Cripto promotes A–P axis specification independently of its stimulatory effect on Nodal autoinduction

Cripto, a glycophosphatidylinositol (GPI)-linked membrane protein, is the founding member of a family of vertebrate signaling molecules, the EGF–Cripto-FRL1-Cryptic (CFC) family, which includes human, mouse, and chick Cripto (Ciccodicola et al., 1989; Dono et al., 1993; Colas and Schoenwolf, 2000), human and mouse Cryptic (Shen et al., 1997), Xenopus laevis FRL-1/XCR1, XCR2, and XCR3 (Kinoshita et al., 1995; Dorey and Hill, 2006; Onuma et al., 2006), and zebrafish one-eyed pinhead (oep; Zhang et al., 1998). During development, members of the EGF-CFC family are required for mesoderm and endoderm formation and patterning of the anterior–posterior (A–P) and left–right axes (Shen and Schier, 2000). Genetic studies and cell-based assays have shown that the EGF-CFC proteins stimulate signaling by the TGF-β–related Nodal (Shen and Schier, 2000). Moreover, receptor reconstitution experiments and coimmunoprecipitation assays suggest that Cripto interacts with Nodal and the activin type IB receptor (activin-like kinase [ALK] 4), thereby activating a complex with the activin type IIB serine/threonine kinase (ActRIIB) receptor (Reissmann et al., 2001; Yeo and Whitman, 2001; Bianco et al., 2002; Sakuma et al., 2002; Yan et al., 2002). Upon receptor activation, the intracellular kinase domain of the type I receptor phosphorylates Smad2 and/or Smad3, which form a hexameric complex with the common Smad4 and translocate into the nucleus to regulate gene expression in conjunction with other transcription factors such as FoxH1 (Massague and Chen, 2000; Adkins et al., 2003; Gray et al., 2003; Harrison et al., 2005). Similarly, Cripto can sensitize a complex of ActRIIB and ALK7 to Nodal (Reissmann et al., 2001), and it also interacts with a subset of related ligands such as GDF1 and 3 (Cheng et al., 2003; Chen et al., 2006). Furthermore, Cripto has been found to bind specific Nodal antagonists, such as the transmembrane protein tomoregulin-1 (TMEFF-1; Harms and Chang, 2003) or the TGF-β–related Lefty proteins (Chen and Shen, 2004). However, the structural determinants that mediate these diverse protein–protein interactions and their relative influence on specific signaling pathways in the embryo are poorly defined.

Consistent with an important role for cripto in Nodal signaling, loss-of-function analysis in the mouse has shown that cripto is essential for both primitive streak formation and conversion of the initial proximal-distal patterning into the A–P axis during gastrulation (Ding et al., 1998; Liguori et al., 2003). However, cripto-null embryos express posterior markers, such as Brachyury and Fgf8, and form anterior neural structures and extraembryonic mesoderm, whereas Nodal mutants do not (Brennan et al., 2001). Thus, Nodal promotes anterior and posterior fates through both Cripto-dependent and -independent pathways.

Cripto has also been implicated in stimulating the progression of a broad spectrum of tumors (Salomon et al., 1999). Expression of cripto is increased severalfold in human colon, gastric, pancreatic, and lung carcinomas and in a variety of different types of mouse and human breast carcinomas (Ciardiello et al., 1991; Baldassarre et al., 1997). Although a specific receptor for Cripto has not yet been identified in mammary gland or cancer cells, mouse and human Cripto can activate a ras–raf–MAP kinase signaling pathway. This response may depend on the ability of Cripto to transactivate erbB-4 and/or FGF receptor 1 or to specifically bind to a membrane-associated heparan sulfate proteoglycan, glypican 1, leading to the activation of a Src-like tyrosine kinase (Bianco et al., 2003). However, without reagents that prevent endogenous Cripto from activating canonical ALK signaling, it has remained difficult to directly assess the physiological role of ALK-independent pathways.

Several structural determinants have been identified in the EGF and the CFC domains that regulate Cripto activity in cell transfection and X. laevis injection assays. Specifically, the CFC domain is essential for ALK4 interaction (Yeo and Whitman, 2001; Adkins et al., 2003), whereas threonine 72 in the EGF domain is O-fucosylated (Schiffer et al., 2001) and, apparently, promotes Nodal binding (Yeo and Whitman, 2001; Yan et al., 2002). It is worth noting that recent data indicate that the threonine residue that carries fucose, but not fucose per se, is required for Cripto to facilitate Nodal signaling (Shi et al., 2007). Furthermore, rescue experiments in cripto^−/− mouse embryonic stem cells and in oep mutant zebrafish established that recombinant Cripto protein also relies on the conserved amino acid F78 (Minchiotti et al., 2001; Parisi et al., 2003). However, whether F78 is essential for all Cripto activities or whether it specifically promotes Nodal signaling has remained unclear.

In this paper, we provide direct evidence that Cripto^F78A is unable to activate detectable amounts of Smad2 in embryonic stem (ES) cell–derived embryoid bodies (EBs) and that it fails to stimulate canonical Nodal–ALK4–Smad2 signaling in cell-based luciferase reporter assays. Further analysis of cripto^F78A/F78A mutant embryos confirms that residue F78 is essential to potentiate autoregulatory feedback signaling mediated by the Nodal–ALK4–Smad–FoxH1 pathway. We show that, unlike cripto-null mutants, cripto^F78A/F78A embryos clearly establish an A–P axis and initiate germ layer formation and gastrulation movements. A subset of known Nodal effector genes that are down-regulated in cripto-null mutants are significantly induced in cripto^F78A/F78A embryos. Collectively, these results suggest that Cripto can promote axis formation and gastrulation movements independently of its known stimulatory effect on the canonical Nodal–ALK4–Smad2 pathway.

To unravel the complex network of molecular interactions of Cripto with its target proteins in vivo, the amino acid residue F78, which is located in the EGF-like domain, was substituted by alanine (F78A) using Cre/loxP-mediated recombination (Fig. 1, A–C). The resulting heterozygous cripto^+/F78A mice appeared phenotypically normal and were fertile; however, homozygosity for the cripto^F78A-targeted allele resulted in embryonic lethality. We first verified the expression of the mutated allele in vivo by whole-mount immunohistochemistry analysis. Cripto protein was consistently detected in homozygous cripto^F78A/F78A embryos, although its expression remained confined to the proximal epiblast (Fig. 1 D). Although this result indicates that the alanine substitution does not abolish the synthesis or stability of Cripto protein, expansion of the expression domain to the distal epiblast is clearly compromised. Upon dissection, cripto^F78A/F78A mutants were recovered at the expected mendelian ratio until 10.5 d past confluence (dpc) and later were resorbed (Table I). However, at 7.5 dpc they already displayed ectopic folds in the embryonic region (Fig. 2, A and A′). At 8.5 dpc, mutant embryos failed to turn and the neural folds were enlarged (Fig. 2, B and B′), apparently at the expense of mesodermal structures, because somites and a beating heart were absent. These results show that residue F78 of Cripto is essential for postimplantation development.

Loss-of-function analysis has shown that cripto converts proximal-distal patterning into an A–P axis and promotes primitive streak formation (Ding et al., 1998; Liguori et al., 2003). To assess whether cripto^F78A/F78A embryos have defects in axis formation, we examined the expression of asymmetrically expressed marker genes such as Brachyury and Otx2 at 7.5 dpc. In normal embryos, Brachyury marks the primitive streak, whereas expression of the anterior neural marker Otx2 by this stage is restricted to the opposite pole (Fig. 3 A; Wilkinson et al., 1990; Simeone et al., 1993). By comparison, cripto-null mutants largely consist of anterior neuroectoderm (Ding et al., 1998; Liguori et al., 2003) and, therefore, ectopically express Otx2 throughout the distal embryonic region (Fig. 3 A″; Ding et al., 1998; Liguori et al., 2003), whereas the mesodermal marker Brachyury is only activated in a few cells along the embryonic–extraembryonic boundary (Fig. 3 A″; Ding et al., 1998). In contrast, in cripto^F78A/F78A mutant embryos, Brachyury expression was normally posteriorized and persisted until 8.5 dpc, indicating the presence of posterior mesoderm populations that are missing in cripto-null mutants (Fig. 3 A′; and Fig. S1, A, A′, and A″). In addition, Otx2 mRNA was consistently localized in the anterior region (Fig. 3 A′; and Fig. S1, A and A′), suggesting that A–P patterning is relatively normal. To monitor posterior neuroectoderm, we also analyzed the expression of Krox20, a marker of rhombomeres three and five, which is absent in cripto-null mutants (Ding et al., 1998). Krox20 mRNA was clearly detected in cripto^F78A/F78A embryos at 8.5 dpc (Fig. S1, B and B′). In addition, Mox1, a marker of paraxial mesoderm that fails to be induced in cripto-null mutants, was expressed in the posterior region of cripto^F78A/F78A embryos (Fig. S1, C and C′). These results demonstrate that cripto^F78A/F78A homozygotes establish an A–P axis and arrest development at a later stage compared with null mutants.

To characterize gastrulation defects in cripto^F78A/F78A mutant embryos, we next visualized derivatives of the anterior primitive streak, such as the node, a structure that expresses Nodal and Foxa2, and the axial mesoderm or anterior definitive endoderm, which are marked by Foxa2, Chordin, or Cer-1 mRNAs (Fig. 3, B–E; Monaghan et al., 1993; Conlon et al., 1994; Biben et al., 1998; Bachiller et al., 2000). In cripto-null mutants, expression of Foxa2 was absent and Nodal expression was confined to the proximal epiblast, confirming that node formation is inhibited (Fig. 3 B″). In cripto^F78A/F78A mutants, Nodal and Foxa2 expression were clearly detected in the distal tip of the rudimentary primitive streak, suggesting that anterior primitive streak derivatives are specified (Fig. 3, B′ and C′). Moreover, Foxa2 mRNA staining revealed that axial mesendoderm populations are also present more anteriorly (Fig. 3 C′, s1), whereas they are missing in cripto-null mutants (Fig. 3 C″). Similarly, Cer-1 was clearly induced in 5 out of 10 cripto^F78A/F78A mutants, even though the mRNA level was reduced and its expression domain extended to more posterior regions compared with wild-type controls (Fig. 3, D and D′). Likewise, Chordin was undetectable in cripto-null embryos but expressed in 3 out of 10 of the cripto^F78A/F78A mutants that were analyzed (Fig. 3, E and E′ [arrowhead]). Thus, compared with a cripto-null mutation, the F78A substitution has only relatively mild inhibitory effects on mesendoderm and primitive streak formation.

Several studies in mice, X. laevis, and zebrafish link Cripto to the Nodal pathway (Shen and Schier, 2000). Therefore, to assess the role of residue F78 of Cripto, we analyzed the expression pattern of Nodal and its target genes, Lefty1 and 2, in cripto^F78A/F78A and cripto-null mutants at 6.75 dpc. At this stage, Nodal is expressed throughout the primitive streak and posterior mesoderm in wild-type embryos (Fig. 4 A; Conlon et al., 1994; Collignon et al., 1996). In contrast, in both cripto^F78A/F78A and cripto-null mutants, Nodal expression was reduced and remained at the rim of the proximal epiblast (Fig. 4, A′ and A″). Next, to assess whether Nodal signaling was induced, we analyzed the expression of Lefty1 and 2. In wild-type embryos at 6.75 dpc, Lefty1 is expressed in the anterior visceral endoderm, whereas Lefty2 marks the nascent mesoderm generated from the primitive streak (Fig. 4 B; Meno et al., 1997). Expression of both Lefty1 and 2 was absent in cripto-null mutants (Fig. 4 B″). Interestingly, both genes were induced in cripto^F78A/F78A embryos, although below normal levels (Fig. 4 B′). To determine whether Nodal signaling is also maintained at later stages in cripto^F78A/F78A embryos, we analyzed the expression pattern of Lefty2, a direct Nodal target gene, and Fgf8 at 7.5 dpc. As expected, both genes were readily detectable in the primitive streak of wild-type embryos (Fig. 4 C) but not in cripto-null mutants (Fig. 4 C″). In contrast, Lefty2 mRNA was detected in a subset of cells in the posterior side of cripto^F78A/F78A embryos (Fig. 4 C′). Furthermore, Fgf8 was expressed in cripto^F78A/F78A mutants and its expression domain was even enlarged and extended into the extraembryonic region (Fig. 4 C′). Collectively, these data strongly suggest that the strength or duration of Nodal signaling in cripto^F78A/F78A embryos is perturbed compared with wild-type embryos, although it significantly exceeds that observed in cripto-null mutants.

Previous analysis of ES cell–derived EBs suggested that F78 is essential for Cripto to stimulate the in vitro differentiation of cardiomyocytes (Parisi et al., 2003). Similarly, substitution of F78 by alanine entirely blocks the ability of Cripto to rescue gastrulation of oep mutant zebrafish embryos (Minchiotti et al., 2001). Given these reports, it was surprising that substitution of F78 by alanine only partially inhibited Cripto activity in the mouse embryo. To determine whether Cripto^F78A can stimulate Nodal signaling in cell culture, we monitored its effect on CAGA-luc, a well characterized and sensitive luciferase reporter of ALK4–Smad3 signaling. Although transfection of wild-type cripto potently stimulated the activity of Nodal, Cripto^F78A was completely inactive in this assay (Fig. 5 A). Analogous results were obtained using the activin response element (ARE)–luc reporter construct in conjunction with wild-type Nodal or a more potent supercleaved and stabilized derivative (Nsc-g; Fig. S2; Yan et al., 2002; Le Good et al., 2005; Chen et al., 2006; Andersson et al., 2007). These results suggest that Cripto^F78A is unable to activate a Nodal–ALK4–Smad signaling complex.

Cripto can also potentiate growth/differentiation factor (GDF) 1 and 3 signaling (Andersson et al., 2007), raising the question of whether these activities rely on F78 in a manner similar to Nodal. Unlike wild-type Cripto, the Cripto^F78A mutant failed to potentiate GDF1 or 3 signaling in this assay (Fig. 5 A and not depicted).

Cripto^F78A might activate the Smad2 pathway only in an embryo-like cell context. To test this, 2-d-old Cripto^−/− ES cell–derived EBs were starved in low serum for 3 h and then stimulated with recombinant soluble Cripto or the F78A mutant protein. Consistent with published data (Watanabe et al., 2007), Cripto without a GPI anchor was poorly active but, when applied at elevated concentrations, it significantly increased the phosphorylation of Smad2 (Fig. 5 B). In contrast, soluble Cripto^F78A failed to detectably stimulate Smad2 phosphorylation (Fig. 5 B). However, mutation of F78 did not diminish the ability of Cripto to stimulate the MAPK signaling pathway (Fig. 5 C). Thus, we conclude that the F78A mutation selectively impairs Nodal–ALK4–Smad2,3 while leaving intact Smad-independent signals mediated by MAPK.

Understanding how Cripto stimulates Nodal-dependent cell movements in the visceral endoderm and epiblast is fundamental to our understanding of how the A–P body axis is established in mammalian embryos. In cripto^−/− embryos, distal visceral endoderm do not move, and the vast majority of cells in the epiblast adopt a neuroectodermal character because mesendoderm progenitors, which form the primitive streak, are either absent or remain confined to the proximal epiblast (Ding et al., 1998; Liguori et al., 2003). In this paper, we show that homozygous mutants carrying the novel cripto^F78A allele display less severe defects than cripto^−/− embryos. In particular, definitive endoderm and axial mesoderm populations marked by Cer-1 and Foxa2 transcripts are readily detectable, and neural progenitors expressing Otx2 mRNA consistently localize to the anterior region. In some instances, anterior-most midline cells also express the axial marker Chordin, which is consistent with their mesendodermal origin. Likewise, posterior cells expressing Fgf8 and Brachyury that are absent or immobilised in the proximal epiblast of cripto-null embryos (Ding et al., 1998) clearly ingress in the primitive streak of cripto^F78A/F78A mutants, even though this structure remains abnormally short and eventually fails to form a morphologically distinguishable node or notochord. Thus, in sharp contrast to cripto-null mutants, cripto^F78A/F78A embryos establish an A–P axis and initiate gastrulation, suggesting that this mutant allele encodes a functional hypomorph.

The phenotype of cripto^F78A/F78A embryos is reminiscent of patterning defects that arise when Nodal autoinduction is inhibited (Hoodless et al., 2001; Yamamoto et al., 2001; Norris et al., 2002). During normal development, Nodal expression is initiated in the proximal epiblast and, upon activation of an autoregulatory enhancer by FoxH1, spreads to the visceral endoderm and distal epiblast (Brennan et al., 2001; Norris et al., 2002). In this paper, we show that both cripto^F78A/F78A and cripto-null mutants fail to expand the Nodal expression domain, confirming that Smad-dependent autoinductive Nodal signaling is inhibited. However, interestingly, mutant Cripto^F78A protein was sufficient to induce or prolong the expression of several other Nodal target genes, including Lefty1, Lefty2, and Fgf8, which were completely silenced in cripto-null mutants at the stages examined.

These results substantiate our conclusion that cripto^F78A is a hypomorphic allele that is sufficient to mediate Alk4–Smad–FoxH1–indpendent Nodal signaling. They can also explain why primitive streak and posterior mesoderm formation are relatively mildly perturbed in cripto^F78A/F78A mutants, because previous analysis of FoxH1 mutants (Hoodless et al., 2001; Yamamoto et al., 2001) and hypomorphic alleles of Nodal (Lowe et al., 2001; Norris et al., 2002; Vincent et al., 2003; Ben-Haim et al., 2006) established that posterior mesoderm formation requires lower levels of Nodal signaling compared with axial midline structures.

Previous studies have shown that Cripto strictly depends on residue F78 to rescue mutant zebrafish embryos lacking the Nodal coreceptor oep. Using cell-based activity assays, we confirmed in this paper that Cripto^F78A protein fails to stimulate well-characterized Nodal luciferase reporter genes, which specifically rely on ALK4–Smad–FoxH1 signaling. Furthermore, Cripto^F78A is also unable to significantly activate Smad2 in ES cell–derived EBs, a model that more closely mimics a physiological environment. Coimmunoprecipitation experiments in transfected cells previously established that Cripto can directly bind both Nodal and ALK4 to potentiate Nodal signaling (Reissmann et al., 2001; Yeo and Whitman, 2001). However, a triple mutant of the EGF-like domain comprising the F78 residue completely abolished the ability of Cripto to stimulate the induction of a Nodal luciferase reporter in mammalian tissue culture cells (Yan et al., 2002). Moreover, chemical cross-linking experiments in 293T cells, followed by coimmunoprecipitation, suggest that this triple mutant fails to bind Nodal, whereas it interacts with the ALK4 receptor in a manner similar to that of the wild type (Yan et al., 2002). The present results are thus consistent with a model in which the F78 residue of Cripto is essential to assemble functional Cripto–ALK4 receptor complexes and thereby potentiate a Nodal autoregulatory feedback loop.

Luciferase reporter assays and coimmunoprecipitation experiments suggest that Cripto can also potentiate Nodal signaling through ALK7 (Reissmann et al., 2001). Therefore, it is formally possible that the loss of F78 selectively blocks the ability of Cripto to activate ALK4 without affecting Nodal signaling via the ALK7 receptor. However, it has previously been shown that ALK7 is dispensable and unable to compensate for the loss of ALK4 in the mouse embryo (Gu et al., 1998; Jornvall et al., 2004). Cripto can also stimulate the induction of CAGA-luc reporter by native forms of GDF1 and 3 (Andersson et al., 2007). In contrast, as observed with Nodal, GDF1 and 3 failed to signal through Cripto^F78A. However, our data do not rule out the possibility that F78 may retain the ability to influence Smad signaling by other ligands, including activins or TGF-β.

Cripto can also regulate several Smad-independent signals in various biological contexts (Adamson et al., 2002; Strizzi et al., 2005). Our results show that Cripto^F78A retains the ability to activate MAPK in embryo bodies, which may contribute to the residual activity of Cripto^F78A observed in vivo. In keeping with this idea, Cripto also stimulates MAPK–AKT phosphorylation in mouse mammary epithelial cells independently of Nodal and ALK4 (Bianco et al., 2002), possibly through the association with the membrane-associated heparan sulfate proteoglycan glypican 1 (Bianco et al., 2003), and overexpression of cripto in mammary epithelial cells promotes epithelial-mesenchymal transitions and cell motility through a MAPK-dependent pathway (Strizzi et al. 2005). In contrast, cripto and its putative homologue FRL-1 have recently also been shown to bind Wnt11 and stimulate the canonical Wnt–β-catenin–Lef-1 signaling pathway in X. laevis (Tao et al., 2005). Finally, it is worth noting that recent data highlight a novel role of Cripto as a TGF-β antagonist and propose that Cripto antagonism of TGF-β signaling may contribute to tumor initiation and progression (Gray et al., 2006).

Based on this consideration, we believe that genetic manipulations, which selectively inhibit a subset of signaling activities, will be crucial to unambiguously dissect the complex functions of Cripto in vivo. Accordingly, our data provide the first in vivo evidence that Cripto initiates gastrulation movements independently of its known stimulatory effect on the canonical Nodal–ALK4–Smad2 signaling and open the way to dissect the complex network of Cripto signaling in vivo. It will be important to determine, in future studies, whether Nodal–ALK4–Smad–independent Cripto activities in vivo are mediated by the Wnt pathway, MAPK, or possibly other signaling pathways.

The targeting vector was derived from pFlox vector (Chen et al., 1998) by excision of a SmaI–BamHI DNA fragment spanning the loxP site and by removing the BglI–SmaI DNA fragment spanning the hsv-tk gene (Fig. 1 A). A 3.5-kb 5′ homologous sequence spanning exons 1 and 2 was inserted upstream of the loxP site–flanked cassette encoding the neo^r gene. A 5.6-kb 3′ homologous sequence spanning exons 3 to 6 was inserted downstream of the neo^r gene (Fig. 1 A). The two overlapping PCR primers 5′F78A (5′-GCATCCTGGGGTCCGCCTGTGCCTGCCCTC-3′) and 3′F78A (5′-GGAGGGCAGGCACAGGCGGACCCCAGGATGC-3′) were used to introduce the F78A point mutation in the targeting vector (underlining in primers indicates the nucleotide sequence that was modified to insert the F78A mutation).

RI ES cells were transfected with the targeting vector and selected in G-418. DNA prepared from individual drug-resistant colonies was digested with EcoRV for Southern blot analysis using a 600-bp 5′ external probe (RH5 in Fig. 1 [A and B]), a 500-bp 3′ internal probe (BE6 in Fig. 1 [A and B]), and a neo^r probe. After identification of the targeted clones, the presence of the point mutation was verified by PCR amplification and sequence analysis. Selected ES cell lines were used to generate germline chimeric mice that were subsequently bred to C57BL/6 females (Charles River Laboratories). F1 cripto^F78Aneo heterozygotes were crossed with a pgk-Cre deletion strain.

Heterozygous mice for the cripto^F78A allele were maintained on a mixed genetic background C57BL/6 × Sv129. Heterozygous mice for the cripto-null allele were maintained on a mixed genetic background (C57BL/6 × Sv129 × Black Swiss) and also backcrossed to an inbred C57BL/6 strain. No phenotypic differences were observed between cripto-null embryos on different genetic backgrounds. Timed matings between heterozygotes were used to obtain both cripto^F78A/F78A and cripto-null homozygous mutant embryos. Embryos were genotyped by PCR at 7.5 dpc using extraembryonic tissues. At 6.75 dpc, DNA was extracted from whole embryos for genotyping.

Embryos were dissected in PBS and fixed in 4% paraformaldehyde in PBS at 4°C for 2–16 h, washed in PBT (0.1% Tween 20 in PBS), dehydrated through graded methanol, and stored in 100% methanol at −20°C. For immunohistochemistry, embryos were rehydrated in PBTx (0.25% Triton X-100 in PBS), bleached with 0.05% H₂O₂ overnight, blocked with PBTsb (10% normal goat serum and 1 mg/ml BSA in PBTx), and incubated overnight at 4°C with 2 μg/ml of affinity-purified α-Cripto antibodies. To remove the unbound antibody, embryos were extensively washed in PBTx (1 h, six times) and labeled with biotinylated secondary antibody overnight at 4°C. After six washes in PBTx, embryos were incubated with biotin–streptavidin complex (AB complex; Vector Laboratories), revealed by incubation for 30 min with 0.5 mg/ml of 3–3′ diaminobenzidine (Sigma-Aldrich), and developed by addition of 0.03% H₂O₂. Stained embryos were examined and photographed using a stereomicroscope (MZ12; Leica). All images were processed in Photoshop 5.0 (Adobe).

Whole mount immunohistochemistry and in situ hybridization was performed according to standard procedures (Liguori et al., 2003). Probes for the following genes were used in this study: Brachyury (Wilkinson et al., 1990), Cerberus-like (Belo et al., 1997), Chordin (Bachiller et al., 2000), Fgf8 (Crossley and Martin, 1995), Foxa2 (Monaghan et al., 1993), Krox20 (Wilkinson et al., 1989), Lefty1 and 2 (Meno et al., 1996; Branford and Yost, 2002), Mox1 (Candia et al., 1992), Nodal (Varlet et al., 1997), and Otx2 (Acampora et al., 1995). J.A. Belo (Istituto Gulbenkian de Ciencia, Oeiras, Portugal), K. Chien (Harvard University, Cambridge, MA), R. Di Lauro (Università Federico II, Napoli, Italy), S. Filosa (Institute of Genetics and Biophysics “A. Buzzati-Traverso,” Naples, Italy), H. Hamada (Osaka University, Osaka, Japan), A. Simeone (Ceinge, Naples, Italy), and M. Studer (Telethon Institute of Genetics and Medicine, Naples, Italy) provided plasmids.

Cripto activity assays were performed as previously described (Yan et al., 2002; Andersson et al., 2007) by transiently transfecting 293T cells with either the ARE-luc or the pCAGA-luc luciferase reporter constructs and expression vectors for Cripto, FoxH1, and wild type or a stabilized form of Nodal (Nsc-g; Le Good et al., 2005).

The ES cell lines RI and cripto^−/− DE7 were used throughout the study and differentiation assays were performed as previously described (Parisi et al., 2003). Western blot analysis was performed as previously described (Parisi et al., 2003). Anti–phospho-Smad2 (Ser465/467), total Smad2 (Cell Signaling Technology), and anti–phospho–extracellular signal–regulated kinase (ERK; Santa Cruz Biotechnology, Inc.) antibodies were used according to the manufacturer's instructions.

Fig. S1 contains additional information on the embryonic development of cripto^F78A/F78A mutants at the head-fold stage. Fig. S2 contains additional information on Cripto activity in the 293 cell ARE-luc reporter assay and shows that F78A mutant is inactive in this assay.

We thank Mrs. M. Terracciano for technical assistance and Raffaele Improta and Ivan Solombrino for animal care. The Biogem transgenic facility is acknowledged for technical assistance in generating targeted ES cells and chimeric mice. J.A. Belo, K. Chien, R. Di Lauro, S. Filosa, H. Hamada, Y. Saga, A. Simeone, and M. Studer are acknowledged for their kind gifts of plasmids. We thank Anna Aliperti for proofreading the manuscript.

This work was supported by grants from the Associazione Italiana Ricerca sul Cancro to M.G. Persico and G. Minchiotti and from the Fondo per gli Investimenti della Ricerca di Base to M.G. Persico. D. D'Andrea was supported by a fellowship from the Fondazione Italiana Ricerca sul Cancro.