Secreted Wnt Antagonists

Shortly after the identification of the Frizzled (Fz) proteins as receptors of the Wnt signaling pathway, members of the secreted Frizzled related protein (SFRPs) family were identified in all vertebrate model organisms. These proteins are composed of a cysteine-rich domain (CRD), homologous to the extracellular portion of Fz, in their N-terminus, and a Netrin module domain, homologous to Netrin metalloproteinases inhibitors, in their C-terminus (Leyns et al., 1997; Lin et al., 1997; Lopez-Rios et al., 2008; Wang et al., 1997a). SFRPs are secreted and thought to bind and sequester Wnt ligands, thus nonautonomously inhibiting their signal transduction (Figure 9.2B) (Leyns et al., 1997; Wang et al., 1997a). Different SFRPs are expressed in different tissues during various developmental stages. During gastrula and neurula stages, SFRP3 (formerly known as Frzb) is expressed in the prechordal plate anterior mesoderm and anterior neural plate in zebrafish, frogs, and mice (Esteve et al., 2004; Leyns et al., 1997; Tendeng and Houart, 2006; Wang et al., 1997a). SFRP1 is also expressed in these same regions in medaka, zebrafish (SFRP1a), and chick embryos (Esteve et al., 2004; Esteve et al., 2000; Tendeng and Houart, 2006) (Figure 9.1A). This expression pattern is consistent with the role of SFRPs as anterior determinants. Indeed, overexpression of SFRP3 in Xenopus results in anteriorized embryos lacking trunk and tail structures (Leyns et al., 1997; Wang et al., 1997a). Moreover, loss of SFRP1 in medaka embryos caused caudalized embryos lacking head structures, indicative of excess Wnt signaling (Lopez-Rios et al., 2008). SFRP1 and SFRP3 were shown to antagonize Wnt1- and Wnt8-dependent caudalization as judged by both the overall morphology of the embryos and neural A–P marker expression levels, in medaka and frogs, respectively (Leyns et al., 1997; Lopez-Rios et al., 2008; Wang et al., 1997a; Wang et al., 1997b). Revealing the complexity of SFRP regulation of the canonical Wnt pathway, Xenopus SFRP3 does not antagonize Wnt3a, Wnt5a, or Wnt11 activity (Wang et al., 1997b). This complex regulation is further demonstrated by the observation that while Wnt5a activity is not inhibited by Xenopus SFRP3, SFRP3 does bind Wnt5a protein in CoIP assays (Lin et al., 1997). In addition, recent evidence supports a novel role for Xenopus SFRP3 and Crescent, another SFRP protein modulating the noncanonical Wnt pathway, in the expansion and propagation of Wnt8 and Wnt11 ligands, thus positively modulating their effective distance of signaling activity (Mii and Taira, 2009).

FIGURE 9.2

Mechanisms of Wnt/β-catenin pathway inhibition by secreted antagonists. Modified from (Kawano and Kypta, 2003). (A) An active pathway. (B) Inhibited pathway due to the sequestering of Wnt ligands by SFRP proteins and their possible binding to (more...)

Dickkopf proteins (Dkk1–4) constitute another family of secreted Wnt antagonists. While Dkk3 is the only family member that does not inhibit the Wnt pathway, Dkk1 is a robust inhibitor of the pathway, and Dkk2 and Dkk4, albeit less potent, are also Wnt inhibitors (Mao and Niehrs, 2003). Like the SFRPs, Dkk1 protein is expressed in the prechordal plate analogous regions of zebrafish, frogs, chick, and mice embryos during gastrula and neurula stages, but without expression in the neural plate (Figure 9.1A) (Glinka et al., 1998; Hashimoto et al., 2000; Kazanskaya et al., 2000; Marvin et al., 2001). Overexpression of Dkk1 protein in frogs and zebrafish embryos results in an enlarged head and malformed trunk and tail regions (see Figure 2.1A) (Glinka et al., 1998; Hashimoto et al., 2000; Kazanskaya et al., 2000). Concomitantly, expression of anterior neural markers is robustly expanded posteriorly, while posterior neural marker expression is inhibited (see example in Xenopus in Figure 3.2A, B) (Hashimoto et al., 2000; Kazanskaya et al., 2000). Moreover, Dkk1 is required for anterior neural development. Injection of anti-Dkk1 inhibitory antibodies to Xenopus embryos disrupts Dkk1 activity and yields a caudalized phenotype, similar to that of canonical Wnt overactivation (Glinka et al., 1998; Kazanskaya et al., 2000). Dkk1 hypomorphic mice exhibit a range of microcephaly, and Dkk1 null mice lack anterior head structures whatsoever (see Figure 2.2C) (MacDonald et al., 2004). Dkk1 was shown to antagonize Wnt3a and Wnt8 caudalization in Xenopus embryos, and Wnt1-induced reporter activity in 293T cells (Glinka et al., 1998; Kazanskaya et al., 2000; Mao et al., 2001). However, the mechanism of this inhibition differs from that of the SFRPs family. Dkk1 does not bind either Wnt ligands or Fz receptors. Instead, it binds Kremen1/2 receptors (Krm1/2). The Dkk1–Krm duplex binds the Fz co-receptor Lrp5/6 and induces its rapid endocytosis (Bafico et al., 2001; Davidson et al., 2002; Mao et al., 2002; Mao et al., 2001; Semenov et al., 2001). Dkk1, thus, prevents the proper formation of the Wnt-Fz-Lrp5/6 complex, preventing Wnt signal transduction (Figure 10.1C). Despite the obvious antagonistic role of Dkk1 as a modulator of the Wnt pathway, the role Dkk2 plays in this context is more complex. Dkk proteins have two CRDs. Interestingly, while the C-terminus CRD of Dkk1,2,4 proteins mediates the inhibitory binding to Lrp5/6, surprisingly, the N-terminus CRD of Dkk2 binds Lrp5/6 to activate Wnt/β-catenin signaling in both Xenopus embryos and NIH3T3 cells (Brott and Sokol, 2002; Li et al., 2002). Krm2 was suggested as a selector between these two opposite activities of Dkk2 because in the absence of Krm2 protein, Dkk2 activates the Wnt/β-catenin pathway in a Wnt ligand-independent manner, while in the presence of Krm2, it binds the Dkk2 N-terminus CRD driving its antagonistic activity (Mao and Niehrs, 2003).

Transcription Factor Wnt Antagonists

A member of the Six family of proteins, Six3, is expressed in the presumptive forebrain region of the anterior neural plate from late gastrula stages onward, in zebrafish, medaka, frog, chick, and mouse embryos (Figure 9.1B) (Bovolenta et al., 1998; Chapman et al., 2002; Kobayashi et al., 1998; Lavado et al., 2008; Loosli et al., 1998; Zhou et al., 2000). Overexpression of Six3 in zebrafish embryos resulted in a dramatic enlargement of the forebrain, mainly the most rostral telencephalon region (Kobayashi et al., 1998). Moreover, Six3 null mice embryos failed to express several forebrain markers, and concomitantly, mid-brain and hindbrain markers were expanded anteriorly in these Six3^–/– embryos (Lagutin et al., 2003; Lavado et al., 2008). Six3 was shown to repress Wnt1 expression, thus preventing its caudalizing activity in the forebrain. In E10.0 Six3^–/– mice embryos, expression of Wnt1 was expanded rostrally, and Six3 overexpression in 10hpf zebrafish embryos repressed Wnt1 expression and rescued the headless mutant phenotype (Tcf3 loss-of-function that results in an overactivated Wnt pathway) (Lagutin et al., 2003). In addition, both EMSA and ChIP assays confirmed the direct binding of Six3 specifically to elements I–III in the Wnt1 promoter in the forebrain of E8.5 mice (Lagutin et al., 2003). In the chick, the Wnt/β-catenin pathway was necessary and sufficient to repress Six3 expression (Lagutin et al., 2003), revealing mutual repression interactions between Six3 and the Wnt/β-catenin pathway. However, Six^–/–; Wnt1^–/– double null mice embryos still failed to express forebrain markers, showing that Six3 possesses an additional activity required for forebrain development that is independent of Wnt1 repression (Lavado et al., 2008). Furthermore, zebrafish Six3 was shown to act as a repressor, binding the Grg3 Groucho co-repressor protein, recruiting it to promoters to repress transcription (Kobayashi et al., 2001).

The Anf1 homeobox protein (called Hesx1 in mice) is expressed in the rostral extremity of the forebrain of sturgeon, zebrafish, newt, frog, chick, mouse, and human embryos from late gastrula to later organogenesis stages (Figure 9.1B) (Kazanskaya et al., 1997; Spieler et al., 2004; Thomas et al., 1995; Zaraisky et al., 1992). Anf family proteins exist only in vertebrates, and its emergence is thought to allow the development of the rostral telencephalon region that is an exclusive vertebrate feature (Ermakova et al., 2007). Xenopus Anf1 and mouse Hesx1 proteins were shown to be necessary for rostral telencephalon development; in their loss-of-function, rostral forebrain markers were not expressed and more caudal telencephalic and diencephalic markers were expanded anteriorly (Andoniadou et al., 2007; Ermakova et al., 2007). Moreover, overexpression of XAnf1 resulted in the posterior expansion of rostral telencephalon markers with a concomitant down-regulation of caudal telencephalon and diencephalon markers (Ermakova et al., 2007). A possible molecular explanation for these phenotypes may be provided by the observation that Hesx1^–/– mouse embryos exhibited a marked anterior expansion of Wnt1 and Wnt3a gene expression at the 1- to 5-somite and 8- to 10-somite stages, respectively. This was followed by dramatic anterior ectopic expression of the Wnt/β-catenin pathway targets, Axin2 and Sp5 (Andoniadou et al., 2007), suggesting that Hesx1 acts to repress Wnt ligand expression, which prevents their consequent caudalizing activity. Experiments in chimeric mouse embryos composed of WT and Hesx1^–/– cells showed that Hesx1 acts cell-autonomously to specify rostral forebrain development (Martinez-Barbera et al., 2000). Furthermore, while conditional Hesx1 expression in the forebrain rescued the Hesx1^–/– phenotypes, ubiquitous Hesx1 overexpression did not alter hindbrain and spinal cord formation (Andoniadou et al., 2007). This implies that Hesx1-mediated Wnt inhibition requires additional forebrain-specific factors. During later development of the forebrain-derived pituitary, Hesx1 was shown to genetically interact with Six3, which is coexpressed in the Hesx1 expression domain (Gaston-Massuet et al., 2008), suggesting that the different anti-Wnt anterior determinants may co-operate to ensure a Wnt-free zone in the anterior neural plate. Interestingly, Anf protein provides an example for the importance of Wnt/β-catenin activity modulation for the fine-tuning patterning of anterior forebrain regions, in addition to its role in the initial crude regionalization of forebrain versus hindbrain and spinal cord regions.

Another Wnt-antagonistic transcription factor, XSalF, was identified in Xenopus embryos (Onai et al., 2004). This Drosofila Spalt gene homologue is expressed in the anterior forebrain region during late gastrula and neurula stages (Figure 9.1B). XSalF was both necessary and sufficient for forebrain development. XSalF overexpression resulted in a posterior expansion of forebrain markers, accompanied by down-regulation of mid- and hindbrain marker expression. Accordingly, XSalF loss-of-function, by either expression of a truncated dominant-negative protein or MO knockdown, prevented expression of forebrain markers, while anteriorly expanding mid- and hindbrain marker expression (Onai et al., 2004). Underlying these phenotypes, XSalF was shown to be necessary and sufficient for the forebrain expression of the antagonistic components of the Wnt/β-catenin pathway, GSK3β and Tcf3 (Onai et al., 2004). Epistasis experiments in both neuralized AC explants and in vivo showed that these negative effectors act downstream of XSalF, as its overexpression could not rescue their inhibition, while in the reciprocal scenario, overexpression of either GSK3β or Tcf3 did rescue XSalF inhibition (Onai et al., 2004). Therefore, unlike Six3 and Hesx1 that prevent cells from expressing the Wnt ligands, XSalF reduces the responsiveness of presumptive forebrain cells to Wnt caudalizing signals.