Genetic interaction between Wnt7a and Lrp6 during patterning of dorsal and posterior structures of the mouse limb



Wnt7a and the Wnt coreceptor Lrp6 are both required for development of posterior digits and dorsal structures of the limb. We report that Lrp6 null mice lack Lmx1b expression in the distal mesenchyme, as previously described for Wnt7a mutants. The loss of Lmx1b expression in Wnt7a−/−Lrp6+/− double mutants did not differ from that in Wnt7a−/− mice. These data suggest that Wnt7a acts through Lrp6 to regulate Lmx1b expression during dorsal specification. The loss of posterior skeletal elements in the Wnt7a−/−Lrp6+/− double mutant was much more severe than in Wnt7a−/− mice, suggesting that the Wnt7a−/− limb is protected by the action of other Lrp6 ligands. The data are consistent with the view that Wnt7a acts through Lrp6 and the canonical Wnt signaling pathway during dorsal and posterior limb development in the mouse. Developmental Dynamics 233:368–372, 2005. © 2005 Wiley-Liss, Inc.


Dorsal–ventral and anterior–posterior axes are established early in development of the vertebrate limb. The dorsal part of the limb, which will form hair and nails in the mouse, is characterized by expression of Wnt7a in the ectoderm and Lmx1b in the underlying mesenchyme. The ventral part of the limb, which will develop hairless footpads, is marked by expression of En1 in the ectoderm. The apical ectodermal ridge (AER) is positioned on the border between dorsal and ventral compartments and is necessary for the outgrowth of the limb. Fibroblast growth factors, mainly Fgf8, are responsible for the function of the AER. The zone of polarizing activity (ZPA) is a group of mesenchymal cells positioned under the posterior end of the AER. Shh expression in the ZPA is necessary for development of the posterior digits (reviewed by Niswander, 2003; Tickle, 2003).

Analysis of mutant phenotypes in the mouse and experimental manipulation of gene expression in chicken limbs have revealed complex interactions among genes in the Wnt signaling pathway (reviewed by Church and Francis-West, 2002). Loss-of-function mutations in mouse Wnt7a affect both dorsal–ventral and anterior–posterior limb development (Parr and McMahon, 1995; Parr et al., 1998). Decreased expression of Shh in the ZPA of Wnt7a null limbs is followed by loss of posterior digits and less penetrant defects in the posterior zeugopod element, the ulna. Loss of Lmx1b expression in the distal mesenchyme of Wnt7a null limbs presages loss of dorsal limb identity and results in formation of dorsal footpads. Null mutation of the Lmx1b gene itself causes dorsal defects similar to Wnt7a mutations (Chen at al., 1998).

Expression of Wnt7a is not only necessary for the distal expression of Lmx1b but is also sufficient to induce Lmx1b expression. Ectopic expression of Wnt7a in the ventral ectoderm of the chicken limb leads to ectopic expression of Lmx1b in the underlying mesenchyme (Kengaku et al., 1998). In the mouse, Wnt7a is ectopically expressed in the ventral ectoderm of limbs null for the transcription factor En1 (Loomis et al., 1996). This expression results in two distinct phenotypes: (1) ectopic expression of Lmx1b in the ventral mesenchyme, leading to development of double-dorsal limbs with round nails, and (2) ventral expansion of the AER leading to formation of fused and ectopic digits. Both features of the En1 null limb can be rescued by removal of Wnt7a. Double mutant En−/−Wnt7a−/− limbs are double-ventral and display loss of posterior digits exactly like Wnt7a−/− limbs (Cygan et al., 1997; Loomis et al., 1998).

There is a difference between mouse and chicken in the effect of ectopic expression of Wnt7a. In mouse, ectopic expression of Wnt7a induces Lmx1b and affects AER formation, as described above. In the chicken, ectopic expression of Wnt7a induces Lmx1b expression without affecting the AER (Kengaku et al., 1998). In contrast, ectopic expression of Wnt3a in chicken does not induce Lmx1b but does affect the AER (Kengaku et al., 1998). The difference between overexpression of Wnt7a and Wnt3a in chicken led to the suggestion that Wnt3a acts through the canonical (beta-catenin) pathway and Wnt7a acts through an unidentified noncanonical pathway (Kengaku et al., 1998). On the other hand, Wnt7a may act through the canonical pathway in the mouse, because mutation of Wnt7a or the Wnt co-receptor Lrp6 both result in loss of posterior structures and formation of double-ventral skeletal elements and tendons (Pinson et al., 2000).

Lrp6 plays a key role in canonical Wnt signaling as a Wnt co-receptor (reviewed by He et al., 2004). Formation of a complex between extracellular Wnt, the frizzled receptor, and the co-receptor Lrp6, leads to binding of axin by Lrp6 and stabilization of beta-catenin. Beta-catenin can subsequently enter the nucleus and regulate gene expression in complex with TCF/LEF transcription factors. Dkk1, a Wnt antagonist, binds Lrp6 and prevents formation of the Wnt/frizzled/Lrp6 complex and stabilization of beta-catenin (Bafico et al., 2001; Mao et al., 2001).

We recently demonstrated genetic interaction between Wnt7a and Dkk1 during limb development (Adamska et al., 2004). The ectopic expression of Wnt7a in En1 null limbs, in combination with reduced Dkk1 expression, resulted in a more severe phenotype than ectopic expression of Wnt7a alone. In addition, the loss of posterior digits in Wnt7a null limbs was rescued by reduction of Dkk1, presumably because reduced concentration of Dkk1 permitted other Wnts to bind Lrp6 and compensate for loss of Wnt7a (Adamska et al., 2004). Thus, Wnt7a interacts genetically with Dkk1, a component of the canonical Wnt pathway, in regulation of posterior limb development in the mouse. The connection between Wnt7a and the canonical pathway is strengthened by the recent demonstration of interaction between Wnt7a, Lrp6, and Dkk1 in a rat cell line (Caricasole et al., 2003). Here, we provide additional evidence that Wnt7a acts through the canonical pathway to regulate dorsal and posterior development of the mouse limb.


Lmx1b Expression Is Reduced in Lrp6 Null Mice

Wnt7a null and Lrp6 null limbs develop ventral structures within the dorsal compartment. In Wnt7a mutants, the double-ventral phenotype is mediated by loss of Lmx1b expression in the distal part of limb mesenchyme (Cygan et al., 1997). In the wild-type limb, the Lmx1b domain extends from the proximal region to the AER, which is marked by Fgf8 expression (Fig. 1a,f,k,p). In Wnt7a null limbs, reduced Lmx1b expression is first visible at embryonic day (E) 11.5 and is clearly evident at E12.5 (Fig. 1b,g,l,q), consistent with previous observations (Cygan et al., 1997). Whereas the distal limb mesenchyme is devoid of Lmx1b transcripts, expression in the proximal area is normal. Variable patches of residual Lmx1b expression can be seen extending along the entire anterior–posterior axis of the limb. Examples are shown with loss (Fig. 1b) or retention (Fig. 1l,q) of Lmx1b expression close to the AER.

Figure 1.

Loss of Lrp6 causes reduction of Lmx1b expression and exacerbates the Wnt7a null phenotype. a–j: dorsal view of limbs at embryonic day (E) 11.5. k–t: Dorsal view of left limb at E12.5 with anterior at the left. Lmx1b expression is indicated by the diffuse blue stain. In the upper panels (E11.5), the apical ectodermal ridge (AER) is stained for Fgf8 (dark blue).

To assess the role of Lrp6 in Lmx1b expression, we examined Lrp6 null embryos. The Lrp6 null embryos appear developmentally delayed in comparison to their wild-type littermates. At E11.5, Lrp6 null forelimbs are smaller than wild-type limbs (see Fig. 2), and Lmx1b expression is normal throughout the mesenchyme, except in a narrow margin adjacent to the AER (Fig. 1c). At E12.5, there is significant loss of Lmx1b expression in the distal mesenchyme of Lrp6 null forelimbs (Fig. 1m).

Figure 2.

Loss of Lrp6 does not affect Wnt7a expression. Wnt7a expression was compared in Dkk1d/+ forelimb at embryonic day (E) 11.0 and in Dkk1d/+Lrp6−/− forelimb at E11.5 to match the size and stage of development of the limbs.

Lmx1b expression could not be evaluated in the hindlimbs of Lrp6 null mice because of their caudal truncation (Fig. 1h,r). Hindlimb development is more complete in Dkk1d/+Lrp6−/− mice, because the 60% reduction of the Wnt antagonist Dkk1 partially rescues digit loss (MacDonald et al., 2004). The dorsal patterning defects are not rescued in these embryos, reflecting the lack of Dkk1 expression in the wild-type dorsal limb. We, therefore, examined Lmx1b expression in Dkk1d/+Lrp6−/− embryos. In these mice, expression of Lmx1b was absent in the distal mesenchyme of both forelimb and hindlimbs (Fig. 1n,s). As shown above for Wnt7a null limbs, loss of Lmx1b was not uniform, with patches of expression sometimes present in anterior (Fig. 1n) or posterior (Fig. 1s) limb.

To test the formal possibility that loss of Lrp6 affects Wnt7a expression, we examined Wnt7a in Lrp6 null limbs. No reduction in Wnt7a expression was observed (Fig. 2).

The data demonstrate that Lrp6 is required for Lmx1b expression in forelimb and hindlimb. The dependence of Lmx1b expression on Wnt7a and Lrp6 can be accounted for if Wnt7a acts as a ligand of Lrp6 to promote expression of Lmx1b in the limb.

Generation of Wnt7a, Lrp6 Double Mutant Mice

We generated Wnt7a+/−Lrp6+/− double heterozygotes by crossing Wnt7a+/− and Lrp6+/− mice. The Wnt7a+/−Lrp6+/− mice displayed no visible defects and were fertile. At E11.5, expression of Lmx1b in the double heterozygotes did not differ from wild-type (Fig. 1d,i). Thus, the combination of 50% of normal levels of Wnt7a and Lrp6 is sufficient for dorsal patterning and formation of the complete limb skeleton.

The double heterozygotes were crossed with Wnt7a+/− mice. Among 197 offspring, we obtained 7 Wnt7a−/−Lrp6+/− individuals and 43 Wnt7a−/−Lrp6+/+individuals (Table 1). This finding is consistent with the genetic linkage of Wnt7a and Lrp6 on mouse chromosome 6, with a genetic distance of 20 cM (Blake et al., 2003). Three additional Wnt7a−/−Lrp6+/− individuals and 10 Wnt7a−/−Lrp6+/+individuals were generated in a Wnt7a+/−Lrp6+/− intercross.

Table 1. Genotypes of Offspring From the Cross Wnt7a+/− × Wnt7a+/−Lrp6+/−*
NonrecombinantNonrecombinant + recombinantRecombinant
  • *

    Observed genotype frequencies are consistent with linkage of Wnt7a and Lrp6 with a distance of 20 cM.


Reduced Expression of Lmx1b in the Dorsal Limb of Wnt7a Null Mice Is Not Exacerbated by Reduction of Lrp6 Expression

Limb development in Wnt7a+/−Lrp6+/− double heterozygotes is normal. In principle, if Wnt7a were the only ligand for Lrp6 in a signaling pathway, then the phenotype of Wnt7a−/−Lrp6+/− double mutants would not be more severe than the Wnt7a−/− single mutant. Lmx1b expression in Wnt7a−/−Lrp6+/− mutants was examined at E11.5 and E12.5 (Fig. 1). The extent of loss of Lmx1b in Wnt7a−/−Lrp6+/− limbs (Fig. 1e,j,o,t) was comparable to that in Wnt7a−/−Lrp6+/+limbs (Fig. 1b,g,l,q). There were no differences in severity of nail defects or number of ectopic dorsal footpads between Wnt7a−/− single mutants and Wnt7a−/−Lrp6+/− double mutants. This result is consistent with the hypothesis that Wnt7a acts as the only ligand for Lrp6 in the pathway regulating Lmx1b expression in the dorsal-distal limb mesenchyme.

Skeletal Defects of the Posterior Limb in Wnt7a Null Mice Are Exacerbated by Reduction of Lrp6 Expression

If Wnt7a were one of several ligands for Lrp6 signaling, then reduction of the Lrp6 receptor could exacerbate the Wnt7a null phenotype by reducing the effectiveness of the other ligands. Wnt7a null limbs on a C3H genetic background exhibit minor skeletal abnormalities (Adamska et al., 2004). The fifth digit of the forelimb is lost or reduced in size (Figs. 1l,q, 3b); in addition the ulna is thin and bent, but the fibula is normal (Fig. 3b).

Figure 3.

Reduction of Lrp6 exacerbates skeletal defects in Wnt7a−/−Lrp6+/− mice. Limbs from postnatal day 0 neonates were stained with Alcian blue (cartilage) and Alizarin red (bone). Asterisks mark the loss of posterior zeugopod bones, the ulna in the forelimb, and the fibula in the hindlimb.

The defects in Wnt7a null, Lrp6 heterozygote mice (Wnt7a−/−Lrp6+/−) are much more severe than the Wnt7a null alone. The limbs are smaller than those of Wnt7a−/−Lrp6+/+littermates at E11.5 (Fig. 1e,j vs. b,g) and at E12.5 (Fig. 1o,t vs. l,q). In the forelimbs, there are only two or three digits and the ulna is missing (Fig. 3c). In the hindlimbs, there are three or four digits and the fibula is missing (Fig. 3c). Reduction of Lrp6 to 50% of wild-type levels, thus, clearly exacerbates the loss of posterior skeletal elements in Wnt7a null limbs. This finding may be accounted for if Wnt7a is one of several Lrp6 ligands involved specification of posterior elements of the limb.


Our data support a model in which Wnt7a acts as a ligand of Lrp6 in dorsal and posterior patterning during limb development. Wnt7a may be the only Lrp6 ligand required for dorsal development, whereas multiple ligands appear to be active in posterior development. One possible Lrp6 ligand for posterior development is Wnt3, which is known to be required for formation of the mouse limb (Barrow et al., 2003). The proposed effect of Lrp6 ligands might be mediated by direct or indirect effects on expression of Shh.

In addition to Lrp6, Lrp5 is also active in development of the mammalian limb. Whereas Lrp5 null limbs display no gross deformations, reduction of Lrp6 in Lrp5−/−Lrp6+/− mice results in loss of posterior digits and posterior skeletal elements of the zeugopod (Kelly et al., 2004). There is a striking similarity between the phenotypes of Wnt7a−/−Lrp6+/− mice and Lrp5−/−Lrp6+/− mice. We previously showed that loss of Lrp6 or Wnt7a in the limb can be balanced by reduction of the Wnt antagonist Dkk1 (MacDonald et al., 2004; Adamska et al., 2004). Removal of one copy of Lrp6 or both copies of Lrp5 prevents the formation of extra digits that is characteristic of Dkk1 deficiency (MacDonald et al., 2004).

Overall, the experimental evidence indicates that the number of skeletal elements in the anterior–posterior limb axis is determined by a balance of positive and negative regulators of Wnt signaling. Such a buffered system would provide robust protection against developmental malformations.


The Dkk1, Wnt7a, and Lrp6 mutant mouse stocks and genotyping assays were described previously (Pinson et al., 2000; MacDonald et al., 2004; Adamska et al., 2004). E0.5 was considered to be noon of the day when the vaginal plug was found. Whole-mount in situ hybridization was carried out with a digoxigenin-labeled probe (Bober et al., 1994) using BM Purple (Roche) as the substrate for alkaline phosphatase. Antisense mRNA probes for Fgf8 (Crossley and Martin, 1995) and Lmx1b (Chen et al., 1998) were prepared as described. Cartilage and bone were stained with Alcian blue and Alizarin red (Kimmel and Trammell, 1981). Images of cleared skeletons and whole embryos were captured with a DEI 750 Optronics digital camera and processed using Adobe Photoshop.


We thank W. Skarnes for providing the Lrp6 null mice, X. Sun and E.B. Crenshaw for cDNA probes, S.A. Camper for sharing equipment, Z.H. Sarmast for technical assistance, and B.T. MacDonald for helpful discussions. M.A. acknowledges a Postdoctoral Fellowship from the Organogenesis Training Program at the University of Michigan.