It is well established that the BMP4 pathway is essential for proper formation of the dorsoventral axis; overexpression of BMP4 in Xenopus embryos can lead to the formation ventralized embryos, i.e. embryos devoid of axial structures, whereas BMP4 depletion can lead to dorsalized embryos, i.e. embryos exhibiting an expansion of dorsal mesoderm 17181920. At tailbud stages, the expression of Xmab21l2 is restricted to the primitive eye field, optic recess, optic cup and retina (excluding the choroid fissure), to the tectum and dorsal neural tube, and to branchial arches (Figure 1A–E). The early-onset and sustained expression of the MAB21L2 protein in oocyte-, blastula, gastrula, tailbud and tadpole lysates suggests that Mab21l2 is expressed both maternally and zygotically, during and after gastrulation (Figure 1F).

To determine whether Xmab21l2 and BMP4 have a synergistic or antagonistic interaction in vivo, we took a gain of function approach utilizing Xenopus laevis embryos. Preliminary RT-PCR experiments conducted on gastrula lysates injected at the one-cell stage with BMP4 mRNA had shown that, as previously reported by others in studies conducted in the nematode, Xenopus Mab21l2 expression is not controlled by BMP4 signaling (not shown).

Prior to addressing the nature of the interaction between Xmab21l2 and BMP signaling molecules, we analyzed the effects of isolated Xmab21l2 overexpression. Embryos were injected at the one-cell stage with 800 pg in vitro-transcribed Xmab21l2 alone. We analyzed the phenotype of injected tailbud embryos as well as the expression of dorsal markers at gastrulation. A significant percentage of injected embryos (48%, N = 200) featured various degrees of dorsalization/anteriorization (DAI 6–8), with a well formed, sometimes enlarged head, a broad cement gland, a short and/or bent axis (Figure 2B). We prepared embryos for histological examination by sectioning them along the longitudinal plane, both sagittally and horizontally. Histological analysis of horizontal sections revealed many tadpoles featuring an enlarged notochord (nc, fig. 2D), suggestive of a possible defect in convergent extension. Finally, wholemount in situ hybridization experiments conducted on gastrulae with markers of DV polarity revealed an increased percentage of embryos (50%, N = 22) with expanded and/or enhanced Chordin gene expression (blue signal, fig. 2F). Perturbation of Xvent2 gene expression was sporadic, with 5 out of 21 Xmab21l2-injected embryos featuring a reduced Xvent2 expression domain (not shown). Our results are in sufficient agreement with those previously obtained by others 9, confirming that Xmab21l2 overexpression produces neither signs of complete dorsalization, such as Janus twins or radial eyes 21, nor axis duplication.

To determine whether Xmab21l2 can compensate the effects of BMP4 overexpression, Xenopus embryos were injected at the one-cell stage with in vitro-transcribed BMP4 mRNA alone (0.6 ng, n = 111), or coinjected with Xmab21l2 mRNA (0.8 ng, n = 83; 1.6 ng, n = 101). We scored the formation of dorsal structures in tadpole stage embryos by using the dorsoanterior index 21. A complete loss of dorsal structures (DAI = 0) (top in Figure 3A) was observed in 11.4% of the embryos after BMP4 overexpression (0.6 ng) (Figure 3B). Coinjection with 0.8 or 1.6 ng Xmab21l2 drastically reduced this class (0% and 3.1%, respectively). Likewise, whilst complete dorsal structures (DAI = 5) (Figure 3A, bottom) were observed in only 34.1% of BMP4-injected embryos, coinjection with 0.8 or 1.6 ng Xmab21l2 increased this class significantly (65.9% and 59.4%, respectively), suggesting that Xmab21l2 antagonizes BMP4 in its ability to ventralize embryonic mesoderm in vivo (see Figure 3B for details).

Next, by wholemount in situ hybridization, we analyzed the expression of molecular markers of dorsal (chordin, Figure 4A–C) and ventral (Xvent2, Figure 4D–F) mesoderm in wildtype embryos, as well as embryos injected with BMP4 alone or coinjected with Xmab21l2. Whilst the expression domain of chd was sharply reduced in 88% of BMP4-injected embryos, 47% of the embryos co-injected with Xmab21l2 showed wild type chd expression (Figure 4B,C,G). Likewise, 64% of the BMP4-injected embryos showed a dorsal expansion of Xvent2 expression, that was rescued in 83% of the embryos co-injected with Xmab21l2 (Figure 4E,F,H).

The antagonistic interaction between Mab21l2 and BMP4 may be either interpreted as evidence of a molecular interaction between MAB21L2 and some BMP signaling transducer, or alternatively suggest that MAB21L2 and some component of the BMP signaling pathway compete for a common cofactor. It is well established that the DNA-binding protein SMAD1, once phosphorylated at the plasma membrane by the BMP receptor heterotetramer, assembles with the receptor-independent protein SMAD4, enters the nucleus, establishes interactions with various cofactors, and acts by regulating the transcription of a host of target genes [reviewed in 22]. Because vertebrate MAB-21 proteins are mainly or exclusively distributed in the nucleus 6, we investigated the possibility of a molecular interaction between MAB21L2 and SMAD proteins. In order to obtain in vitro evidence of this interaction, P19 cells were either treated with BMP4 or left untreated, and subsequently lysed. P19 cell extracts were then incubated with a resin-bound His-ZZ-tagged form of MAB21L2, or with a control His-ZZ-coupled resin. Eluates were analyzed by immunoblotting with a polyclonal anti-SMAD1 antibody. The experiment revealed a clear interaction between synthetic His-ZZ-MAB21L2 and

endogenous

These results prompted more questions, such as whether the interaction between MAB21L2 and SMAD1 is direct and whether it is dependent on prior formation of a SMAD1-SMAD4 complex. Once again, we resorted to affinity chromatography, incubating in vitro-translated, ³⁵S-Met-labeled SMAD1 and SMAD4 with a resin containing His-ZZ-MAB21L2. This experiment suggested that the interaction with SMAD1 is direct, and that MAB21L2 does not assemble with SMAD4 directly but only through SMAD1, indicating that the formation of a SMAD4-MAB21L2 complex is mediated by SMAD1 (Figure 5B). Indirectly, this experiment also showed that MAB21L2 does not obviously compete with SMAD4 to assemble with SMAD1.

Finally, we investigated whether the in vitro interaction observed between MAB21L2 and SMAD1 could be replicated in vivo. P19 cells co-transfected with myc-MAB21L2 and flag-SMAD1 were either treated with BMP4 or left untreated. In parallel, Xenopus laevis embryos were injected at the four-cell stage into the animal pole of each blastomere with myc-Mab21l2 and flag-Smad1 mRNAs, and coinjected with either BMP4 mRNA or H₂O. The embryos were grown until mid-gastrula stage (st. 10.5/11). Both cell- and embryo lysates were subjected to immunoprecipitation, using an anti flag monoclonal antibody, whereas an anti-myc antibody was utilized for immunodetection in Western blotting. Experiments conducted in transfected P19 cells showed that SMAD1 coprecipitates with MAB21L2, and that the interaction is enhanced by activation of BMP signaling, which is required for nuclear localization of receptor-dependent SMADs (Figure 5C). Likewise, MAB21L2-SMAD1 co-precipitation took place especially, albeit not exclusively, in BMP4-coinjected embryos (Figure 5D). Interaction of the two proteins in the absence of exogenous BMP4 can likely be explained by the abundance of endogenous BMP2/4 ligands in stage 11 gastrulae.

Because in vivo MAB21L2 acts as a BMP4 antagonist, and because BMP4 signaling is transduced in the nucleus by SMAD transactivators, we asked whether MAB21L2 has any intrinsic repressor functions. To address this point, we generated fusion transcripts adding the GAL4 DNA binding domain (residues 1–147) 23, either to the N- or C-terminus of MAB21L2. For as yet unexplained reasons, the latter fusion protein proved extremely unstable and was discarded. Conversely, the N-terminal fusion protein is stable and was shown to bind DNA by gel-retardation assay (Figure 6A). Thus, we were able to assess its activity in cotransfection assays, using as a reporter system the GAL4 target promoter (5XUAS) fused to luciferase. Luciferase activity in Mab21l2-cotransfected COS7 cells was compared to the baseline reporter activity scored in cells transfected with the GAL4 DBD alone (G-D) or with a transcriptionally inactive GAL4 DBD fusion protein, (GAL4HOXB3NT, G-N) 24. Reporter activity was clearly downregulated in Gal4-Mab21l2-transfected cells only, suggesting that MAB21L2 may function as a transcriptional repressor or co-repressor (Figure 6B). The same luciferase assay was performed in P19 cells, co-transfected with plasmid DNA, G-D, and G-M, revealing a 12-fold downregulation of 5XUAS-luc in G-M-transfected cells (not shown). This confirmed the evidence of a strong repressor activity of MAB21l2 when targeted to DNA.

Images were obtained either by traditional photography (slides 1–3) or by electronic scanning of autoradiography films. All images were processed through the Adobe Photoshop 7.0 software or through the Adobe Illustrator 10 software.

The Xenopus laevis homolog of mouse Mab21L2 was isolated by screening a Xenopus stage 28–30 lambdaZapII library (Courtesy of Richard Harland) using the murine homolog as a probe. The 2.8 Kb clone isolated contained the full length Xmab21l2 cDNA (GenBank AF040992) as well as 5' and 3' regions. The coding region of Xmab21 was excised using NotI and BalI (1.3 Kb) and subcloned into pBluescript for whole mount in situ hybridization, or BamHI-StuI (1.1 Kb) and subcloned into pT7TS for overexpression in embryos and into pCDNA3 for cell transfection.

All animal experimentation was conducted according to the stipulations of the Institutional Animal Care and Use Committee, San Raffaele Scientific Institute. In order to obtain embryos, Xenopus females were primed 1000 U of human chorionic gonadotropin (Profasi HP 5000, Serono) the night before collection. Ovulated eggs were fertilized with testis homogenates and allowed to develop in 0.1X MMR (1X MMR is 0.1 M NaCl, 2.0 mM KCl, 1.0 mM MgCl2, 2.0 mM CaCl2, 5 mM HEPES, pH7.4). Jelly coats were removed in 3.2 mM DTT, 0.2 M Tris pH 8.8. Embryos were staged according to Nieuwkoop and Faber 36 and fixed in MEMFA 37 for in situ hybridization.

For histological examination, wt or injected and stained embryos were fixed in MEMFA, embedded in wax, cut into 10 μm sections, dried onto slides, dewaxed with xylene and dehydrated with alcohol. Sections were rehydrated and stained with an orange G solution (2 g orange G, 8 ml glacial acetic acid, 100 ml water) and subsequently with an orange G-aniline blue solution (2 g orange G, 0.5 g aniline blue, 8 ml glacial acetic acid, 100 ml water), and dehydrated 3638.

Whole-mount in situ hybridization was performed on staged embryos as described by 37. The antisense or control sense-strand were generated from the following linearized plasmids: pBS-Xmab21 (antisense linearized NotI, trancribed from T7; sense linearized HindIII, transcribed from T3), Engrailed 2, Krox20, Emx2 (Maria Pannese), CS2+-Chordin (a gift of Stefano Piccolo), Xvent1, Xvent2 (a gift of Christof Niehrs). In single whole-mount hibridizations the probe was labeled with digoxigenin. Digoxigenin-labeled probes were immunolabeled with alkaline phosphatase-conjugated anti-digoxigenin antibody and with the BM Purple substrate (Boehringer). In the double whole-mount one probe was digoxigenin-labeled the second was labeled with fluorescein-UTP. Each probe was used at a concentration of 1 g/ml. Sequential detection with alkaline phosphatase-conjugated anti-fluorescein antibodies (1:8000) and anti-digoxigenin (1:2000) was done. The first color reaction was revealed with NBT/BCIP, while for the second color reaction the Vector Black kit II (Vector Laboratories) was used.

A 14aa C-terminal stretch from the mMab21l2 protein was chosen as immunogenic peptide to obtain antisera in rabbits. Peptide was synthesized, lyophilized and coupled to Keyhole limped hemocyanin (KLH). Coupled peptide was used for immunization of two rabbits. Immune and preimmune serum were controlled by ELISA using the same peptide coated at different serial dilutions. Immune serum was then purified by affinity chromatography: briefly, 20 mg peptide was coupled on CNBr-Sepharose-4B resin (Pharmacia), according to the standard procedure suggested. 50 ml antiserum diluted in 1:1 in 1X PBS was applied on the peptide/CNBr-Sepharose-4B over night at 4°C. After washing the resin with 50 ml PBS and 50 ml 3 M NaCl, bound antibodies were eluted using 100 mM Glycine pH 1.8. Amount, specificity and purity of the antiserum was tested by the Bradford assay, ELISA and SDS-PAGE/Coomassie staining. Purified antibody was tested for specificity in western blot on total protein extracts from cos-7 cells transfected with mMab21l2-myc fusion protein or mock vector, and on total brain extract from mouse embryos. One single band of 41 KDa was detected by anti Mab21l2 antibody. The same 41 kD band was detected by WB in lysates from transfected COS-7 cells, and was missing in untransfected cells. Cross-reactivity of anti mouse Mab21l2 antibody with the Xenopus Mab21l2 protein wastested by Western blot on total extracts from X. laevis embryos. The antibody did not work successfully in IHC or IP experiments, suggesting that it fails to recognize the native epitope.

Protein concentrations in extracts were quantitated through Bradford assays. Extracts were gel-fractionated by denaturing polyacrylamide gel electropohoresis SDS-PAGE. Gels were transfered onto nitrocellulose filters. Even loading was confirmed by Ponceau staining of nitrocellulose filters. Imunostaining was conducted as described 39.

The vector Xmab21-T7TS was linearized with EcoRI and transcribed with the T7 RNA polimerase, N-MycMab21L2-CS2+ was linearized with Asp718 and transcribed with the SP6 RNA polymerase. The Flag-Smad1 was excised from the human construct Flag-Smad1-CMV5 (Courtesy of J. Massagué) and subcloned into the vector pCS2+. The Flag-Smad1-pCS2+ vector was linearized with Asp718 and transcribed with SP6. BMP4 expression constructs were a kind gift of N. Ueno. Capped mRNA was synthesized using the Ambion Message Machine kit according to manufacturer's instructions. Injections were performed in 4% Ficoll in 1X MMR. Embryos were injected animally at the one-cell or two-cell stage into one or both blastomeres. The amount of mRNA injected is given in the text.

COS7 cells were maintained in DMEM Medium (Gibco Brl) supplemented with 10% FBS (Euro Clone), 2 mM L-glutamine (Euro Clone), 100 IU/ml penicillin, and 100 É g/ml streptomycin. Calcium phosphate transfection was performed in a 60 mm Petri dish with different amounts of the expression constructs, 5 É g of reporter plasmids, and 100 ng of pRL-TK (Promega) as an internal control. 18 h after transfection the cells were washed twice with PBS1X and the medium was replaced with DMEM Medium supplemented with 10% FBS, 2 mM L-glutamine (Euro Clone), 100 IU/ml penicillin, and 100 É g/ml streptomycin. Cells were harvested 36 h after the transfection, lysed, and assayed for luciferase activity (Dual Luciferase Reporter Assay System Kit, Promega). All luciferase assays were performed in triplicate.

P19 cells were maintained in MEM alpha medium (Gibco Brl) supplemented with 10% FBS (Euro Clone), 2 mM L-glutamine (Euro Clone), 100 IU/ml penicillin, and 100 μg/ml streptomycin. Calcium phosphate transfection was performed in a 100 mm Petri dish with 10 μg of mycMab21l2 expression construct and 10 μg of flagSmad1 expression construct. 24 h after transfection the medium was replaced with MEM Alpha Medum supplemented with 1% FBS 2 mM L-glutamine (Euro Clone), 100 IU/ and 40 ng/ml BMP4 (R&D Systems). Cells were harvested 36 h after the transfection with TEN Solution (40 mM Tris pH7.5; 1 mM EDTA; 150 mM NaCl) and centrifugated for 15 minutes at 5000 RPM. Cell pellets were frozen at -80°C o/n, thawed at RT and resuspended in 5 volumes of Extraction Buffer (10 mM Hepes pH 7.9; 400 mM NaCl; 0,1 mM EGTA, 5% glicerol). The samples were centrifugated at 34000 RPM for 30 minutes at 4°C and the supernatants were harvested. The concentration of the proteins was determined by BCA assay (Pierce). Xenopus laevis lysates were obtained in the same way starting from injected embryos harvested by centrifugation.

After preclearing with 50 μl of protein G-sepharose o/n at 4°C, lysates were incubated with 25 μg of M2 anti-Flag monoclonal antibody (Sigma) and 50 μl of protein G-Sepharose o/n at 4°C. The resin was washed twice with Dilution Buffer (10 mM Tris-Cl pH8; 140 mM NaCl; 0.025% NaN₃; 0.1% triton), once with TSA solution (10 mM Tris-Cl pH8; 140 mM NaCl; 0.025% NaN₃) and once with 50 mM Tris-Cl pH 6.8. The protein complexes were eluted by addition of Sample buffer (Tris-Cl 125 mM pH 6.8; 0.1 M 2-mercaptoethanol; 2% SDS; 20% glycerol; 25 mg/ml Bromphenol Blue) followed by boiling for 5 minutes and separated on an SDS-polyacrylamide gel followed by anti-myc (1 μg/ml of clone n°9E10) immunoblotting. The secondary antibody was a goat anti-mouse antibody HRP-conjugated (1:30000 Bio-Rad). The blots were developed with the Supersignal West Pico reagent (Pierce).

The protocol was based on manufacturer's recommendations for Ni-NTA Agarose (Qiagen). His-ZZ fusion protein expression in bacterial cultures (37°C, BL21 E. coli) was induced by β-D-thiogalactopyranoside (IPTG) (0.4 mM for His-ZZ-Mab21; 1 mM for His-ZZ and for His-ZZ-Smad1) when the optical density (OD600) reached 0.7–0.9. Cells were harvested after 3 hours, resuspended in lysis buffer (50 mM NaH₂PO₄; 300 mM NaCl; lysozyme 2 mg/ml), and sonicated. The purification protocol was performed with Ni-NTA Agarose (Quiagen) beads and the elution with 250 mM Imidazole (Sigma).

120 μg of His-ZZ-Mab21l2 fusion protein was incubated 1 h at 4°C with 25 μl of IgG-Sepharose beads (Amersham). The beads were recovered by centrifugation and washed ten times with 0.4 M KCl. The beads were incubated 1 h with 250 μg of precleared lysates of P19 treated and untreated with 40 ng/ml of BMP4. The beads were washed 5 times with 0.2 M KCl; the protein complexes were eluted by addition of Sample buffer followed by boiling for 5 minutes, and separated on an SDS-polyacrylamide gel for anti-Smad1 (1:2000 Santa Cruz) immunoblotting. The secondary antibody was a goat anti-rabbit HRP-conjugated antibody (Bio-Rad). The blots were developed with the Supersignal West Pico reagent (Pierce).

120 μg of His-ZZ-Smad1 fusion protein was incubated 1 h at 4°C with 25 μl of IgG-Sepharose resin (Amersham). The beads were recovered by centrifugation and washed ten times with 0.4 M KCl.

The beads were incubated 1 h with 250 μg of precleared lysates of COS7 cells transfected with mycMab21l2. The beads were washed 5 times with 0.2 M KCl; the protein complexes were eluted by addition of Sample buffer,boiled for 5 minutes and separeted on a SDS-polyacrylamide gel for anti-myc immunoblotting.

120 μg of His-ZZ-Mab21l2 fusion protein was incubated 1 h at 4°C with 25 μl of IgG-Sepharose beads (Amersham) in Binding buffer (50 mM Tris pH8; 50 mM KCl; 5 mM MgCl₂; 1 mM DTT; 0.2% NP-40, 10% glycerol). The beads were recovered by centrifugation and washed with Binding buffer. In-vitro translated, ³⁵S-methionine-labeled Smad1 and Smad4 were prepared with the TNT coupled transcription/translation system (Promega). 45 μl of the TNT reaction were mixed with the His-ZZ-Mab21l2 bound to the IgG-Sepharose beads. The mixture was incubated for 2 hours at 4°C in Binding buffer. The Sepharose-protein complex was washed four times with Wash buffer (50 mM Tris pH7.5; 150 mM NaCl; 1 mM EDTA; 0.2% NP-40), eluted by addition of Sample buffer followed by boiling for 5 minutes and separated on SDS-polyacrylamide gels. Gels were fixed and stained with 50% methanol, 10% acetic acid and 0.1% Comassie Blue for 20 minutes. After the incubation with Sodium salicylate 1 M (Fluka) for 20 minutes to improve the radioactive signal the gels were dried and exposed to Kodak X-Omat film o/n at -80°C.