MEMFA/MeOH were re-hydrated in a graded methanol series and blocked in PBT (1× PBS, 2 mg/ml bovine serum albumin, 0.2% Triton X-100) for at least 1 hr at room temperature. Samples were then incubated in primary antibody diluted in PBT overnight at 4°C; whole Bcl9 antiserum (and preimmune serum controls) was used at 1:350 dilution and anti Myc (Developmental Studies Hybridoma Bank) at 1:250 dilution. Embryos were washed 5 × 20 min in PBT with agitation, and incubated 1.5–2 hr with fluorescently labeled secondary antibody diluted 1:250 in PBT at room temperature (secondary antibodies: anti-rabbit fluorescein isothiocyanate [FITC], BD Biosciences; anti-mouse-Cy5, Cedarlane), followed by washing 5 × 20 min in PBT. Nuclei were stained using DAPI (4′-6-diamidino-2-phenylindole, diluted to 200 ng/ml). Stained embryos/animal caps were mounted in 10% glycerol/1× PBS and visualized using Confocal Laser Scanning Microscopy (Olympus) and Fluoview software. Preimmune controls were visualized by using the same PMT settings as well as the same thickness and number of optical sections scanned. EGFP fluorescence was detected without antibody staining.


This work was performed in partial fulfillment for the degree of Doctor of Philosophy for M.W.K., supported by a CIHR grant to K.K. We thank Karen Stapleton for assistance with confocal microscopy and Peter Klein and Dave Turner for plasmids.