Introduction

Transforming growth factor-β (TGF-β) superfamily members are secreted signaling proteins determining the development, maintenance, and regeneration of tissues and organs. TGF-β superfamily members signal through receptor serine/threonine kinases and intracellular Smad proteins (Shi and Massague 2003). The Smad pathways are evolutionary conserved signaling modules that transmit signals to the nucleus. However, biochemical and developmental evidence supports the notion that alternative, non-Smad pathways also participate in TGF-β signaling (Moustakas and Heldin 2005). The Xenopus nodal-related protein, Xnr3, is similar in amino acid sequence to other types of Xnr but lacks the dorsal mesoderm-inducing activity of other Nodal-related factors (Ezal et al. 2000; Hansen et al. 1997). Xnr3 cannot activate the Smad pathway (Dorey and Hill 2006) but activates the MAP kinase signaling pathway and Xbra expression (Yokota et al. 2003). Xnr3 has unique structural features that diverge from the TGF-β superfamily consensus. Xnr3 is lacking the seventh cysteine that is highly conserved throughout the TGF-β superfamily and also has a serine instead of a glycine between the second and third cysteines within the consensus sequence CXGXC (Smith et al. 1995). The members of the TGF-β superfamily are usually homodimeric proteins, and the mature monomer typically has a cysteine-knot structure described as a left hand with the N-terminal end as thumb, the β-sheets as the knuckle which consists of two fingers, and central α-helix as the wrist (Sebald et al. 2004). Within the monomer, there are three disulfide bond pairs. The consensus sequences CXGXC between the second and the third cysteines, and the last two cysteines CXC constitute a ring of eight amino acids by forming disulfide bonds between the second and sixth, and third and seventh cysteines. Another disulfide bond is formed between the first and the fifth cysteines. Xnr3 lacks this essential ring, suggesting that Xnr3 does not form the characteristic cysteine-knot structure. In the dimer of TGF-β superfamily members, the wrist of one monomer fits into the hollow of the fingers of the other monomer. The dimer is stabilized by an interchain bond. The prediction of protein secondary structure indicates that Xnr3 lacks this α-helix described as the wrist. We report in this study that the mature region of Nodal-related 3 is required for ectopic Xbra expression and functions as a monomer. We previously reported that the proregion of Xnr3 inhibits BMP signaling (Haramoto et al. 2004, 2006), and several reports have revealed the activities of Xnr3 (Ezal et al. 2000; Glinka et al. 1996; Hansen et al. 1997; Smith et al. 1995; Yokota et al. 2003); however, the mechanism of Xnr3 action remains unresolved. Orthologs of Xnr3 have been identified only in a closely related species, Xenopus tropicalis (Haramoto et al. 2004). Further functional analyses of Nodal-related 3 should therefore present a unique insight into the divergent activities of TGF-β superfamily members.

Materials and methods

Embryos and microinjection

Xenopus laevis eggs were obtained as described (Tanegashima et al. 2004). Microinjection was performed in 100% Steinberg’s solution containing 5% Ficoll with synthesized mRNAs at a dose of 5 nl per blastomere. For Western blotting, each mRNA (1 ng) was injected into the marginal zone of all blastomeres of four-cell-stage X. laevis embryos. Embryos were staged according to Nieuwkoop and Faber (1956).

Constructs and mRNAs

To construct pCS2-pXtnr3, encoding the proregion of Xtnr3A, the corresponding region of pCS2-Xtnr3A was amplified by PCR with the following primers: forward (SP6) 5′-ATTTAGGTGACACTATAGAATAC-3′ and reverse 5′-CACACTCGAGTTAAAATCCATTGACCTTGTTGGC-3′. The PCR products were digested with EcoRI and XhoI and ligated into the EcoRI–XhoI site of the pCS2+ vector. pCS2-cmXtnr3 and pCS2-cmXtnr3-6myc were designed to have mutated cleavage sites changed from “RRLRR” to “GVDGG”. pCS2-Xtnr3-6myc, pCS2-cmXtnr3-6myc, and pCS2-cmXnr5-6myc all have six myc-epitope repeats at the C terminus. pCS2-Xtnr3-HA has an HA tag at the C terminus. In pCS2-Xnr5-HA, amino acid sequences containing an HA tag, “YPYDVPDYALE”, were inserted just after aspartic acid at position 263. pCS2-Xtnr3-6myc+C7 and pCS2-cmXtnr3-6myc+C7 have the seventh cysteine instead of phenylalanine 398. In pCS2-Xtnr3-6myc+C7sh and pCS2-cmXtnr3-6myc+C7sh, the C-terminal amino acids, “CGFKDI”, were changed to “CGCY”, as occurs in wild-type Xnr5. pCS2-Xnr5-6myc-mC7S and pCS2-Xnr5-6myc-mC7F have serine and phenylalanine, respectively, instead of the seventh cysteine. Chimera constructs 1–7 were generated by PCR using mutagenic primers. All constructs were sequence-verified. Capped mRNAs were synthesized using SP6 mMESSAGE mMACHINE (Ambion) with the above-mentioned and the following plasmids as templates: pCS2-Xtnr3-A (Haramoto et al. 2004), pCS2-mXtnr3 (Haramoto et al. 2006), pCS2-Xnr5-6myc (Onuma et al. 2005), and pCS2-nuclear localizing signal-β-galactosidase (β-gal) (Takahashi et al. 2000).

Immunoprecipitation and Western blotting

Coimmunoprecipitation assays were performed as described previously (Haramoto et al. 2004). Synthetic mRNAs were injected into the marginal zone of all blastomeres of four-cell-stage embryos. Injected embryos were harvested at stage 10. Samples were treated with or without 100 mM dithiothreitol (DTT) to give reducing or nonreducing conditions, respectively. The following antibodies were used for immunoprecipitation and Western blot analyses: IP, anti-c-myc (A-14-G) polyclonal antibody (Santa Cruz Biotechnology); anti-HA (Y-11) polyclonal antibody HA-probe (Santa Cruz Biotechnology); detection, anti-c-myc (9E10) HRP-conjugated monoclonal antibody (Santa Cruz Biotechnology); and anti-HA-Peroxidase High Affinity (Roche).

Whole-mount in situ hybridization

In situ hybridization analyses were performed as described (Sive et al. 2000). For lineage tracing, 250 pg β-galactosidase mRNA was coinjected and stained using Red-gal (Research Organics, USA). Probes were synthesized using pSP73-Xbra (Smith et al. 1991) as a template.

Results and discussion

Xtnr3 mature region is indispensable for ectopic Xbra activation

Xnr3 and Xtnr3 are unorthodox members of the TGF-β superfamily in that they lack the last of seven conserved cysteines and have peculiar activities that differ remarkably from those of other Nodal-related factors. The molecular mechanism underlying this difference in activities has not been defined. In Xenopus Nodal proteins, cleavage mutants function as dominant-negative inhibitors by forming inactive dimers (Onuma et al. 2002; Osada and Wright 1999). The Xnr3 cleavage mutant (cmXnr3) was previously reported to retain its function as a neural inducer (Ezal et al. 2000). We examined the effects of overexpression cmXtnr3 on Xenopus embryonic development (Fig. 1a–c and Table 1). Western blot analysis using epitope-tagged construct (cmXtnr3-6myc) showed that there are no apparent cryptic processing sites (e.g., Fig. 2h), indicating that cmXtnr3 is indeed a cleavage mutant. cmXtnr3 induced similar anterior defects and tail-like protrusions (Fig. 1c) to those seen with wild-type Xtnr3 (Fig. 1b), but with relatively reduced activity. Xnr3-knockdown embryos injected with specific morpholino oligo have gastrulation and convergent extension defects (Yokota et al. 2003). So it is conceivable that cmXtnr3 did not function in a dominant-negative fashion. To investigate the function of the Xtnr3 mature region, we made a construct encoding the presumptive signal peptide and mature region of Xtnr3 (mXtnr3). A previous study reported that Nodal synthesized without any proregion has strong activity (Le Good et al. 2005). Ectopic expression of mXtnr3 (Fig. 1d and Table 1), like wild-type Xtnr3 (Fig. 1b), induced tail-like protrusions as previously reported (Haramoto et al. 2006). Proregion of Xtnr3 (pXtnr3) did not induce these phenotypes, but pXtnr3-injected embryos have a bump on the head resulting from BMP inhibition (data not shown) (Haramoto et al. 2004). Overexpression of Xnr3 in animal caps induces Xbra expression (Yokota et al. 2003), and ectopic expression of Xbra alone is sufficient to cause formation of a tail-like protrusion (Tada et al. 1997). Next, we used whole-mount in situ hybridization to examine whether overexpression of the Xtnr3 mature region can induce Xbra activation. Messenger RNAs encoding wild-type Xtnr3, cmXtnr3, mXtnr3, and pXtnr3 were overexpressed in the animal pole of one blastomere of two-cell-stage embryos, and the embryos were allowed to develop to stage 12. β-gal mRNA was injected together with each mRNA as a lineage tracer. Compared to controls injected with β-gal mRNA alone (Fig. 1e,e’), ectopic or enhanced expression of Xbra was seen in embryos injected with Xtnr3 (Fig. 1f,f’), cmXtnr3 (Fig. 1g,g’), and mXtnr3 (Fig. 1h,h’) mRNAs, whereas overexpression of pXtnr3 (Fig. 1i,i’) had no apparent effect on ectopic Xbra expression. This finding is consistent with the results that Xtnr3, cmXtnr3, and mXtnr3, but not pXtnr3, can induce tail-like protrusions, [Fig. 1b–d and (Haramoto et al. 2004)]. These experiments indicated that the Xtnr3 mature region is essential for Xbra activation and that the Xtnr3 proregion is dispensable for this activity. Ectopic expression of Xbra was induced in both Xtnr3-injected and Xtnr3-uninjected regions (Fig. 1j,j’). This ectopic Xbra expression is irregular, with Xbra-expressing cells seeming to stray off course. Such a patchy Xbra expression is difficult to be explained by an Xtnr3 gradient from the injected region, so further investigation is required to understand the precise mechanism by which Xtnr3 induces Xbra activation.

Fig. 1
figure 1

Xtnr3 mature region is indispensable for ectopic Xbra expression when overexpressed. a Uninjected control embryo at stage 35. b, c, d Examples of anterior defects and protrusions induced by injection of 1 ng of Xtnr3 mRNA (b), cmXtnr3 mRNA (c), or mXtnr3 mRNA (d) into the animal pole of one blastomere of two-cell-stage embryos. e–j’ Xbra expression in late-gastrula embryos (St. 12). mRNAs were injected into the animal pole of one blastomere of two-cell-stage embryos. β-gal mRNA (250 pg) was injected together with each mRNA (1 ng) as a lineage tracer. In situ hybridization signals are indicated in purple, and the injected regions shown as red. e, e’ Control embryo injected with β-gal mRNA alone. f, f’ Xtnr3-injected embryo. g, g’ cmXtnr3- (cleavage mutant of Xtnr3) injected embryo. h, h’ mXtnr3- (Xtnr3 mature region) injected embryo. i, i’ pXtnr3- (Xtnr3 proregion) injected embryo. e–i dorsal view and e’–i’ lateral view. Xtnr3, cmXtnr3, and mXtnr3 induced ectopic or enhanced expression of Xbra, whereas pXtnr3 did not. j Another example of Xtnr3-injected embryo (lateral view). j’ Magnified view of (j). Ectopic Xbra expression induced by Xtnr3 was detected in both injected (red arrowhead) and uninjected (black arrowhead) regions. The RedGal-labeled region is bordered with a red broken line

Full size image
Table 1 Phenotypes of cmXtnr3- or mXtnr3-injected embryos
Full size table
Fig. 2
figure 2

Xtnr3 mature regions cannot homodimerize. a Diagram of Xtnr3 or Xnr5 constructs used in coimmunoprecipitation assays. Myc tag is shown in dark blue, and HA tag is shown in red. b Uninjected control embryo at stage 35. c, d, e Examples of anterior defect and protrusions induced by injection of 250 pg of Xtnr3 mRNA (c), Xtnr3-6myc mRNA (d), or Xtnr3-HA mRNA (e) into the animal pole of one blastomere of two-cell-stage embryos. f, g Dimerization was determined by coimmunoprecipitation and Western blot analysis. Each RNA (1 ng) was injected into the marginal zone of all blastomeres of four-cell-stage X. laevis embryos. The lysates from stage-10 embryos were immunoprecipitated with anti-HA (f) or anti-myc (g) antibodies. The Xtnr3 mature region exists as a monomer. h Western blot analyses under reducing or nonreducing conditions. Apparent molecular weight of Xtnr3-6myc and cmXtnr3-6myc under reducing and nonreducing conditions did not change, suggesting that Xtnr3 and cmXtnr3 do not form covalent homodimers. Deduced structures of Xtnr3 and Xnr5 are shown at the right side. Myc epitope is shown in dark blue

Full size image

The Xtnr3 mature region cannot form homodimers

cmXtnr3 induced anterior defects and tail-like protrusions (Fig. 1c), like wild-type Xtnr3 (Fig. 1b), indicating that cmXtnr3 did not function in a dominant-negative fashion. So we hypothesized that Nodal-related protein 3 functions as a monomer. To address this, we made constructs of Myc-tagged Xtnr3 and HA-tagged Xtnr3 (Fig. 2a), overexpression of which also induced tail-like protrusions like wild-type Xtnr3 (Fig. 2b–e), and performed coimmunoprecipitation (Fig. 2f,g) and Western blot analyses under reducing and nonreducing conditions (Fig. 2h). Coimmunoprecipitation assays revealed that both the precursor and the mature region of Xnr5 formed homodimers. In contrast, although the Xtnr3 precursor was present in low amounts, the Xtnr3 mature region was not detected in the precipitates (Fig. 2f,g). These results suggest that Xtnr3 precursors may weakly associate with each other but that Xtnr3 mature regions cannot homodimerize. We confirmed this finding in Western blot analyses (Fig. 2h). Under nonreducing conditions, the Xnr5-6myc and cmXnr5-6myc bands were shifted, but the molecular mass of Xtnr3-6myc and cmXtnr3-6myc remained unchanged under reducing and nonreducing conditions. These results indicate that Xtnr3 cannot form disulfide-linked homodimers and confirm that the Xtnr3 mature region functions as a monomer.

Involvement of the seventh cysteine in dimerization

The most divergent feature of Xnr3 compared to other TGF-β family members is the lack of the seventh conserved cysteine residue. We next investigated whether this difference underlies the inability of Xtnr3 to dimerize. In Xtnr3, the amino acid sequence “CGFKDI” occurs at the C terminus. We therefore made mutants with the C terminus mutated to either CGCKDI (+C7) or the shorter sequence, CGCY (+C7sh) (Fig. 3a). The former has the TGF-β-family seventh cysteine instead of phenylalanine, and the latter has the C terminus of wild-type Xnr5. The addition of the seventh cysteine to Xtnr3-6myc or cmXtnr3-6myc did not restore dimerization (Fig. 3b). On the other hand, changing the seventh cysteine of Xnr5-6myc to serine (mC7S) or phenylalanine (mC7F) inhibited dimerization (Fig. 3a,c). This result indicates that the seventh cysteine is essential for Xnr5 dimerization, which raises the possibility that this sequence change also accounts for the lack of dimerization in Xtnr3. However, there appears to be at least one other factor underlying the peculiarity of the Xtnr3 monomeric mature region.

Fig. 3
figure 3

Involvement of the seventh cysteine in dimerization. a C-terminal amino acid sequences of constructed mutants. Mutated residues were shaded. F (position 398) in Xtnr3 and C (position 383) in Xnr5 were mutated to C and S or F, respectively. b, c Western blot analyses under reducing or nonreducing conditions. b A gel shift of the mutants was not detected under nonreducing conditions, indicating that mutations adding the seventh cysteine to Xtnr3-6myc and cmXtnr3-6myc did not restore covalent homodimerization. c A gel shift of the mutants was not detected under nonreducing conditions, indicating that elimination of the seventh cysteine of Xnr5-6myc results in failure to covalently homodimerize

Full size image

Chimeras of Xtnr3 and Xnr5 reveal decisive differences for dimerization

The addition of the seventh cysteine did not enable Xtnr3 or cmXtnr3 to dimerize, so this structural difference is not the only distinctive mechanistic feature of Xtnr3. The TGF-β subfamily proteins, Lefty/Antivin, were thought to function as monomers because they lack the large α-helix required for dimerization with other TGF-β members as well as the cysteine that participates in the covalent stabilization of homodimers or heterodimers (Meno et al. 1996; Sakuma et al. 2002; Thisse and Thisse 1999). From the prediction of protein secondary structure using the New Joint Method served by the Computational Biology Research Center, AIST, Japan (Akiyama et al. 1998), we propose that Xnr3 and Xtnr3 might also lack this α-helix required for dimerization. Another possible feature of Xnr3 and Xtnr3 is a serine residue between the second and third cysteines. All other superfamily members have a glycine in this position that is thought to be required for proper folding of the protein (Griffith et al. 1996; Schlunegger and Grutter 1992). Chimera analyses of Xnr3 and Xnr2 revealed that the region between the second and fifth cysteines of Xnr2 is required for muscle-actin-inducing activity (Ezal et al. 2000). This region contains the conserved glycine and α-helix required for proper folding and dimerization. Therefore, we estimated that, in the case of Nodal-related 3, the lack of this conserved glycine and/or α-helix may be an additional contributor to the failure to dimerize. Xnr5 has strong mesendoderm-inducing activity. Xnr5 is numerously amplified in the Xenopus genome, and its expression pattern is ideal for mesendoderm inducer, indicating that Xnr5 is the endogenous key candidate (Takahashi et al. 2006). So we compared Xtnr3 and Xnr5 and investigated the domains that distinguish the ability to dimerize of these two factors. We made various chimeras of Xtnr3-6myc and Xnr5-6myc (Fig. 4a) and performed a Western blot analysis under reducing and nonreducing conditions (Fig. 4b). Only chimera 7 efficiently formed dimers (Fig. 4a,b), suggesting that all α-helix, conserved glycine, and seventh cysteine are required for dimerization. However, in the case of chimeras 3, 5, and 6, gel shifts were slightly detected when overexposed (data not shown). This α-helix may be the most important among them. Overexpression of all chimeras except chimera 7 induced anterior defects and protrusions (Fig. 5a–h). Chimera 7-injected embryos showed exogastrulation phenotype (Fig. 5i–l) and died at neurula stage, indicating that chimera 7 has different activities from Xtnr3. These results reveal that monomeric Xtnr3 mature region is important to exhibit the activity inducing anterior defects and protrusions. If Nodal-related 3 can form dimers, it will inhibit other Nodal-related proteins or TGF-β superfamily members by forming heterodimers. Monomer is favorable to exhibit its own activity without disturbing other members.

Fig. 4
figure 4

The regions required for dimerization. a Amino acid sequences of Xtnr3/Xnr5 chimeras 1–7. Residues derived from Xnr5 were shaded. b Western blot analysis under reducing or nonreducing conditions. Only chimera 7 efficiently formed dimers

Full size image
Fig. 5
figure 5

Phenotypes induced by chimeras. a Uninjected control embryo at stage 35. b–h Examples of anterior defects and protrusions induced by injection of 500 pg mRNAs; b Xtnr3-6myc, c–h chimeras 1–6, respectively. i Uninjected control embryo at stage 19. j Embryo injected with 500 pg Xtnr3 mRNA showing protrusions and an opened neural plate. k Exogastrulation embryo induced by injection of 500 pg chimera 7 mRNA. l Another view of the exogastrulation embryo shown in (k). mRNAs were injected into the animal pole of both blastomeres of two-cell-stage embryos

Full size image

In this report, we showed that Xtnr3 mature region is essential for ectopic Xbra activation and functions as a monomer. We also found that the lack of all α-helix, conserved glycine, and seventh cysteine defines this specific character of Nodal-related 3. These results will give a different point of view for the direction of the study of Nodal-related 3. The mechanism by which Xbra expression occurs remains to be investigated. A detailed research on Nodal-related 3 may present a unique insight into the divergent activities of TGF-β superfamily members.