One of the fundamental problems in developmental biology is to understand how the three germ layers—ectoderm, mesoderm, and endoderm—are established. The molecular mechanisms of ectoderm and mesoderm formation have been analyzed extensively and our understanding of their development is fairly advanced. In contrast, the molecular analysis of endoderm formation has only begun relatively recently, but already several Xenopus and zebrafish genes have been implicated in this process. In Xenopus, several related paired-class homeodomain proteins, the Mix and Bix proteins, have been proposed to have important roles in endoderm formation (Rosa et al. 1989; Henry and Melton 1998; Mead et al. 1998; Tada et al. 1998). The genes encoding these proteins are all expressed in the prospective endoderm during gastrulation; most are also expressed in marginal cells that will form mesoderm, with the exception of Mixer, which is expressed exclusively in the endoderm (Henry and Melton 1998). When overexpressed in animal caps, these genes exhibit different abilities to activate endodermal gene expression: Mixer and Bix2/Milk appear to promote endodermal gene expression strongly, and at the expense of mesodermal gene expression (Ecochard et al. 1998; Henry and Melton 1998); Bix1 overexpression at low and high levels induces mesodermal and endodermal gene expression, respectively (Tada et al. 1998); and Mix.1 is able to activate endodermal gene expression only when co-expressed with the dorso-vegetal homeobox gene siamois (Lemaire et al. 1998). Additionally, Bix4 is able to restore endodermal gene expression to VegT depleted embryos, which lack all endoderm and most mesoderm (Zhang et al. 1998b), indicating that Bix4 acts downstream of VegT in endoderm formation (Casey et al. 1999). Therefore, whereas the various Xenopus Mix and Bix genes can regulate endoderm development, their individual roles and degree of importance in this process remain unresolved. In contrast to the situation in Xenopus, only a single Mix-like gene has been identified so far in other vertebrates. The chick CMIX gene is expressed in the epiblast of the posterior marginal zone of early chick embryos, and starting at the onset of gastrulation, along the entire primitive streak except for its most posterior part (Peale et al. 1998; Stein et al. 1998). In mouse, a Mix-like gene, Mml, has recently been identified (Pearce and Evans 1999); it is expressed in the visceral endoderm at embryonic day 5.5 (E5.5), and starting at E6.5 throughout the entire primitive streak except for its most anterior part. In zebrafish, a Mix-like gene has also been described and provisionally named mixer, because of its homology with Xenopus Mixer and ability to induce endodermal gene expression in wild-type and mutant zebrafish embryos (Alexander et al. 1999; Alexander and Stainier 1999). The Xenopus genes Xsox17α, and Xsox17β (here referred to collectively as Xsox17), which encode proteins containing a high mobility group DNA-binding domain, are also implicated in endoderm development (Hudson et al. 1997). Expression of Xsox17 becomes restricted to the endoderm at the onset of gastrulation, and when overexpressed in animal caps, Xsox17 activates endodermal gene expression. Mixer induces Xsox17 expression but Xsox17 does not induce Mixer expression in animal caps, suggesting that Mixer acts upstream of Xsox17; however, Xsox17 expression appears 3 hr before that of Mixer during normal development (Henry and Melton 1998). Therefore, it may be that these genes initiate expression independently of each other, with Mixer then regulating the maintenance of Xsox17 expression (Henry and Melton 1998; Yasuo and Lemaire 1999). Additionally, coexpression of Mixer and a dominant interfering form of Xsox17 (Xsox17–enR) in animal caps blocks the induction of endodermal gene expression, whereas coexpression of Xsox17 and Mixer–enR does not, suggesting that Mixer functions through Xsox17 in endoderm formation (Henry and Melton 1998). In the absence of genetic analyses of the Mix, Bix, and Xsox17 genes, however, it remains difficult to describe precisely the relationships between these various genes. Several mutations have recently been reported to affect endoderm development in zebrafish (Schier et al. 1997; Feldman et al. 1998; Alexander et al. 1999; Reiter et al. 1999). The cyclops (cyc) and squint (sqt) genes encode Nodal-related proteins, and embryos mutant for both of these genes lack all endodermal and most mesodermal derivatives (Feldman et al. 1998). The same phenotype is seen in embryos that lack both maternal and zygotic one-eyed pinhead (oep), a gene encoding a member of the EGF–CFC protein family essential for Nodal signaling (Zhang et al. 1998a; Gritsman et al. 1999), whereas zygotic oep mutants lack endoderm as well as the prechordal plate and ventral neuroectoderm. Nodal-related genes have also been implicated in Xenopus endoderm development (Clements et al. 1999; Osada and Wright 1999; Yasuo and Lemaire 1999). casanova (cas) and faust (fau), two zebrafish mutants that exhibit bilateral hearts (cardia bifida), have also been shown to be essential for endoderm development. In cas mutants, the early endodermal expression of genes such as axial, sox17, and fkd2 does not initiate, but expression of the Mix-like gene is normal, suggesting that cas acts downstream of, or in parallel to, this Mix-like gene to promote endoderm formation (Alexander et al. 1999). The cas gene has not yet been isolated. The fau locus is essential for multiple aspects of heart and endoderm development; fau mutants exhibit reduced amounts of cardiac and endodermal tissue and abnormal morphogenesis of the heart, pharyngeal endoderm, and gut (Reiter et al. 1999). fau encodes Gata5, a zinc finger-containing transcription factor, and is expressed in the endodermal and some mesodermal progenitors before to the onset of gastrulation (Reiter et al. 1999). The bonnie and clyde (bon) mutation, like oep, cas, and fau, causes cardia bifida (Stainier et al. 1996). In this report, we show that in bon mutants the number of sox17-expressing endodermal precursors formed during gastrulation is significantly reduced, and that the gut tube is almost entirely absent at later stages. The endodermal precursors that do form in bon mutants, however, appear to differentiate normally—indicating that bon is not required at later steps of endoderm development. We also show that bon corresponds to the Mix-like gene that we isolated previously (Alexander et al. 1999). We further examined the relationship between bon and Nodal signaling and found that bon can induce a few cells to express sox17 expression in cyc;sqt double mutants, which normally completely lack sox17 expression. In contrast, bon cannot induce ectopic cyc and sqt expression in wild-type embryos, indicating that bon functions exclusively downstream of these nodal-related genes. Our results provide the first genetic evidence that Mix genes are critical for early endoderm development, show that bon is required only at a very early step of endoderm formation (i.e., the generation of sox17-expressing endodermal precursors), and suggest that additional factors act downstream of cyc and sqt to generate the full complement of endodermal precursors.