Mesendoderm Induction and Reversal of Left–Right Pattern by Mouse

Developmental Biology 227, 495–509 (2000) doi:10.1006/dbio.2000.9926, available online at http://www.idealibrary.com on View metadata, citation and similar papers at core.ac.uk brought to you by CORE Mesendoderm Induction and Reversal of Left–Rightprovided by Elsevier - Publisher Connector Pattern by Mouse Gdf1, a Vg1-Related Gene

Nancy A. Wall,* Eileen J. Craig,† Patricia A. Labosky,† and Daniel S. Kessler†,1 *Biology Department, Lawrence University, Appleton, Wisconsin 54912; and †Department of Cell & Developmental Biology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104-6058

TGF␤ signals play important roles in establishing the body axes and germ layers in the vertebrate embryo. Vg1 is a TGF␤-related gene that, due to its maternal expression and vegetal localization in Xenopus, has received close examination as a potential regulator of development in Xenopus, zebrafish, and chick. However, a mammalian Vg1 ortholog has not been identified. To isolate mammalian Vg1 we screened a mouse expression library with a Vg1-specific monoclonal antibody and identified a single cross-reactive clone encoding mouse Gdf1. Gdf1 is expressed uniformly throughout the embryonic region at 5.5–6.5 days postcoitum and later in the node, midbrain, spinal cord, paraxial mesoderm, lateral plate mesoderm, and limb bud. When expressed in Xenopus embryos, native GDF1 is not processed, similar to Vg1. In contrast, a chimeric protein containing the prodomain of Xenopus BMP2 fused to the GDF1 mature domain is efficiently processed and signals via Smad2 to induce mesendoderm and axial duplication. Finally, right-sided expression of chimeric GDF1, but not native GDF1, reverses laterality and results in right-sided Xnr1 expression and reversal of intestinal and heart looping. Therefore, GDF1 can regulate left–right patterning, consistent with the Gdf1 loss-of-function analysis in the mouse (C. T. Rankin, T. Bunton, A. M. Lawler, and S. J. Lee, 2000, Nature Genet. 24, 262–265) and a proposed role for Vg1 in Xenopus. Our results establish that Gdf1 is posttranslationally regulated, that mature GDF1 activates a Smad2-dependent signaling pathway, and that mature GDF1 is sufficient to reverse the left–right axis. Moreover, these findings demonstrate that GDF1 and Vg1 are equivalent in biochemical and functional assays, suggesting that Gdf1 provides a Vg1-like function in the mammalian embryo. © 2000 Academic Press Key Words: mouse; Xenopus; Vg1; Gdf1; mesoderm; endoderm; left–right pattern; situs inversus.

INTRODUCTION of a type I and type II receptor. The receptor complex transduces signals via the Smad family of cytoplasmic In the vertebrate embryo, formation of the primary germ proteins to regulate transcriptional targets (Massague, 1998; layers and establishment of the body axes (dorsal–ventral, Whitman, 1998). In the mouse, mutation of TGF␤-related anterior–posterior, and left–right) is initiated in the pregas- ligands, their receptors, and cytoplasmic signaling compo- trula embryo and is completed during the process of gastru- nents has demonstrated multiple roles for TGF␤ pathways lation. These events are regulated in part by secreted in the events of gastrulation and pattern formation (Bed- proteins and several gene families, including WNTs, FGFs, dington and Robertson, 1999). Embryos with mutations in and TGF␤s, have been implicated in the regulation of early type I TGF␤ receptors, including Alk2, Alk3, and Alk4, development in a variety of model systems (Kessler and have defects in gastrulation and mesoderm formation Melton, 1994; Conlon and Beddington, 1995; Heasman, (Mishina et al., 1995; Gu et al., 1998, 1999). Type II TGF␤ 1997). receptors are also required for gastrulation and embryos TGF␤-related proteins such as Activin, Nodal, and the mutant for both the ActRIIA and the ActRIIB genes, or for BMPs, signal through a heterodimeric complex consisting the BMPRII gene, fail to gastrulate or form mesoderm (Song et al., 1999; Beppu et al., 2000). Mutations in Smad2 and 1 To whom correspondence should be addressed. Fax: (215) 573- Smad4 demonstrate that these downstream components 7601. E-mail: [email protected]. are also involved in axis and mesoderm formation (Nomura

0012-1606/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved. 495 496 Wall et al. and Li, 1998; Sirard et al., 1998; Waldrip et al., 1998; (zVg1) and chick (cVg1) using degenerate PCR and low- Weinstein et al., 1998; Yang et al., 1998). stringency hybridization (Helde and Grunwald, 1993; Se- Although many mutations have been generated in TGF␤ leiro et al., 1996; Shah et al., 1997). Like xVg1, zVg1 and ligands that disrupt gastrulation, none of these mutations cVg1 are initially expressed before gastrulation, are post- have a complete absence of mesoderm. Mutations in the translationally regulated, and when processed to form ma- Activin␤A and Activin␤B genes result in mice that survive ture protein are potent mesoderm inducers (Helde and to term (Matzuk et al., 1995a,b). Although Nodal mutant Grunwald, 1993; Dohrmann et al., 1996; Seleiro et al., 1996; embryos have defects in gastrulation related to primitive Shah et al., 1997). While expression of xVg1 and zVg1 streak maintenance, some mesoderm is present and declines at the end of gastrulation, cVg1 expression is NodalϪ/Ϫ ES cells can form differentiated mesoderm (Con- detected at later stages in paraxial mesoderm, neural tis- lon et al., 1994). Embryos mutant for Bmp2 or Bmp4 sues, and branchial arches (Seleiro et al., 1996; Shah et al., initiate gastrulation normally and form mesoderm (Winnier 1997). Two mouse genes, Vgr1 and Vgr2/Gdf3, have been et al., 1995; Zhang and Bradley, 1996; Lawson et al., 1999; named to indicate their Vg1-relatedness (Lyons et al., Kanzler et al., 2000). Even though disruption of TGF␤ 1989a; Jones et al., 1992; McPherron and Lee, 1993). Despite signaling results in defects in mesoderm induction and axis the relation this nomenclature implies, the degree of se- formation, none of the individual TGF␤ ligand mutants quence similarity, the expression patterns, and the bio- show a complete absence of these early gastrulation events, chemical and functional features of Vgr1 and Vgr2/Gdf3 and the analysis of compound mutants effecting gastrula- suggest that these are not orthologs of xVg1 (Lyons et al., tion has so far been limited. Clearly, the isolation and 1989b; Jones et al., 1992; McPherron and Lee, 1993; Mas- analysis of additional mammalian TGF␤ ligands is required sague et al., 1994; Wall and Hogan, 1994). Sequence-based to further elucidate the role of TGF␤ signaling in the mouse screening has been attempted by ourselves and others to embryo. identify a mouse Vg1 ortholog, but no candidate has been TGF␤ signaling is also implicated in left–right (L/R) axis isolated with the high degree of sequence similarity found formation (Burdine and Schier, 2000; Capdevila et al., 2000). for zVg1 and cVg1. Loss-of-function and chimera analyses in the mouse indi- Mammalian Vg1-related genes are potential candidates cate that the TGF␤-related ligands, Nodal, Lefty1, and for regulating the early steps of gastrulation and L/R pat- Lefty2, as well as the downstream components ActRIIB, terning perturbed in TGF␤ receptor and Smad mutants. To Cryptic, Smad2, and Smad5, regulate L/R patterning (Col- isolate mammalian Vg1-related genes an expression library lignon et al., 1996; Oh and Li, 1997; Meno et al., 1998; was screened with a Vg1 antibody and a single clone, Nomura and Li, 1998; Gaio et al., 1999; Heyer et al., 1999; encoding mouse Gdf1, was identified (Lee, 1990). Sequence Chang et al., 2000). Nodal, Lefty1, and Lefty2 are expressed analysis shows that GDF1 shares higher homology with on the left and current models of L/R patterning in the xVg1, zVg1, and cVg1 than with other TGF␤ proteins. We chick embryo implicate an Activin-like factor, acting up- show that Gdf1 expression in mouse embryos is similar to stream of the asymmetric genes, in breaking symmetry to cVg1 expression in the chick. Furthermore, the posttrans- establish the L/R axis. The identity of this upstream TGF␤ lational processing and inducing activities of GDF1 mimic ligand is yet to be determined, but recent analyses suggest that of xVg1 in Xenopus assays. Finally, we demonstrate that the TGF␤-related gene, Gdf1, maybe act early in L/R that mature Gdf1 can reverse L/R pattern in the Xenopus patterning (Rankin et al., 2000; this work). embryo, concordant with the recent loss-of-function analy- Xenopus Vg1 (xVg1) was among the first TGF␤-related sis for Gdf1 in the mouse (Rankin et al., 2000). Therefore, factors identified in vertebrate embryos (Rebagliati et al., GDF1 is an important activator of TGF␤ signaling in the 1985; Weeks and Melton, 1987). The localization of xVg1 to mouse embryo and, either alone or in combination with the vegetal pole of Xenopus oocytes and embryos (Dale et other TGF␤ ligands, regulates formation of the embryonic al., 1989; Tannahill and Melton, 1989) demonstrated an body axes. association with the inducing properties of vegetal blastomeres (Nieuwkoop, 1969). This expression pattern suggests a role for xVg1 in inducing and patterning embryonic endoderm and mesoderm (Vize and Thomsen, 1994). Pro- MATERIALS AND METHODS teolytic processing of xVg1 into its active form is tightly regulated (Dale et al., 1989; Tannahill and Melton, 1989), Western blot analysis. In vitro translated protein was prepared but mature xVg1 protein, produced using chimeric con- using the TNT-coupled transcription–translation system (Pro- structs, is a potent mesendodermal inducer (Dale et al., mega). Templates were pSP64TEN-xVg1, pSP64TEN-zVg1 (Dohr- mann et al., 1996), pSP64TEN-cVg1 (Shah et al., 1997), pSP64T- 1993; Thomsen and Melton, 1993; Kessler and Melton, Activin ␤B (Sokol et al., 1991), and pCS2-Gdf1 (this work). For 1995; Henry et al., 1996; Kessler, 1999). More recently, Xenopus extracts, embryos were lysed in 0.1 M Tris–HCl (pH 6.8) processed xVg1 has been shown to influence L/R patterning supplemented with protease inhibitors (10 ␮l per embryo), the of the heart and viscera in Xenopus (Hyatt et al., 1996; extract was cleared by centrifugation, and half an embryo was Hyatt and Yost, 1998). loaded per well for Western analysis. For mouse extracts, embryos Orthologs of xVg1 have been isolated from zebrafish were lysed in 1ϫ sample buffer (2.5 ␮l per embryo for 7.5 days

postcoitum (days p.c.) and 20 ␮l per embryo for 9.0 days p.c.), genomic DNA was sheared, and four 7.5 days p.c. embryos or half a 9.0 days p.c. embryo was loaded per well for Western analysis. An anti-xVg1 monoclonal antibody (clone 4F6 or D5) (Tannahill and Melton, 1989) was used at a 1:1000 dilution and was detected with a 1:3000 dilution of anti-mouse IgG–peroxidase by chemilumines- cence (Amersham). Cloning of mouse Gdf1. An 8.5 days p.c. mouse cDNA library (a gift of K. Mahon) was screened with an anti-xVg1 antibody (clone D5) and one positive clone was isolated from 1.8 ϫ 106 plaques after three rounds of immunoscreening. The clone was excised as a pBluescript-SK plasmid and both strands were sequenced. The insert was a fusion of a partial mouse Gdf1 cDNA (Lee, 1990) and a mouse Nucleolin cDNA (Bourbon et al., 1988). A full-length Gdf1 ORF was obtained by removing the Nucleolin sequences and combining the 3Ј Gdf1 clone with 5Ј Gdf1 sequences isolated by a standard cDNA library screen. The reconstructed Gdf1 ORF (bases 190–1310) was subcloned into FIG. 1. The Vg1 orthologs are antigenically related and mouse pCS2ϩ (Rupp et al., 1994). To construct the BMP2–GDF1 fusion embryos express a Vg1-cross-reactive protein. (A) Western blot of in a fragment of Xenopus BMP2 (Nishimatsu et al., 1992), contain- vitro translated proteins using an anti-Vg1 monoclonal antibody. ing the signal sequence, prodomain, and cleavage site (residues Lane 1, no template control; lane 2, Activin␤B; lane 3, Xenopus 1–288) was joined by an introduced XhoI site to the GDF1 Vg1; lane 4, zebrafish Vg1; lane 5, chick Vg1. (B) Western blot mature domain (residues 240–357) and subcloned in pCS2ϩ. The analysis of Xenopus and mouse embryo extracts. Lanes 1 and 4, protein sequence at the fusion site is RQKRQARHLEPRVEVG Xenopus gastrula embryo (stage 11); lane 2, 7.5 days p.c. mouse with BMP2 sequence in bold, GDF1 sequence in italic, and an embryo; lane 3, 9.0 days p.c. mouse embryo. Endogenous xVg1 is a introduced leucine residue at the junction. The cleavage site is glycosylated doublet (46–48 kDa) and a major cross-reacting pro- underlined and an alternative cleavage site is present 30 residues tein (44 kDa) was detected in mouse embryo extracts. Size markers upstream of the indicated cleavage site, as described for BMP4 indicated on left (kDa). (Cui et al., 1998). In situ hybridization and RT-PCR. For whole-mount in situ hybridization of mouse embryos a 772-bp SmaI fragment of Gdf1, subcloned into pBluescript-KS, was used as template to synthesize RESULTS digoxygenin-labeled riboprobe. Mouse embryos (6.5–10.5 days p.c.) were fixed and processed as previously described (Hogan et al., GDF1 Is Antigenically Related to xVg1 1994). In situ hybridization of Xenopus embryos was performed using standard methods (Sive et al., 2000). A digoxygenin-labeled The synthesis and posttranslational regulation of xVg1 Xnr1 riboprobe was prepared using a previously described have been analyzed using several monoclonal antisera spe- pBluescript-Xnr1 construct (a gift of C. Wright) (Jones et al., 1995). cific for the mature domain of xVg1 (Dale et al., 1989; For the RT-PCR assay, RNA extraction, cDNA synthesis, and gel Tannahill and Melton, 1989). To determine if the Vg1 electrophoresis were as previously described (Wilson and Melton, orthologs have antigenic relatedness in addition to se- 1994). PCR conditions and primers were as previously described quence similarity, in vitro translated xVg1, zVg1, and cVg1 (Wilson and Melton, 1994; Sasai et al., 1996; Zhang et al., 1998; were analyzed by Western blotting with the 4F6 Vg1 anti- Agius et al., 2000). serum and each protein was recognized (Fig. 1A). Consis- Xenopus embryo culture and microinjection. Xenopus em- tent with previous analysis of zVg1 and cVg1 (Dohrmann et bryos and animal pole explants were collected, fertilized, in- al., 1996; Seleiro et al., 1996; Shah et al., 1997), this result jected, and cultured as previously described (Yao and Kessler, indicates that a conserved epitope is present within the 1999), and embryonic stages were determined according to mature domain of the Vg1 orthologs. Despite similar induc- Nieuwkoop and Faber (1967). Ventral blastomeres were identi- ing activities (Smith et al., 1990; Thomsen et al., 1990), fied at the four-cell stage by pigmentation (Klein, 1987). For Activin␤B did not cross-react with the Vg1 antiserum (Fig. right-sided injection, the right side of the embryo was identified 1A). In addition, other TGF␤-related proteins, including at the four-cell stage relative to dorsal–ventral pigmentation BMP2, BMP4, and Xnr1, were not recognized (data not differences and a single right blastomere was injected adjacent to the second cleavage plane. Capped, in vitro transcribed RNA was shown). Therefore, the Vg1 antiserum detects a specific synthesized using the Message Machine kit (Ambion). Tem- epitope present in the Xenopus, zebrafish, and chick Vg1 ␤ plates for in vitro transcription were pSP64TEN-xVg1 (Dohr- proteins, but not found in a number of other TGF -related mann et al., 1996), pSP64T3-BMP2–Vg1 (Thomsen and Melton, proteins. This result offered a distinct strategy for identify- 1993), pCS2-Xnr1 (Sampath et al., 1997), pCS2-Cerberus-short ing mouse Vg1-related genes based on antigenic relatedness. (Piccolo et al., 1999), pSP64T-⌬1XAR1 (Hemmati-Brivanlou and Extracts of mouse embryos were analyzed by Western Melton, 1992), pCS2-MT-SID (Chen et al., 1997), pCS2-Gdf1, blotting to determine the presence of mouse proteins that and pCS2-BMP2–Gdf1 (this work). cross-react with the Vg1 antiserum. At 7.5 and 9.0 days p.c.

Copyright © 2000 by Academic Press. All rights of reproduction in any form reserved. 498 Wall et al. a single major band of ϳ44 kDa was detected with weaker reacting bands in the range of 29–65 kDa (Fig. 1B). At ϳ44 kDa, the major cross-reacting protein has the approximate size of an unprocessed TGF␤ protein. In addition, no proteins were detected in the range of 14–21 kDa, the predicted size of a processed mature TGF␤ protein. Therefore, the protein detected in mouse extracts by the Vg1 antiserum may be an unprocessed TGF␤-related protein. We note that the cross-reacting mouse protein migrates faster than endogenous xVg1 which migrates as a glycosylated doublet at 46–48 kDa (Fig. 1B) (Dale et al., 1989; Tannahill and Melton, 1989). Since a single major cross-reacting protein was detected in mouse embryo extracts with the Vg1 antiserum, this antibody was used to screen an 8.5 days p.c. mouse expression library. A single positive clone was isolated following three rounds of immunoscreening and sequencing revealed the presence of 902 nucleotides identical to the 3Ј coding sequence of mouse Gdf1 (Lee, 1990) fused to the partial coding sequence of Nucleolin (Bourbon et al., 1988). The second codon present in this partial mouse Gdf1 sequence encoded an inframe methionine and, therefore, this partial clone encoded a truncated GDF1 protein lacking the first 58 amino acids of the wild-type protein (Fig. 2A). A full-length Gdf1 clone was obtained by combining a 5Ј fragment of Gdf1, isolated in a standard cDNA library screen, with the 3Ј sequences isolated in the expression screen. The resulting full-length mouse Gdf1 was sequenced and used in subsequent experiments. To verify that GDF1 protein was recognized by the Vg1 FIG. 2. Mouse GDF1 is antigenically related to xVg1. (A) Protein antiserum, Western blot analysis of in vitro translated sequence alignments of full-length GDF1 and the partial GDF1 proteins was performed. Full-length GDF1, xVg1, and clone isolated by immunoscreening. The internal methionine Activin␤B were produced in vitro and analyzed by Western residue that serves as a start codon in the partial clone is under- blotting. Both GDF1 and xVg1 were detected, while lined. (B) Western blot of in vitro translated proteins with an Activin␤B was not (Fig. 2B). In addition, the truncated form anti-Vg1 monoclonal antibody. Lane 1, no template control; lane 2, Activin␤B; lane 3, Xenopus Vg1; lane 4, full-length GDF1. Size of GDF1 derived from the partial clone was also recognized markers indicated on left (kDa). by the Vg1 antiserum (data not shown). It is interesting to note that the GDF1 in vitro translation product migrates more quickly than the xVg1 protein (44 vs 46 kDa), a difference in size identical to that observed for the cross- explore the possibility of functional homology. The crystal reacting proteins in Xenopus and mouse extracts (Fig. 1B). structures of multiple TGF␤-related proteins have been reported and in each case a carboxy terminal loop, between ␤ sheets 7 and 8, has been identified as a potential receptor Primary Structure of the GDF1 Mature Domain binding site (Schlunegger and Grutter, 1992; Griffith et al., The mature domain of GDF1 was analyzed by BLAST 1996; Hinck et al., 1996; Mittl et al., 1996; Scheufler et al., search to identify related proteins. The most closely related 1999). Of the five residues in this loop, the second and sequences are not mouse TGF␤-related proteins, but are the fourth residues are proposed to be most critical in receptor Vg1 orthologs. While GDF1 is highly related to xVg1, zVg1, specificity (Scheufler et al., 1999). A 20-amino-acid se- and cVg1 (54–57% identity and 70–74% similarity), the quence from the carboxy terminus of xVg1, which includes degree of similarity between xVg1, zVg1, and cVg1 is the putative receptor binding loop, was compared to other significantly higher (69–75% identity and 83–87% similar- TGF␤ family members. The results show that, following ity) and, therefore, GDF1 does not appear to be a mouse zVg1 and cVg1, the most closely related molecule is GDF1 ortholog of Vg1 at the primary sequence level (Fig. 3A). (Fig. 3B). Within the receptor binding loop, GDF1 was Given that the specificity of TGF␤ function is determined identical at four of five residues, including the critical by the binding of the mature domain to specific receptors second and fourth residues. While GDF1 differs from xVg1 (Massague, 1998; Massague and Chen, 2000), we examined and zVg1 at residue three of the loop, GDF1 and cVg1 are the putative receptor binding regions of GDF1 and Vg1 to identical at all five positions. Therefore, the putative recep-

expression becomes restricted to speciﬁc tissues including the node (Fig. 4D), midbrain (Fig. 4C, arrowhead), developing spinal cord (Fig. 4C, arrow), and areas of the somitic and lateral plate mesoderm (Fig. 4C, asterisk). At 9.5 days p.c. Gdf1 transcripts are detected in the tailbud, ventrolateral mesoderm, posterior notochord, and posterior spinal cord (Figs. 4E and 4F). Expression is also maintained in the ventral midbrain at the cephalic ﬂexure. By 10.5 days p.c. Gdf1 is weakly expressed in the midbrain and more strongly expressed in the posterior limb bud (Fig. 4G). The expression of Gdf1 prior to and during gastrulation suggests a role for Gdf1 in patterning the early mouse embryo. Furthermore, the sites of Gdf1 expression in the mouse are strikingly similar to the expression of cVg1 in the chick (Seleiro et al., 1996; Shah et al., 1997). These results are consistent with the recent analysis by Rankin et al. (2000).

Regulation of GDF1 Processing TGF␤-related proteins are synthesized as preproproteins FIG. 3. Sequence similarity of the GDF1 mature domain and the that undergo maturation by endoproteolytic cleavage, re- Vg1 orthologs. (A) Percentage of identity and similarity compari- leasing the C-terminal region as active mature protein sons (identity/similarity) of GDF1 with GDF and Vg1 proteins. (B) (Gentry et al., 1988). While many members of the TGF␤ BLAST results for the carboxy-terminal 20 amino acids of xVg1 superfamily, such as Activin␤B, are constitutively pro- containing a putative receptor binding site. Amino acids high- lighted in gray indicate the five-residue loop between ␤ sheets 7 and cessed, proteolysis is an important step in the posttransla- ␤ 8 and the two residues in light gray are proposed to be critical in tional regulation of a subset of TGF -related proteins, conferring receptor binding specificity. x, Xenopus; z, zebrafish; sb, including Vg1, Nodal, and BMPs (Sha et al., 1989; Cui et al., red sea bream; n, Japanese common newt; c, chick; m, mouse; h, 1998; Constam and Robertson, 1999). In Xenopus, xVg1 is human; su, sea urchin; am, amphioxus Branchiostoma belcheri; present as an abundant precursor protein, but no processed, amphi, amphioxus Branchiostoma floridae; s, European starfish; mature protein is detectable (Dale et al., 1989; Tannahill as, ascidian Ptychodera flava. and Melton, 1989). Similarly, endogenous zVg1 and cVg1 accumulate as unprocessed precursors (Dohrmann et al., 1996; Seleiro et al., 1996; Shah et al., 1997), suggesting that regulation of Vg1 processing is a conserved mechanism to tor binding loops of GDF1 and the Vg1 orthologs are highly temporally or spatially limit production of active Vg1. To conserved, supporting the possibility that Gdf1 is a func- determine whether proteolytic cleavage of GDF1 is regu- tional homolog of Vg1. We note that mouse Vgr2/Gdf3 lated, GDF1 was expressed in Xenopus embryos. At the (Jones et al., 1992; McPherron and Lee, 1993), another one-cell stage embryos were injected with Gdf1 mRNA and mouse gene highly related to xVg1, differs at three of five extracts were analyzed by Western blotting at the gastrula residues of the receptor binding loop, including the critical stage. GDF1 precursor was detected as an abundant 44-kDa second and fourth positions (Fig. 3B). protein and, like xVg1, mature protein was not detected (Fig. 5A, lanes 1 and 3). This result suggests that processing of GDF1 is regulated, consistent with the detection of a Expression of Gdf1 precursor-sized protein, but no mature protein, in mouse First cloned by Lee (1990), Gdf1 mRNA was shown by embryo extracts (Fig. 1B). Northern blotting to be expressed in embryos from 8.5 to Chimeric proteins containing the BMP2 or Activin␤B 18.5 days p.c. Additionally, at 14.5 days p.c. GDF1 protein is prodomain fused to the mature domain of xVg1 have been expressed in the brain, spinal cord, and peripheral nerves used to overcome the regulation of processing and facilitate (Lee, 1991). To determine if Gdf1 was expressed early in the functional analysis of mature Vg1 (Fig. 5A, lane 2) (Dale mouse development, consistent with a potential role in et al., 1993; Thomsen and Melton, 1993; Kessler and early patterning, the spatial expression pattern of Gdf1 was Melton, 1995). In an attempt to generate mature GDF1 determined by in situ hybridization of 5.5–10.5 days p.c. protein, the signal sequence, prodomain, and cleavage site embryos. Gdf1 is expressed ubiquitously in the embryonic of BMP2 were fused to the mature domain of GDF1 (BMP2– region of 5.5–6.5 days p.c. embryos, but excluded from GDF1). BMP2–GDF1 was expressed in Xenopus embryos extraembryonic tissues (Fig. 4A). Ubiquitous expression and both precursor (46 kDa) and mature protein were continues at 7.5–8.0 days p.c. with more robust expression detected (18 kDa) (Fig. 5A, lane 4). Therefore, efficient in the developing head folds (Fig. 4B). By 8.5 days p.c. processing of GDF1 can be accomplished by substituting

Copyright © 2000 by Academic Press. All rights of reproduction in any form reserved. Induction and Patterning by Gdf1 501 the N-terminal sequences of GDF1 with those of BMP2, native GDF1, BMP2–GDF1, or BMP2–Vg1. At the tadpole consistent with a proposed role for the prodomain in stage no ectopic axes were observed in response to native regulating TGF␤ processing (Gray and Mason, 1990; Cui et GDF1 (0%, n ϭ 29; Fig. 6C), while BMP2-GDF1 induced al., 1998; Constam and Robertson, 1999). ectopic axial structures at high frequency (62%, n ϭ 30; Fig. 6D), similar to the response to BMP2-Vg1 (69%, n ϭ 29; Fig. 6B). At the doses used, most BMP2–GDF1-induced Dorsal Mesendoderm Induction by Mature GDF1 axes had an anterior limit at the midbrain/otic vesicle, but The inducing activity of GDF1 was assessed by expres- ectopic head structures were observed in a minority (3–7%) sion in Xenopus animal pole explants. At the one-cell stage of the injected embryos (data not shown). BMP2–GDF1- the animal region was injected with mRNA encoding induced axes contained somitic muscle and neural tube, but native GDF1, BMP2–GDF1, or BMP2–Vg1. Explants were most lacked notochord (data not shown). Therefore, mature prepared from injected and uninjected embryos at the GDF1 protein is an inducer of mesendoderm in both ex- blastula stage and cultured to the gastrula or tailbud stages plants and intact embryos, identical to the activity of for analysis (Figs. 5B–5F). At the tailbud stage, explants mature xVg1. expressing native GDF1 were indistinguishable from uninjected explants (Figs. 5B and 5D), consistent with the failure Smad2-Dependent Signaling by Mature GDF1 of GDF1 processing. In contrast, explants expressing BMP2– GDF1 underwent convergent-extension movements char- The dorsal mesoderm inducing activity of mature GDF1 acteristic of dorsal mesoderm induction, identical to the is similar to that observed for a subgroup of the TGF␤ response to BMP2–Vg1 (Figs. 5C and 5E). Explants were superfamily, including Activin, Nodal, and mature Vg1. analyzed by RT-PCR for tissue-specific marker expression These TGF␤ ligands activate signals that are transduced via at the gastrula and tailbud stages (Fig. 5F). Consistent with a Smad2-containing complex (Massague, 1998; Whitman, the morphogenetic response of explants, BMP2–GDF1 and 1998). To assess the signaling properties of GDF1, BMP2– BMP2–Vg1 induced expression of both mesodermal and GDF1 was coexpressed in animal pole explants with three endodermal markers, including Brachyury, Goosecoid, distinct inhibitors of TGF␤ signaling. A truncated form of Sox17, Xnr1, and VegT at the gastrula stage and Muscle Cerberus (Cer-short) is a secreted factor that specifically Actin and Collagen Type II at the tailbud stage. The binds to and inhibits Nodal, but does not interfere with presence of differentiated muscle, notochord, and neural Activin or Vg1 activity (Piccolo et al., 1999). A truncated tissue in BMP2–GDF1-expressing explants was confirmed form of an Activin type II receptor (⌬1XAR1) that lacks the by histology (data not shown). In contrast, native GDF1 did cytoplasmic kinase domain inhibits signaling by several not induce mesendodermal marker expression or the differ- TGF␤-related proteins including Activin (Hemmati- entiation of mesodermal or neural tissues (Fig. 5F and data Brivanlou and Melton, 1992), Vg1 (Schulte-Merker et al., not shown). 1994; Kessler and Melton, 1995), BMP4 (Wilson and To further assess the activity of GDF1, native or chimeric Hemmati-Brivanlou, 1995), and Nodal (this paper). The GDF1 was expressed in ventral mesoderm and axial devel- Smad2-interaction domain (SID) of FAST1 inhibits forma- opment was examined. At the four-cell stage a single tion of a functional Smad2–Smad4 complex, thus blocking ventral blastomere was injected with mRNA encoding signaling by Smad2-dependent TGF␤ proteins such as Ac-

FIG. 4. Expression pattern of Gdf1 mRNA in the mouse embryo. (A) 5.5 (left) and 6.5 (right) days p.c. embryos, lateral views. Line indicates the boundary between extraembryonic (top) and embryonic (bottom) tissues. (B) 7.5 days p.c. embryo, lateral view (white asterisk, head fold; al, allantois). (C) 8.5 days p.c. embryo, ventral view (white arrowhead, presumptive midbrain; white arrow, neural tube; white asterisk, lateral plate mesoderm). (D) 8.5 days p.c. embryo (dorsal view, anterior to left) showing Gdf1-positive cells at the node (arrowhead). (E) 9.5 days p.c. embryo, lateral view (white arrowhead, ventral midbrain; white asterisk, lateral plate mesoderm). (F) 9.5 days p.c. embryo, dorsal view (white arrowhead, ventral midbrain; black arrowhead, notochord). (G) 10.5 days p.c. embryo, lateral view (black arrow, posterior limb bud). Bars, 50␮m. FIG. 5. Posttranslational processing and mesendoderm-inducing activity of GDF1. (A) Embryos were injected at the one-cell stage with 2 ng of RNA encoding Xenopus Vg1 (xVg1), BMP2–Vg1 (BVg1), GDF1, or BMP2–GDF1 (BGdf1) and extracts were prepared at the gastrula stage for Western blotting with an anti-Vg1 monoclonal antibody. Extracts of uninjected embryos were also analyzed (Control). The unprocessed precursor proteins (P) and processed mature proteins (M) are indicated. Size markers indicated on left (kDa). (B–E) Embryos were injected at the one-cell stage with 100 pg of RNA encoding BMP2–Vg1 (C), GDF1 (D), or BMP2–GDF1 (E), animal pole explants were isolated from injected and uninjected embryos (B) at the blastula stage and were cultured to the tailbud stage (stage 25). (F) Embryos were injected as above and explants were collected at the gastrula stage (top panel) or tailbud stage (bottom panel) for RT-PCR analysis. Gastrula markers are Brachyury (Xbra, mesoderm), Goosecoid (Gsc, dorsal mesoderm), Sox17 (endoderm), Nodal-related-1 (Xnr1, mesendoderm), and VegT (mesendoderm). Tailbud markers are Muscle Actin (M. Actin, somitic muscle) and Collagen type II (Coll II, notochord). EF1␣ is a control for RNA recovery and loading. Intact embryos (Embryo) served as a positive control and an identical reaction without reverse transcriptase controlled for PCR contamination (Embryo-RT). Bar, 400␮m.

visceral organs relative to the L/R axis (Burdine and Schier, 2000; Capdevila et al., 2000). Experiments in Xenopus suggest that xVg1, or a Vg1-like factor, acts upstream of asymmetrically expressed genes, like Nodal, to establish L/R pattern (Hyatt et al., 1996; Hyatt and Yost, 1998). The ability of GDF1 to alter L/R patterning in Xenopus was examined by expressing mature GDF1 on the right. At the four-cell stage a single right blastomere was injected adjacent to the second cleavage plane with native GDF1, BMP2–GDF1, or BMP2–Vg1. At the tailbud stage Xnr1 expression, normally restricted to the left lateral plate mesoderm (Lohr et al., 1997; Sampath et al., 1997), was examined by in situ hybridization. Right-sided expression of BMP2–GDF1 resulted in altered Xnr1 expression in 71% of injected embryos (n ϭ 28), with 46% expressing Xnr1 bilaterally, 7% in the right lateral plate mesoderm, and 18% lacking Xnr1 expression on the left and right (Figs. 8J and 8K). Right-sided expression of BMP2–Vg1 also altered Xnr1 expression with 85% of injected embryos (n ϭ 13) expressing Xnr1 bilaterally (Fig. 8D and 8E). In contrast, Xnr1 expression in uninjected and native GDF1-injected embryos was normal in nearly every case (93%, n ϭ 14; and 97%, n ϭ 29, respectively) (Figs. 8A, 8B, 8G, and 8H). The influence of mature GDF1 on L/R asymmetry was also assessed during the period of organogenesis (late tadpole stage) by examining the direction of intestinal and heart looping. Consistent with the alteration of Xnr1 FIG. 6. Axis induction by mature GDF1. At the four-cell stage a expression, BMP2–GDF1 and BMP2–Vg1 induced a rever- single ventral blastomere was injected with 100 pg of RNA encod- sal of heart and intestinal looping. Situs inversus (both ing BMP2–Vg1 (B), native GDF1 (C), or BMP2–GDF1 (D). (A) Uninjected embryo (Control). Arrowhead indicates anterior end of organs reversed) was observed in 17% of BMP2–GDF1- ϭ ectopic axis. Bar, 500 ␮m. injected embryos (n 29) and 15% of BMP2–Vg1- injected embryos (n ϭ 13), but not in embryos injected with native GDF1 (0%, n ϭ 35) (Figs. 8F, 8I, and 8L and data not shown). Heterotaxia (heart or intestine reversed) tivin (Chen et al., 1997). At the one-cell stage the animal pole region was injected with BMP2–GDF1, BMP2–Vg1, or Xnr1 alone or in combination with each inhibitor. Animal pole explants were prepared at the blastula stage and expression of Brachyury and Goosecoid was examined at the gastrula stage by RT-PCR (Fig. 7). Mesoderm induction by BMP2–GDF1 was blocked by ⌬1XAR1 and SID, but not Cer-short, identical to the results obtained for BMP2–Vg1. The response to Xnr1 was blocked by all three inhibitors. The inhibition of GDF1 activity by the SID confirms that mature GDF1 signals in a Smad2-dependent manner. In addition, the identical inhibition profiles for mature GDF1 and xVg1 supports the idea that GDF1 may be a functional homolog of xVg1.

FIG. 7. Smad2-dependent signaling by mature GDF1. Embryos Reversal of Left–Right Patterning by Mature GDF1 were injected at the one-cell stage with 100 pg of BMP2–Vg1 (BVg1), BMP2–GDF1 (BGdf1), or Nodal-related-1 (Xnr1) alone or in combi- During vertebrate embryogenesis, TGF␤ signals function nation with 1 ng of a truncated form of Cerberus (CerS), 2 ng of a early to establish and pattern the germ layers and later to truncated Activin type II receptor (⌬1XAR1), or 2 ng of the regulate tissue differentiation and morphogenesis resulting Smad2-interaction domain of FAST1 (SID). Explants were collected in normal organogenesis. In vertebrates, Nodal-related at the gastrula stage for RT-PCR analysis of Brachyury (Xbra) and genes are expressed in the left lateral plate mesoderm and Goosecoid (Gsc) expression. Positive and negative PCR controls are are implicated in the morphogenesis and positioning of the as described in the legend to Fig. 5.

FIG. 8. Reversal of left–right patterning by mature GDF1. A single right blastomere was injected at the four-cell with 100 pg of BMP2–Vg1 (D–F), GDF1 (G–I), or BMP2–GDF1 (J–L) or was not injected (A–C, Control). At the early tailbud stage (stage 23–24) expression of Xnr1 in lateral plate mesoderm was examined by in situ hybridization (A, B, D, E, G, H, J, K). Left lateral (A, D, G, J) and right lateral (B, E, H, K) views of single embryos are presented. At the late tadpole stage (stage 45) the direction of intestinal looping was examined. Ventral view (C, F, I, L); arrow indicates direction of looping. Bar, 300 ␮m (lateral views) or 400 ␮m (ventral views).

was observed in 79% of BMP2–GDF1-injected embryos DISCUSSION (48% intestine/31% heart) and 85% of BMP2–Vg1- injected embryos (46% intestine/38% heart). None of the The TGF␤ superfamily is a conserved collection of sig- GDF1-injected embryos (n ϭ 35) and only a single naling molecules that act individually and in combination uninjected embryo (n ϭ 37) had reversal of heart or to regulate vertebrate and invertebrate development. The intestinal looping (Figs. 8C and 8I and data not shown). cellular response to TGF␤ ligands is mediated by families of We note that at the doses used very few embryos dis- cell surface, cytoplasmic, and nuclear proteins that consti- played morphological defects in axis formation and orga- tute a conserved signal transduction cassette that has been nogenesis, but those that did were not scored for lateral- used many times in the evolution of animal development ity. The high frequency of heterotaxia indicates that (Massague, 1998; Whitman, 1998). Here we report the right-sided expression of BMP2–GDF1 or BMP2–Vg1 re- identiﬁcation and functional analysis of GDF1, a mouse sults in randomization of L/R pattern. This result con- protein antigenically related to xVg1. While protein se- trasts with previous reports that BMP2–Vg1 induces situs quence comparisons of GDF1 and xVg1 argue against gene inversus (Hyatt et al., 1996; Hyatt and Yost, 1998), but orthology, we describe regulatory and functional similari- differences in the stage and site of injection may account ties that suggest functional homology for GDF1 and xVg1. for the distinct results. Therefore, right-sided expression Like the Vg1 orthologs, Gdf1 is broadly expressed in the of mature GDF1 can alter laterality in Xenopus, consis- early embryo, GDF1 is regulated at the posttranslational tent with loss-of-function analysis in the mouse demon- level, and following maturation, GDF1 signals via Smad2 to strating that Gdf1 is essential for normal L/R patterning induce mesendodermal fates and alter L/R pattern. In light (Rankin et al., 2000). of our gain-of-function results in Xenopus and recent Gdf1

Copyright © 2000 by Academic Press. All rights of reproduction in any form reserved. 504 Wall et al. loss-of-function analysis in the mouse (Rankin et al., 2000), Activin␤A/Activin␤B mutant mice (Matzuk et al., we discuss the role of Gdf1 in embryonic patterning and the 1995a,b). potential significance of the Gdf1–Vg1 relationship. In light of the “nodal flow” model and the potential Vertebrate embryos display bilateral symmetry exter- involvement of an Activin-like signal in breaking symme- nally, but the internal organs, including the heart, lungs, try, mouse Gdf1 is a candidate for an early signal in L/R liver, pancreas, and intestines, show L/R asymmetry in patterning. Moreover, Gdf1Ϫ/Ϫ mice display a variety of structure and/or position. The L/R axis is established early L/R patterning defects, including situs inversus of the during embryogenesis and asymmetric gene expression is visceral organs, pulmonary isomersim, and cardiac defects first detected within or adjacent to the node (Burdine and (Rankin et al., 2000). In addition, left-sided expression of Schier, 2000; Capdevila et al., 2000). In the chick, asym- Nodal, Lefty1, Lefty2, and Pitx2 was absent in Gdf1Ϫ/Ϫ metric expression of Sonic Hedgehog within the node leads embryos. This loss of left-sided gene expression, as opposed to perinodal expression of Nodal on the left (Levin et al., to right-sided or bilateral expression observed in other L/R 1995; Pagan-Westphal and Tabin, 1998). This local expres- mutants, suggests that Gdf1 acts upstream of the asymmet- sion of Nodal maintains Caronte expression on the left and ric genes and may play a role in establishing the L/R axis. this Cerberus-related BMP-inhibitor (Rodriguez Esteban et Unlike other TGF␤ factors involved in L/R patterning, Gdf1 al., 1999; Yokouchi et al., 1999) induces a broad Nodal mRNA is expressed symmetrically with respect to the L/R expression domain in the left lateral plate mesoderm, a axis at all stages. Gdf1 is expressed throughout the embryo feature common to all vertebrates (Collignon et al., 1996; proper during gastrulation and is later enriched in cells of Lowe et al., 1996; Lohr et al., 1997; Sampath et al., 1997; the node. Significantly, Gdf1 is expressed by node cells Rebagliati et al., 1998). Lateral Nodal expression induces during the period of nodal flow and prior to the onset of left-sided expression of Pitx2 in the heart, gut, and stomach asymmetric gene expression and therefore, Gdf1 is present primordia, resulting in asymmetric morphogenesis of these at the appropriate place and time to influence L/R axis organs (Logan et al., 1998; Piedra et al., 1998; Ryan et al., formation. Furthermore, our results demonstrate that 1998; Yoshioka et al., 1998; Campione et al., 1999; Kita- GDF1 has Activin-like signaling properties. We have shown mura et al., 1999; Lin et al., 1999). The left-sided signals are that mature GDF1 is a potent inducer of mesendoderm and negatively regulated by the TGF␤-related proteins Lefty1, axis formation and this activity is blocked by the FAST1 which prevents Caronte from acting across the midline, and Smad2-interaction domain, demonstrating that GDF1 sig- Lefty2, which competes with Nodal for receptor binding nals via Smad2. In addition, GDF1 is inhibited by a trun- (Meno et al., 1996, 1998, 1999; Ishimaru et al., 2000). cated Activin type II receptor, consistent with the involve- Components of this highly conserved regulatory network ment of the closely related receptor, ActRIIB, in L/R of asymmetric gene expression have been studied by gain- patterning of the mouse. Finally, right-sided expression of of-function in the chick and Xenopus, and loss-of-function mature GDF1 in Xenopus results in Xnr1 expression on the in the mouse, and altered expression of each of the genes right and reversal of intestinal and heart looping. Therefore, discussed above perturbs L/R pattern. In the mouse, the L/R Gdf1 is both necessary and sufficient for L/R patterning and axis is defined as the A/P and D/V axes are established, but may provide the proposed Activin-like function that con- the embryo appears bilaterally symmetrical for several fers L/R asymmetry on the node. additional days (Beddington and Robertson, 1999). The The symmetrical expression of Gdf1 mRNA raises an upstream events that break bilateral symmetry and result intriguing question: how does a uniformly distributed fac- in asymmetric gene expression at the node are not well tor confer asymmetric pattern? Regulation by translational understood. In the mouse node a single monocilium control, proteolytic processing, or protein stability could projects from each cell and the concerted movement of result in asymmetric expression or activity of GDF1 pro- these cilia generates a leftward flow of extracellular fluid. tein. The absence of mature GDF1 protein in mouse em- This may result in the asymmetric distribution of a se- bryos and Xenopus embryos overexpressing GDF1 could creted factor that regulates L/R polarity. Supporting this result from tight regulation of proteolytic cleavage or rapid “nodal flow” hypothesis, mutations in specific kinesin and degradation of mature protein once produced. For BMP4 and dynein genes disrupt nodal cilia formation and function and Nodal, the cleavage site sequence and the presence of result in L/R defects (Supp et al., 1997; Nonaka et al., 1998; alternative cleavage sites determines cleavage efficiency Marszalek et al., 1999; Supp et al., 1999; Takeda et al., (Cui et al., 1998; Constam and Robertson, 1999). In addi- 1999; Wagner and Yost, 2000). In addition, an Activin type tion, stability of mature Nodal protein is effected by prodo- II receptor, ActRIIB, is expressed within the node and is main sequences of the Nodal precursor (Constam and required for normal L/R patterning, suggesting that an Robertson, 1999). Although unresolved at this point, we Activin-like signal generates node asymmetry (Oh and Li, favor control of GDF1 cleavage as a regulatory mechanism 1997). Consistent with the involvement of Activin-like for the following reasons. If GDF1 was processed and the signals, L/R patterning defects were observed in chimeric mature domain rapidly degraded, the prodomain might mice with Smad2-deficient embryonic tissues (Heyer et al., accumulate as a product of precursor cleavage. Metabolic 1999). However, the identity of the Activin-like ligand is labeling of Xenopus oocytes expressing native GDF1 did not yet to be determined given the absence of L/R defects in reveal the accumulation of the prodomain (data not shown).

Furthermore, despite the instability of mature Nodal, na- receptor, in which maternal xVg1 is selectively processed tive Nodal does induce mesoderm in Xenopus, so mature on the left to generate L/R asymmetry. However, mature protein must accumulate to levels sufficient for biological xVg1 has not been detected in left blastomeres of the response (Jones et al., 1995). We observed no response to Xenopus embryo; mature xVg1 stimulates a Smad2- native GDF1 even at high expression levels (data not dependent signal identical to that produced by Nodal- shown), suggesting regulation of GDF1 processing rather related proteins such as Xnr1; and the truncated Activin than stability. Consistent with regulation of GDF1 cleavage receptor used to inhibit endogenous xVg1 also inhibits in L/R patterning, loss-of-function analyses for two en- other TGF␤-related ligands, including Xnr1. For these rea- zymes that cleave TGF␤ precursors, the proprotein conver- sons, it is possible that mature xVg1 phenocopies the tases SPC1/Furin and SPC4/Pace4, result in L/R defects, activity of Xnr1 or other TGF␤ ligands. For example, the although these proteins may regulate several TGF␤ pro- right-sided injection of Activin protein was shown to teins, including Nodal and GDF1 (Roebroek et al., 1998; mimic Nodal activity and reverse L/R patterning (Toyoi- Constam and Robertson, 2000a,b). zumi et al., 2000). Alternatively, given the similarities of On the other hand, if GDF1 is constitutively processed xVg1 and Gdf1, xVg1 may phenocopy an, as yet unidenti- and stable, other mechanisms could generate asymmetric fied, Xenopus GDF1 activity that acts upstream of Xnr1. It GDF1 activity. Given that Gdf1 mRNA is uniformly ex- should be noted, however, that our studies do not distin- pressed during the period of nodal flow, the leftward flow of guish between a specific activity of mature GDF1 and an extracellular fluid could result in asymmetrical accumula- ability to phenocopy Xnr1 activity. tion of mature GDF1 on the left side of the node and lead to Analysis of the Vg1 orthologs has demonstrated a conser- asymmetrical gene expression. Alternatively, asymmetric vation of expression pattern, posttranslational regulation, expression of a GDF1 inhibitor could produce an asymme- and inducing activity, suggesting a conservation of Vg1 try of GDF1 activity. function as well. In addition, the successful isolation of Vg1 We identified GDF1 based on antigenic relatedness to orthologs from the zebrafish and chick argues for the xVg1, and the analysis of GDF1 function and regulation has existence of a mouse ortholog. A number of reasons may revealed several similarities to xVg1 including: pregastrula account for the difficulty in identifying mouse Vg1, includ- expression; sequence conservation of the mature domain ing very low mRNA abundance or unexpectedly high se- and the putative receptor binding loop; regulation of pro- quence divergence. The attempt to identify mouse Vg1 teolytic cleavage; mesendoderm and axis induction; and resulted in the identification of Gdf1, a gene with striking L/R axis reversal. Previous analyses of Vg1 in Xenopus, sequence, regulatory, and functional similarity to the Vg1 zebrafish, and chick have suggested a number of potential orthologs, and our analysis of Gdf1 points to an important functions in vertebrate development. Gain-of-function ex- role in establishing the L/R axis. Although not a mamma- periments have suggested a role for Vg1 in Nieuwkoop lian Vg1 ortholog, Gdf1 may be a functional homolog of Vg1 Center formation and/or induction of mesendoderm (Dale that regulates development of the early mouse embryo. et al., 1993; Thomsen and Melton, 1993; Schulte-Merker et Further efforts will be required to identify the elusive al., 1994; Kessler and Melton, 1995; Cui et al., 1996; mouse Vg1 gene and determine the role of Vg1-related genes Dohrmann et al., 1996; Henry et al., 1996; Seleiro et al., in vertebrate development. 1996; Shah et al., 1997; Kessler, 1999). A role for xVg1 in mesendoderm formation is supported by studies using dominant inhibitory mutants of xVg1 and downstream ACKNOWLEDGMENTS signaling components (Kessler and Melton, 1995; Henry et al., 1996; Henry and Melton, 1998; Joseph and Melton, We are grateful to Beth DeStasio, Mark Engleka, and Peter Klein for constructive comments on the manuscript. We thank Ruth 1998; Hoodless et al., 1999; Kessler, 1999). In particular, Foreman and Valerie Cluzet for assistance with the mouse in situ Xenopus mesendoderm formation can be inhibited by a experiments. We thank Eddy De Robertis, Stephano Piccolo, Mal- dominant negative form of xVg1 that blocks the activity of colm Whitman, and Chris Wright for providing plasmids. This xVg1, but not other TGF␤ ligands (Joseph and Melton, work was supported by grants from the NIH (HD35159), the 1998). Analyses of VegT, a maternally expressed Xenopus American Digestive Health Foundation, the Pew Scholars Program, T-box gene (Zhang et al., 1998), and Nodal-related genes and the American Heart Association. (Schier and Shen, 2000) have established these genes as regulators of mesendodermal development, raising the possibility that xVg1 cooperates with VegT and/or Nodal to REFERENCES initiate mesendoderm formation. Agius, E., Oelgeschlager, M., Wessely, O., Kemp, C., and De A role for xVg1 in establishing the L/R axis has also been Robertis, E. M. (2000). Endodermal Nodal-related signals and proposed (Hyatt et al., 1996; Hyatt and Yost, 1998), and this mesoderm induction in Xenopus. Development 127, 1173–1183. is noteworthy given the involvement of Gdf1 in L/R pat- Beddington, R. S., and Robertson, E. J. (1999). Axis development terning and the similarities between GDF1 and xVg1. Yost and early asymmetry in mammals. Cell 96, 195–209. and colleagues have proposed a model, based on ectopic Beppu, H., Kawabata, M., Hamamoto, T., Chytil, A., Minowa, O., expression of chimeric xVg1 and a truncated Activin type II Noda, T., and Miyazono, K. (2000). BMP type II receptor is

required for gastrulation and early development of mouse em- Gentry, L. E., Lioubin, M. N., Purchio, A. F., and Marquardt, H. bryos. Dev. Biol. 221, 249–258. (1988). Molecular events in the processing of recombinant type 1 Bourbon, H. M., Lapeyre, B., and Amalric, F. (1988). Structure of the pre-pro-transforming growth factor beta to the mature polypep- mouse nucleolin gene: The complete sequence reveals that each tide. Mol. Cell. Biol. 8, 4162–4168. RNA binding domain is encoded by two independent exons. J. Gray, A. M., and Mason, A. J. (1990). Requirement for activin A and Mol. Biol. 200, 627–638. transforming growth factor-␤1 pro-regions in homodimer assem- Burdine, R. D., and Schier, A. F. (2000). Conserved and divergent bly. Science 247, 1328–1330. mechanisms in left–right axis formation. Genes Dev. 14, 763– Griffith, D. L., Keck, P. C., Sampath, T. K., Rueger, D. C., and 776. Carlson, W. D. (1996). Three-dimensional structure of recombi- Campione, M., Steinbeisser, H., Schweickert, A., Deissler, K., van nant human osteogenic protein 1: Structural paradigm for the Bebber, F., Lowe, L. A., Nowotschin, S., Viebahn, C., Haffter, P., transforming growth factor beta superfamily. Proc. Natl. Acad. Kuehn, M. R., and Blum, M. (1999). The homeobox gene Pitx2: Sci. USA 93, 878–883. Mediator of asymmetric left–right signaling in vertebrate heart Gu, Z., Nomura, M., Simpson, B. B., Lei, H., Feijen, A., van den and gut looping. Development 126, 1225–1234. Eijnden-van Raaij, J., Donahoe, P. K., and Li, E. (1998). The type Capdevila, J., Vogan, K. J., Tabin, C. J., and Izpisua Belmonte, J. C. I activin receptor ActRIB is required for egg cylinder organization (2000). Mechanisms of left–right determination in vertebrates. and gastrulation in the mouse. Genes Dev. 12, 844–857. Cell 101, 9–21. Gu, Z., Reynolds, E. M., Song, J., Lei, H., Feijen, A., Yu, L., He, W., Chang, H., Zwijsen, A., Vogel, H., Huylebroeck, D., and Matzuk, MacLaughlin, D. T., van den Eijnden-van Raaij, J., Donahoe, M. M. (2000). Smad5 is essential for left–right asymmetry in P. K., and Li, E. (1999). The type I serine/threonine kinase mice. Dev. Biol. 219, 71–78. receptor ActRIA (ALK2) is required for gastrulation of the mouse Chen, X., Weisberg, E., Fridmacher, V., Watanabe, M., Naco, G., embryo. Development 126, 2551–2561. and Whitman, M. (1997). Smad4 and FAST-1 in the assembly of Heasman, J. (1997). Patterning the Xenopus blastula. Development activin-responsive factor. Nature 389, 85–89. 124, 4179–4191. Collignon, J., Varlet, I., and Robertson, E. J. (1996). Relationship Helde, K. A., and Grunwald, D. J. (1993). The DVR-1 (Vg1) tran- between asymmetric nodal expression and the direction of em- script of zebrafish is maternally supplied and distributed bryonic turning. Nature 381, 155–158. throughout the embryo. Dev. Biol. 159, 418–426. Conlon, F. L., Lyons, K. M., Takaesu, N., Barth, K. S., Kispert, A., Hemmati-Brivanlou, A., and Melton, D. A. (1992). A truncated Herrmann, B., and Robertson, E. J. (1994). A primary requirement activin receptor dominantly inhibits mesoderm induction and for nodal in the formation and maintenance of the primitive formation of axial structures in Xenopus embryos. Nature 359, streak in the mouse. Development 120, 1919–1928. 609–614. Conlon, F., and Beddington, R. (1995). Mouse gastrulation from a Henry, G. L., Brivanlou, I. H., Kessler, D. S., Hemmati-Brivanlou, frog’s prospective. Semin. Dev. Biol. 6, 249–256. A., and Melton, D. A. (1996). TGF-␤ signals and a prepattern in Constam, D. B., and Robertson, E. J. (1999). Regulation of bone Xenopus laevis endodermal development. Development 122, morphogenetic protein activity by pro domains and proprotein 1007–1015. convertases. J. Cell Biol. 144, 139–149. Henry, G. L., and Melton, D. A. (1998). Mixer, a homeobox gene Constam, D. B., and Robertson, E. J. (2000a). SPC4/PACE4 regu- required for endoderm development. Science 281, 91–96. lates a TGFbeta signaling network during axis formation. Genes Heyer, J., Escalante-Alcalde, D., Lia, M., Boettinger, E., Edelmann, Dev. 14, 1146–1155. W., Stewart, C. L., and Kucherlapati, R. (1999). Postgastrulation Constam, D. B., and Robertson, E. J. (2000b). Tissue-specific Smad2-deficient embryos show defects in embryo turning and requirements for the proprotein convertase furin/SPC1 during anterior morphogenesis. Proc. Natl. Acad. Sci. USA 96, 12595– embryonic turning and heart looping. Development 127, 245– 12600. 254. Hinck, A. P., Archer, S. J., Qian, S. W., Roberts, A. B., Sporn, M. B., Cui, Y., Tian, Q., and Christian, J. L. (1996). Synergistic effects of Weatherbee, J. A., Tsang, M. L., Lucas, R., Zhang, B. L., Wenker, Vg1 and Wnt signals in the specification of dorsal mesoderm and J., and Torchia, D. A. (1996). Transforming growth factor beta 1: endoderm. Dev. Biol. 180, 22–34. Three-dimensional structure in solution and comparison with Cui, Y., Jean, F., Thomas, G., and Christian, J. L. (1998). BMP-4 is the X-ray structure of transforming growth factor beta 2. Bio- proteolytically activated by furin and/or PC6 during vertebrate chemistry 35, 8517–8534. embryonic development. EMBO J. 17, 4735–4743. Hogan, B., Beddington, R., Constantini, F., and Lacy, E. (1994). Dale, L., Matthews, G., Tabe, L., and Colman, A. (1989). Develop- “Manipulating the Mouse Embryo: A Laboratory Manual.” Cold mental expression of the protein product of Vg1, a localized Spring Harbor Laboratory Press, Cold Spring Harbor, NY. maternal mRNA in the frog Xenopus laevis. EMBO J. 8, 1057– Hoodless, P. A., Tsukazaki, T., Nishimatsu, S., Attisano, L., Wrana, 1065. J. L., and Thomsen, G. H. (1999). Dominant-negative Smad2 Dale, L., Matthews, G., and Colman, A. (1993). Secretion and mutants inhibit activin/Vg1 signaling and disrupt axis formation mesoderm-inducing activity of the TGF-␤-related domain of in Xenopus. Dev. Biol. 207, 364–379. Xenopus Vg1. EMBO J. 12, 4471–4480. Hyatt, B. A., Lohr, J. L., and Yost, H. J. (1996). Initiation of Dohrmann, C. D., Kessler, D. S., and Melton, D. A. (1996). vertebrate left–right axis formation by maternal Vg1. Nature 384, Induction of axial mesoderm by zDVR-1, the zebrafish ortho- 62–65. logue of Xenopus Vg1. Dev. Biol. 175, 108–117. Hyatt, B. A., and Yost, H. J. (1998). The left–right coordinator: The Gaio, U., Schweickert, A., Fischer, A., Garratt, A. N., Muller, T., role of Vg1 in organizing left–right axis formation. Cell 93, Ozcelik, C., Lankes, W., Strehle, M., Britsch, S., Blum, M., and 37–46. Birchmeier, C. (1999). A role of the cryptic gene in the correct Ishimaru, Y., Yoshioka, H., Tao, H., Thisse, B., Thisse, C., Wright, establishment of the left–right axis. Curr. Biol. 9, 1339–1342. C. V. E., Hamada, H., Ohuchi, H., and Noji, S. (2000). Asymmet-

ric expression of antivin/lefty1 in the early chick embryo. Mech. Conserved left–right asymmetry of nodal expression and alter- Dev. 90, 115–118. ations in murine situs inversus. Nature 381, 158–161. Jones, C. M., Simon-Chazottes, D., Guenet, J. L., and Hogan, B. L. Lyons, K. M., Graycar, J. L., Lee, A., Hashmi, S., Lindquist, P. B., (1992). Isolation of Vgr-2, a novel member of the transforming Chen, E. Y., Hogan, B. L. M., and Derynck, R. (1989a). Vgr-1, a growth factor-beta-related gene family. Mol. Endocrinol. 6, 1961– mammalian gene related to Xenopus Vg-1, is a member of the 1968. transforming growth factor-␤ superfamily. Proc. Natl. Acad. Sci. Jones, C. M., Kuehn, M. R., Hogan, B. L., Smith, J. C., and Wright, USA 86, 4554–4558. C. V. (1995). Nodal-related signals induce axial mesoderm and Lyons, K. M., Pelton, R. M., and Hogan, B. L. M. (1989b). Patterns dorsalize mesoderm during gastrulation. Development 121, of expression of murine VgR-1 and BMP-2a RNA suggest that 3651–3662. transforming growth factor-␤-like genes ccordinately regulate Joseph, E. M., and Melton, D. A. (1998). Mutant Vg1 ligands disrupt aspects of embryonic development. Genes Dev. 3, 1657–1668. endoderm and mesoderm formation in Xenopus embryos. Devel- Marszalek, J. R., Ruiz-Lozano, P., Roberts, E., Chien, K. R., and opment 125, 2677–2685. Goldstein, L. S. (1999). Situs inversus and embryonic ciliary Kanzler, B., Foreman, R. K., Labosky, P. A., and Mallo, M. (2000). morphogenesis defects in mouse mutants lacking the KIF3A BMP signaling is essential for development of skeletogenic and subunit of kinesin-II. Proc. Natl. Acad. Sci. USA 96, 5043–5048. neurogenic cranial neural crest. Development 127, 1095–1104. Massague, J., Attisano, L., and Wrana, J. L. (1994). The TGF-␤ Kessler, D. S., and Melton, D. A. (1994). Vertebrate embryonic family and its composite receptors. Trends Cell Biol. 4, 172–178. induction: Mesodermal and neural patterning. Science 266, 596– Massague, J. (1998). TGF-beta signal transduction. Annu. Rev. 604. Biochem. 67, 753–791. Kessler, D. S., and Melton, D. A. (1995). Induction of dorsal Massague, J., and Chen, Y. G. (2000). Controlling TGF-beta signal- mesoderm by soluble, mature Vg1 protein. Development 121, ing. Genes Dev. 14, 627–644. 2155–2164. Matzuk, M. M., Kumar, T. R., and Bradley, A. (1995a). Different Kessler, D. S. (1999). Maternal signaling pathways and the regula- phenotypes for mice deficient in either activins or activin recep- tion of cell fate. In “Cell Lineage and Fate Determination” (S. A. tor type II. Nature 374, 356–360. Moody, Ed.), pp. 323–340. Academic Press, San Diego. Matzuk, M. M., Kumar, T. R., Vassalli, A., Bickenbach, J. R., Roop, Kitamura, K., Miura, H., Miyagawa-Tomita, S., Yanazawa, M., D. R., Jaenisch, R., and Bradley, A. (1995b). Functional analysis of Katoh-Fukui, Y., Suzuki, R., Ohuchi, H., Suehiro, A., Motegi, Y., activins during mammalian development. Nature 374, 354–356. Nakahara, Y., Kondo, S., and Yokoyama, M. (1999). Mouse Pitx2 McPherron, A. C., and Lee, S. J. (1993). GDF-3 and GDF-9: Two new deficiency leads to anomalies of the ventral body wall, heart, members of the transforming growth factor-beta superfamily extra- and periocular mesoderm and right pulmonary isomerism. containing a novel pattern of cysteines. J. Biol. Chem. 268, Development 126, 5749–5758. 3444–3449. Klein, S. L. (1987). The first cleavage furrow demarcates the Meno, C., Saijoh, Y., Fujii, H., Ikeda, M., Yokoyama, T., Yokoyama, dorsal–ventral axis in Xenopus embryos. Dev. Biol. 120, 299– M., Toyoda, Y., and Hamada, H. (1996). Left–right asymmetric 304. expression of the TGF beta-family member lefty in mouse Lawson, K. A., Dunn, N. R., Roelen, B. A., Zeinstra, L. M., Davis, embryos. Nature 381, 151–155. A. M., Wright, C. V., Korving, J. P., and Hogan, B. L. (1999). Bmp4 Meno, C., Shimono, A., Saijoh, Y., Yashiro, K., Mochida, K., is required for the generation of primordial germ cells in the Ohishi, S., Noji, S., Kondoh, H., and Hamada, H. (1998). lefty-1 is mouse embryo. Genes Dev. 13, 424–436. required for left-right determination as a regulator of lefty-2 and Lee, S. J. (1990). Identification of a novel member (GDF-1) of the nodal. Cell 94, 287–297. transforming growth factor-beta superfamily. Mol. Endocrinol. 4, Meno, C., Gritsman, K., Ohishi, S., Ohfuji, Y., Heckscher, E., 1034–1040. Mochida, K., Shimono, A., Kondoh, H., Talbot, W. S., Robertson, Lee, S. J. (1991). Expression of growth/differentiation factor 1 in the E. J., Schier, A. F., and Hamada, H. (1999). Mouse Lefty2 and nervous system: conservation of a bicistronic structure. Proc. zebrafish antivin are feedback inhibitors of nodal signaling Natl. Acad. Sci. USA 88, 4250–4254. during vertebrate gastrulation. Mol. Cell 4, 287–298. Levin, M., Johnson, R. L., Stern, C. D., Kuehn, M., and Tabin, C. Mishina, Y., Suzuki, A., Ueno, N., and Behringer, R. R. (1995). (1995). A molecular pathway determining left–right asymmetry Bmpr encodes a type I bone morphogenetic protein receptor that in chick embryogenesis. Cell 82, 803–814. is essential for gastrulation during mouse embryogenesis. Genes Lin, C. R., Kioussi, C., O’Connell, S., Briata, P., Szeto, D., Liu, F., Dev. 9, 3027–3037. Izpisua-Belmonte, J. C., and Rosenfeld, M. G. (1999). Pitx2 Mittl, P. R., Priestle, J. P., Cox, D. A., McMaster, G., Cerletti, N., regulates lung asymmetry, cardiac positioning and pituitary and and Grutter, M. G. (1996). The crystal structure of TGF-beta 3 tooth morphogenesis. Nature 401, 279–282. and comparison to TGF-beta 2: Implications for receptor binding. Logan, M., Pagan-Westphal, S. M., Smith, D. M., Paganessi, L., and Protein Sci. 5, 1261–1271. Tabin, C. J. (1998). The transcription factor Pitx2 mediates Nieuwkoop, P. D., and Faber, J. (1967). “Normal Table of Xenopus situs-specific morphogenesis in response to left–right asymmet- laevis (Daudin).” North-Holland, Amsterdam. ric signals. Cell 94, 307–317. Nieuwkoop, P. D. (1969). The formation of mesoderm in urodelean Lohr, J. L., Danos, M. C., and Yost, H. J. (1997). Left–right amphibians. I. Induction by the endoderm. Roux Arch. Entw- asymmetry of a nodal-related gene is regulated by dorsoanterior Mech. Org. 162, 341–373. midline structures during Xenopus development. Development Nishimatsu, S., Suzuki, A., Shoda, A., Murakami, K., and Ueno, N. 124, 1465–1472. (1992). Genes for bone morphogenetic proteins are differentially Lowe, L. A., Supp, D. M., Sampath, K., Yokoyama, T., Wright, transcribed in early amphibian embryos. Biochem. Biophys. Res. C. V., Potter, S. S., Overbeek, P., and Kuehn, M. R. (1996). Commun. 186, 1487–1495.

Nomura, M., and Li, E. (1998). Smad2 role in mesoderm formation, Schulte-Merker, S., Smith, J. C., and Dale, L. (1994). Effects of left–right patterning and craniofacial development. Nature 393, truncated activin and FGF receptors and of follistatin on the 786–790. inducing activities of BVg1 and activin: does activin play a role in Nonaka, S., Tanaka, Y., Okada, Y., Takeda, S., Harada, A., Kanai, mesoderm induction? EMBO J. 13, 3533–3541. Y., Kido, M., and Hirokawa, N. (1998). Randomization of left– Seleiro, E. A. P., Connolly, D. J., and Cooke, J. (1996). Early right asymmetry due to loss of nodal cilia generating leftward development and experimental axis determination by the flow of extraembryonic fluid in mice lacking KIF3B motor chicken Vg1 gene. Curr. Biol. 6, 1476–1486. protein. Cell 95, 829–837. Sha, X., Brunner, A. M., Purchio, A. F., and Gentry, L. E. (1989). Oh, S. P., and Li, E. (1997). The signaling pathway mediated by the Transforming growth factor beta 1: Importance of glycosylation type IIB activin receptor controls axial patterning and lateral and acidic proteases for processing and secretion. Mol. Endocri- asymmetry in the mouse. Genes Dev. 11, 1812–1826. nol. 3, 1090–1098. Pagan-Westphal, S. M., and Tabin, C. J. (1998). The transfer of Shah, S. B., Skromne, I., C. R., H., Kessler, D. S., Lee, K. J., Stern, left–right positional information during chick embryogenesis. C. D., and Dodd, J. (1997). Misexpression of Chick Vg1 in the Cell 93, 25–35. marginal zone induces primitive streak formation. Development Piccolo, S., Agius, E., Leyns, L., Bhattacharyya, S., Grunz, H., 124, 5127–5138. Bouwmeester, T., and De Robertis, E. M. (1999). The head Sirard, C., de la Pompa, J. L., Elia, A., Itie, A., Mirtsos, C., Cheung, inducer Cerberus is a multifunctional antagonist of Nodal, BMP A., Hahn, S., Wakeham, A., Schwartz, L., Kern, S. E., Rossant, J., and Wnt signals. Nature 397, 707–710. and Mak, T. W. (1998). The tumor suppressor gene Smad4/Dpc4 Piedra, M. E., Icardo, J. M., Albajar, M., Rodriguez-Rey, J. C., and is required for gastrulation and later for anterior development of Ros, M. A. (1998). Pitx2 participates in the late phase of the the mouse embryo. Genes Dev. 12, 107–119. pathway controlling left–right asymmetry. Cell 94, 319–324. Sive, H. L., Grainger, R. M., and Harland, R. M. (2000). “Early Rankin, C. T., Bunton, T., Lawler, A. M., and Lee, S. J. (2000). Development of Xenopus laevis: A Laboratory Manual.” Cold Regulation of left–right patterning in mice by growth/ Spring Harbor Laboraty Press, Cold Spring Harbor, NY. differentiation factor-1. Nature Genet. 24, 262–265. Smith, J. C., Price, B. M. J., Van Nimmen, K., and Huylebroeck, D. Rebagliati, M. R., Weeks, D. L., Harvey, R. P., and Melton, D. A. (1990). Identification of a potent Xenopus mesoderm-inducing (1985). Identification and cloning of localized maternal mRNAs factor as a homolog of activin A. Nature 345, 729–731. from Xenopus eggs. Cell 42, 769–777. Sokol, S., Christian, J. L., Moon, R. T., and Melton, D. A. (1991). Rebagliati, M. R., Toyama, R., Fricke, C., Haffter, P., and Dawid, Injected wnt RNA induces a complete body axis in Xenopus I. B. (1998). Zebrafish nodal-related genes are implicated in axial embryos. Cell 67, 741–752. patterning and establishing left–right asymmetry. Dev. Biol. 199, Song, J., Oh, S. P., Schrewe, H., Nomura, M., Lei, H., Okano, M., 261–272. Gridley, T., and Li, E. (1999). The type II activin receptors are Rodriguez Esteban, C., Capdevila, J., Economides, A. N., Pascual, J., essential for egg cylinder growth, gastrulation, and rostral head Ortiz, A., and Izpisua Belmonte, J. C. (1999). The novel Cer-like development in mice. Dev. Biol. 213, 157–169. protein Caronte mediates the establishment of embryonic left– Supp, D. M., Witte, D. P., Potter, S. S., and Brueckner, M. (1997). right asymmetry. Nature 401, 243–251. Mutation of an axonemal dynein affects left–right asymmetry in Roebroek, A. J., Umans, L., Pauli, I. G., Robertson, E. J., van inversus viscerum mice. Nature 389, 963–966. Leuven, F., Van de Ven, W. J., and Constam, D. B. (1998). Failure Supp, D. M., Brueckner, M., Kuehn, M. R., Witte, D. P., Lowe, L. A., of ventral closure and axial rotation in embryos lacking the McGrath, J., Corrales, J., and Potter, S. S. (1999). Targeted proprotein convertase Furin. Development 125, 4863–4876. deletion of the ATP binding domain of left–right dynein confirms Rupp, R. A., Snider, L., and Weintraub, H. (1994). Xenopus embryos its role in specifying development of left–right asymmetries. regulate the nuclear localization of XMyoD. Genes Dev. 8, Development 126, 5495–5504. 1311–1323. Takeda, S., Yonekawa, Y., Tanaka, Y., Okada, Y., Nonaka, S., and Ryan, A. K., Blumberg, B., Rodriguez-Esteban, C., Yonei-Tamura, Hirokawa, N. (1999). Left–right asymmetry and kinesin super- S., Tamura, K., Tsukui, T., de la Pena, J., Sabbagh, W., Green- family protein KIF3A: New insights in determination of lateral- wald, J., Choe, S., Norris, D. P., Robertson, E. J., Evans, R. M., ity and mesoderm induction by kif3AϪ/Ϫ mice analysis. J. Cell Rosenfeld, M. G., and Izpisua Belmonte, J. C. (1998). Pitx2 Biol. 145, 825–836. determines left–right asymmetry of internal organs in verte- Tannahill, D., and Melton, D. A. (1989). Localized synthesis of the brates. Nature 394, 545–551. Vg1 protein during early Xenopus development. Development Sampath, K., Cheng, A. M., Frisch, A., and Wright, C. V. (1997). 106, 775–785. Functional differences among Xenopus nodal-related genes in Thomsen, G., Woolf, T., Whitman, M., Sokol, S., Vaughan, J., Vale, left-right axis determination. Development 124, 3293–3302. W., and Melton, D. A. (1990). Activins are expressed early in Sasai, Y., Lu, B., Piccolo, S., and De Robertis, E. M. (1996). Xenopus embryogenesis and can induce axial mesoderm and Endoderm induction by the organizer-secreted factors chordin anterior structures. Cell 63, 485–493. and noggin in Xenopus animal caps. EMBO J. 15, 4547–4555. Thomsen, G. H., and Melton, D. A. (1993). Processed Vg1 protein is Scheufler, C., Sebald, W., and Hulsmeyer, M. (1999). Crystal an axial mesoderm inducer in Xenopus. Cell 74, 433–441. structure of human bone morphogenetic protein-2 at 2.7 A Toyoizumi, R., Mogi, K., and Takeuchi, S. (2000). More than 95% resolution. J. Mol. Biol. 287, 103–115. reversal of left–right axis induced by right-sided hypodermic Schier, A. F., and Shen, M. M. (2000). Nodal signaling in vertebrate microinjection of activin into Xenopus neurula embryos. Dev. development. Nature 403, 385–389. Biol. 221, 321–336. Schlunegger, M. P., and Grutter, M. G. (1992). An unusual feature Vize, P. D., and Thomsen, G. H. (1994). Vg1 and regional specifi- revealed by the crystal structure at 2.2 A resolution of human cation in vertebrates: a new role for an old molecule. Trends transforming growth factor-beta 2. Nature 358, 430–434. Genet. 10, 371–376.

Wagner, M. K., and Yost, H. J. (2000). Left–right development: The SMAD4/DPC4 is essential for epiblast proliferation and meso- roles of nodal cilia. Curr. Biol. 10, R149–R151. derm induction in mice. Proc. Natl. Acad. Sci. USA 95, 3667– Waldrip, W. R., Bikoff, E. K., Hoodless, P. A., Wrana, J. L., and 3672. Robertson, E. J. (1998). Smad2 signaling in extraembryonic tis- Yao, J., and Kessler, D. S. (1999). Mesoderm induction in Xenopus: sues determines anterior–posterior polarity of the early mouse Oocyte expression system and animal cap assay. In “Methods in embryo. Cell 92, 797–808. Molecular Biology, Vol. 37, Developmental Biology Protocols, Wall, N. A., and Hogan, B. L. (1994). TGF-beta related genes in Vol. III” (R. S. Tuan and C. W. Lo, Eds.), pp. 169–178. Humana development. Curr. Opin. Genet. Dev. 4, 517–522. Press, Totowa, NJ. Weeks, D. L., and Melton, D. A. (1987). A maternal mRNA Yokouchi, Y., Vogan, K. J., Pearse, R. V., 2nd, and Tabin, C. J. localized to the vegetal hemisphere in Xenopus eggs codes for a (1999). Antagonistic signaling by Caronte, a novel Cerberus- growth factor related to TGF-␤. Cell 51, 861–867. related gene, establishes left–right asymmetric gene expression. Weinstein, M., Yang, X., Li, C., Xu, X., Gotay, J., and Deng, C. X. Cell 98, 573–583. (1998). Failure of egg cylinder elongation and mesoderm induc- Yoshioka, H., Meno, C., Koshiba, K., Sugihara, M., Itoh, H., Ishimaru, Y., Inoue, T., Ohuchi, H., Semina, E. V., Murray, J. C., tion in mouse embryos lacking the tumor suppressor smad2. Hamada, H., and Noji, S. (1998). Pitx2, a bicoid-type homeobox Proc. Natl. Acad. Sci. USA 95, 9378–9383. gene, is involved in a lefty-signaling pathway in determination of Whitman, M. (1998). Smads and early developmental signaling by left–right asymmetry. Cell 94, 299–305. the TGFbeta superfamily. Genes Dev. 12, 2445–2462. Zhang, H., and Bradley, A. (1996). Mice deﬁcient for BMP2 are Wilson, P. A., and Melton, D. A. (1994). Mesodermal patterning by nonviable and have defects in amnion/chorion and cardiac de- an inducer gradient depends on secondary cell–cell communica- velopment. Development 122, 2977–2986. tion. Curr. Biol. 4, 676–686. Zhang, J., Houston, D. W., King, M. L., Payne, C., Wylie, C., and Wilson, P. A., and Hemmati-Brivanlou, A. (1995). Induction of Heasman, J. (1998). The role of maternal VegT in establishing the epidermis and inhibition of neural fate by Bmp-4. Nature 376, primary germ layers in Xenopus embryos. Cell 94, 515–524. 331–333. Winnier, G., Blessing, M., Labosky, P. A., and Hogan, B. L. (1995). Received for publication May 31, 2000 Bone morphogenetic protein-4 is required for mesoderm forma- Revised August 16, 2000 tion and patterning in the mouse. Genes Dev. 9, 2105–2116. Accepted August 30, 2000 Yang, X., Li, C., Xu, X., and Deng, C. (1998). The tumor suppressor Published online October 13, 2000