Report

Distinct WNT Pathways Regulating AER Formation and Dorsoventral Polarity in the Chick Limb Bud

See allHide authors and affiliations

Science  22 May 1998:
Vol. 280, Issue 5367, pp. 1274-1277
DOI: 10.1126/science.280.5367.1274

Abstract

The apical ectodermal ridge (AER) is an essential structure for vertebrate limb development. Wnt3a is expressed during the induction of the chick AER, and misexpression of Wnt3ainduces ectopic expression of AER-specific genes in the limb ectoderm. The genes β-catenin and Lef1 can mimic the effect of Wnt3a, and blocking the intrinsic Lef1activity disrupts AER formation. Hence, Wnt3a functions in AER formation through the β-catenin/LEF1 pathway. In contrast, neither β-catenin nor Lef1 affects theWnt7a-regulated dorsoventral polarity of the limb. Thus, two related Wnt genes elicit distinct responses in the same tissues by using different intracellular pathways.

The Wnt gene family encodes a group of signaling molecules that are implicated in numerous aspects of morphogenesis in both vertebrates and invertebrates. Several chick Wnt genes are expressed in a specialized epithelial structure running along the distal margin of the limb bud, called the apical ectodermal ridge (AER), which is essential for limb morphogenesis (1, 2). Wnt3ais the first of these genes to be expressed in the limb. We therefore examined the spatiotemporal pattern of expression of Wnt3ain developing limb buds with respect to that of Fgf8, the earliest known AER marker during chick development (3,4) (Fig. 1, A through D) (5).

Figure 1

Wnt3a induces AER gene expression and regulates AER formation. (A throughD) Spatiotemporal relationship between Wnt3aand Fgf8 expression during early limb development. Images are of whole-mount in situ hybridization showing expression patterns ofWnt3a [(A) and (B)] and Fgf8 [(C) and (D)]. At stage 15, Wnt3a expression is detected in the surface ectoderm overlying the lateral plate mesoderm at approximately the level of somites 14 to 18 [between arrowheads in (A)], whereasFgf8 expression is not detected in the limb field [between arrowheads in (C)] before stage 16 (5). As the limb buds emerge, expression of Wnt3a becomes elevated in the distal limb ectoderm forming the AER, whereas it is maintained at a lower level in the proximal dorsal ectoderm and is reduced in the proximal ventral ectoderm of the limb (8). Once the AER has formed, both Wnt3a (B) and Fgf8 (D) are primarily confined to the AER through at least stage 26. (E andF) Misexpression of Wnt3a induces ectopic expression of an AER-specific gene, Fgf8, in patches on both the dorsal and ventral ectoderm. Occasionally there are also morphological effects on the AER. In some cases, the AER appears wider in places, giving it an irregular appearance [green arrowhead in (E)]. In other cases, the AER is disrupted [red arrowhead in (F)] or ectopic AERs form, branching toward the ventral side [yellow arrowhead in (F)]. Panels show lateral views of the limbs, dorsal side up. Single- and double-color whole-mount in situ hybridizations of chick embryos were performed as described (36). The probes used were Wnt3a [372 base pairs (bp) (32)] and Fgf8 [800 bp (4)].

Wnt3a transcripts are detected before Fgf8transcripts in the limb field ectoderm but not in the flank outside the limb fields. Subsequently, Wnt3a expression is up-regulated in the ectoderm cells near the dorsoventral (DV) border.Fgf8 expression is initiated and then up-regulated within the region of high Wnt3a expression during AER formation. From stage 20 on, Wnt3a and Fgf8 expression are confined primarily to the mature AER. Thus, Wnt3a expression appears to presage Fgf8 expression and AER formation.

To verify the epistatic relationship between Wnt3a andFgf8 that is suggested by the expression data, we ectopically delivered each factor to developing limb buds. We misexpressed Wnt3a in the limb ectoderm using a replication-competent retroviral vector and assayed for the expression patterns of the various AER markers (6). Misexpression ofWnt3a induced ectopic expression of AER-specific genes, including Bmp2, Fgf4, and Fgf8, in broad patchy domains in the ectoderm of nearly 100% of infected limbs (Fig. 1E) (5). However, Wnt3a expression was not induced in the ectoderm by either fibroblast growth factor 4 (FGF4) protein or Fgf8-virus (5). This suggests thatWnt3a acts upstream of FGFs in establishing AER gene expression.

In addition to its effect on AER gene expression, Wnt3amisexpression occasionally led to disruption of the AER or to formation of an ectopic AER extending ventrally, or both (Fig. 1, E and F). These morphological effects on the AER are reminiscent of those seen after misexpression of Radical fringe (7). We therefore examined Radical fringe expression and found that it was ectopically expressed in Wnt3a-infected limbs (8). Disruption of the AER morphology was only seen in a subset of Wnt3a-infected limb buds, which is consistent with the finding that Radical fringe only affects AER formation when it is misexpressed at the earliest stages of limb development (7).

The FGFs produced in the AER are responsible for maintaining the proliferative state of the undifferentiated mesoderm at the distal tip of the limb bud, the progress zone (PZ) (1). To verify that the FGFs induced in the limb bud ectoderm by Wnt3a are functional signals, we examined the expression of several PZ markers:Fgf10 (9), Msx1 (10),Nmyc (11), and Slug (12). Equivalent results were obtained with each of these markers (Fig.2) (8). When the AERs were removed from experimental limb buds, expression of the PZ markers was rapidly lost (Fig. 2, C and D). Application ofWnt3a-expressing cells to the AER-deprived limbs inducedFgf8 expression and restored expression of the PZ markers (Fig. 2, E and F). To show that this response was due to the ectopic expression of FGFs and not to a direct action of Wnt3aitself, we removed the adjacent distal ectoderm as well as the AER so that the Wnt3a cells could not induce ectodermalFgf8 expression (Fig. 2G). Under these conditions,Wnt3a induced little or no expression of the PZ genes (Fig.2H). The maintenance of the PZ is critical for outgrowth of the limb bud. The long-term effects of virally mediated Wnt3amisexpression, and consequent ectopic FGF production by the ectoderm, included some cases in which extra outgrowth formed digitlike structures (5). Wnt3a thus appears to influence both morphological AER formation and induction of AER-specific genes in the early limb bud.

Figure 2

Induction of PZ gene expression byWnt3a requires the ectoderm expressing Fgf8. Embryos were manipulated at stage 20 and harvested after 24 hours for in situ hybridization for Fgf8 [(A), (C), (E), and (G)] and Fgf10 (probed with a 555-bp fragment; a gift from H. Ohuchi) [(B), (D), (F), and (H)]. (A andB) Normal expression patterns in the contralateral limbs.Fgf8 (A) is specifically expressed in the AER, whereasFgf10 (B) is strongly expressed in the PZ. (Cand D) The AERs were removed with fine tungsten needles and forceps and replaced with control line 0 cell pellets fixed with a platinum staple at the distal rim of the mesoderm. In the absence of the AER, there is no expression of Fgf8 (C) or of other members of the Fgf family, and expression ofFgf10 disappears within 20 hours (D). (E andF) The AERs were removed and replaced withWnt3a cell pellets. (E) Wnt3a inducesFgf8 expression (purple staining) in the ectoderm adjacent to implanted cell pellets expressing Wnt3a (pink staining pointed to by an arrow, reflecting a second hybridization with a 900-bp probe detecting the viral vector). (F) Expression ofFgf10 is restored in the distal mesenchyme next toWnt3a cell pellets. (G and H) The limb buds were denuded of the distal ectoderm on both the dorsal and the ventral sides as well as the AERs, and then Wnt3a cell pellets were implanted. The proximal ectoderm was left intact.Wnt3a cell pellets induced neither Fgf8 (G) norFgf10 expression (H).

The members of the vertebrate Wnt gene family have been categorized by their relative ability to transform murine mammary epithelial cells (13). A similar classification can be made on the basis of the ability to promote axis duplications in earlyXenopus embryos (14). Yet even Wntgenes categorized into the same functional group can play distinct roles during limb development. During the period of AER formation,Wnt7a, which is placed in the same highly transforming axis-duplicating class as Wnt3a, is exclusively expressed in the dorsal limb ectoderm (15, 16) and functions as the signal for DV patterning in the distal limb (17-19). In early stages of limb development, expression domains of Wnt3a and Wnt7a overlap in the dorsal ectoderm. To assess whether the two Wnt family genes are functionally redundant, we compared their abilities to induce expression of AER markers and affect DV patterning. Early limb buds were infected with either Wnt3a or Wnt7a virus (5) and examined for Fgf8 expression (Fig.3, A, C, and E). Unlike misexpression ofWnt3a, misexpression of Wnt7a never affected the expression pattern of Fgf8 or the morphology of the AER, which suggests that Wnt7a is not involved in AER formation. The effects of Wnt3a and Wnt7a on DV patterning were assayed by monitoring expression of the dorsal mesenchymal markerLmx1 (Fig. 3B). Misexpression of Wnt7a induced strong ectopic expression of Lmx1 in the distal half of ventral mesenchyme (Fig. 3F), and produced a biconvex-shaped limb bud, in contrast to the normal ventrally curving limb bud morphology (Fig.3, B and F) (17, 18). Wnt3amisexpression did not produce the morphological bidorsal phenotype and had little or no effect on Lmx1 expression (Fig. 3D).

Figure 3

WNT3a and WNT7a elicit distinct activities that are differentially mediated by β-catenin. Embryos were infected with virus carrying Wnt3a [(C) and (D)], Wnt7a(16) [(E) and (F)], or an activated form of β-catenin (6) [(G) and (H)] at stage 10 and harvested at stage 24 to analyze expression of Fgf8 [(A), (C), (E) and (G)] and Lmx1 (17) [(B), (D), (F), and (H)]. The different viruses all resulted in comparable infections, spread throughout the limb ectoderm and mesoderm (5). Ectodermal targets are, in general, induced in a patchy pattern, whereas mesodermal targets are more uniformly induced, perhaps reflecting the relative kinetics of viral spread in the two tissues. (A and B) Normal expression ofFgf8 and Lmx1. Fgf8 expression is ectopically induced in the limb ectoderm by Wnt3a(C) in nearly 100% of infections, but never byWnt7a (E). In contrast, Lmx1expression is strongly induced throughout the ventral mesenchyme byWnt7a in nearly 100% of infections (F) but not by Wnt3a (D). (D) In 20 to 30% of infected limbs, Wnt3a induces patchy weak expression in the ventral mesenchyme and shifts the distal border of Lmx1expression slightly ventral to the AER (5), which appeared to be correlated with the formation of widened and ectopic AERs. Activated β-catenin is able to mimic Wnt3a in inducing ectopic Fgf8 expression in the dorsal and the ventral ectoderm (G) and both the punctate ventral expression and the ventral shift in the distal border ofLmx1 expression (H) (5) at a comparable frequency. However, activated β-catenin does not induce strong ventral expression of the Wnt7a targetLmx1 in the mesenchyme.

To understand how these two Wnt genes elicit distinct responses, we investigated whether Wnt3a andWnt7a act through the same signaling pathway. In other systems, signaling by the highly transforming Wnt genes, including Wnt3a, has been shown to be transduced by preventing degradation of cytoplasmic β-catenin, a protein that is ubiquitously expressed in vertebrate embryos (20-22). We misexpressed an activated mutant form of β-catenin (20, 23,24) in limb buds (5) and found that β-catenin activity simulated the effect ofWnt3a but not of Wnt7a. Misexpression of the activated form of β-catenin induced ectopic expression ofFgf8 (Fig. 3G) and expression of other AER markers, including Fgf4 and Bmp2, in the ectoderm and up-regulated the downstream PZ markers Msx1 andNmyc in the mesoderm (8). In contrast, misexpression of β-catenin did not induce the strong mesenchymal expression of Lmx1 seen in response toWnt7a (Fig. 3H). Moreover, like Wnt3a-infected limb buds, activated β-catenin–infected limbs retain a concave ventral morphology in contrast to the biconvex morphology ofWnt7a-infected ones (Fig. 3, D, F, and H). Thus,Wnt3a and Wnt7a function in limb morphogenesis through β-catenin–dependent and –independent pathways, respectively.

β-catenin forms a complex with members of the LEF/XTCF family to activate transcription of downstream genes (25-28). To explore the possible involvement of LEF in WNT3a signaling, we cloned a chick homolog of Lef1.Lef1 is strongly expressed in the AER and distal mesenchyme of the chick limb primordia (Fig. 4A). In addition to the limb buds, intense expression is observed in the developing medial somites and in the tail bud (8), both of which are regions believed to be targets of Wnt3a signaling (29, 30).

Figure 4

LEF1 mediates signaling by WNT3a but not WNT7a. (A) Expression of chick Lef1 in a stage-23 limb bud. (B and C) Lef1expression is enhanced and ectopically induced in the entire limb bud by Wnt3a misexpression [(B), dorsal view] but not byWnt7a misexpression (C). (D) Misexpression ofLef1 in the limb bud induces ectopic expression ofFgf8 (arrowheads). (E) BlockingLef1 activity by ΔLef1 misexpression suppresses Fgf8 expression in the AER and disrupts formation of the morphological AER. Gaps in Fgf8 expression (between arrowheads) are seen where the morphological AER is disrupted. Limb outgrowth at the gaps is markedly retarded. (F) Expression of Lmx1 in the dorsal mesenchyme is not suppressed by misexpression of ΔLef1 in the infected limb, which is reduced in size because of disruption of the AER. The full-length chick Lef1 gene was cloned by screening a cDNA library of a chick limb bud at stage 20 to 22 under reduced stringency conditions by hybridization with an approximately 1.0-kbLef1 mouse cDNA. The full-length sequence has been deposited in the GenBank database under accession number A-F 064462. The entire coding frame of chick Lef1 and the deletion mutant of chick Lef1 deprived of 31 NH2-terminal amino acids (ΔLef1) were cloned into RCAS(BP)A as described in (6).

To examine whether Wnt3a regulates Lef1expression during chick development, embryos were injected withWnt3a virus at stage 10 and assayed for Lef1expression at stages 22 to 24. Misexpression of Wnt3amarkedly enhanced Lef1 expression in both the ectoderm and mesoderm of the limb bud and also in the tail bud (Fig. 4B) (8). Misexpression of Wnt7a had no effect onLef1 expression (Fig. 4C). Thus, Lef1 is a specific target of WNT3a signaling.

To test whether the up-regulation of Lef1 mediates the WNT3a effects on AER marker genes, we misexpressed a retrovirus carryingLef1 in the ectoderm of developing limb buds.Fgf8 expression was ectopically induced in the infected limb ectoderm (Fig. 4D). However, like Wnt3a and β-catenin, Lef1 was unable to mimic the ability of Wnt7a to induce Lmx1 expression (8).

To assess whether LEF1 activity is necessary for the induction of AER morphogenesis by WNT3a, we constructed a retrovirus carrying a deleted form of Lef1 cDNA (ΔLef1) that binds to DNA but not to β-catenin, thereby blocking intrinsic Lef1 activity as a dominant negative mutant (27, 28). Misexpression of ΔLef1 disrupted the AER and subsequently perturbed normal growth of the underlying mesenchyme, resulting in severe defects in limb outgrowth and patterning (Fig. 4E). However, expression of the WNT7a target Lmx1 was not suppressed by misexpression of ΔLef1 (Fig. 4F). Thus the β-catenin/LEF1 pathway is necessary and sufficient to induce a WNT3a response but not a WNT7a response.

Apparently, Lef1 is both a target and a mediator of WNT3a signaling. In the early Xenopus embryo, the preexisting level of the Lef1 homolog XTCF3 is effectively saturating (25, 27). In contrast, we find that in the chick limb bud, increased expression of LEF1 activates Wnt3atargets. Obviously, ectodermal cells must be capable of responding to WNT3a before LEF1 up-regulation. This could be mediated by the low preexisting levels of LEF1, which we observe throughout the ectoderm, in which case the LEF1 induction would constitute a positively reinforcing feedback loop. Alternatively, the induction ofLef1 could be supported by another member of the LEF1 family (31).

Our results suggest that WNT3A/β-catenin/LEF1 signaling plays a key role in endogenous chick AER formation. Surprisingly, Wnt3ais neither expressed in the AER nor implicated in its formation during mouse limb development (15, 30). Several otherWnt genes have also been shown to be expressed differentially between mouse and chick embryos, both in the developing central nervous system and in limb buds, including some in the murine AER (15, 32). Different Wnt genes could substitute for one another as long as they activate the same intracellular signaling pathway mediated by β-catenin/LEF1. It is therefore probable that another species of Wnt that is expressed in the mouse AER plays the same role as Wnt3a in the chick.

Both WNT3a and WNT7a proteins act, at least in part, on the mesoderm, where they activate distinct targets; WNT3a induces Lef1whereas WNT7a induces Lmx1, which implies that receptors for both factors must be present on the surface of mesenchymal cells. In spite of previous data suggesting that all members of the highly transforming class of Wnt genes act through β-catenin, our results indicate that the induction of Lmx1 expression by WNT7a signaling is not mediated by β-catenin and LEF1. Precedent exists for more divergent Wnt genes, such asWnt5a, to act through distinct signaling cascades (33). Transcriptional activation of downstream genes by distinct WNT pathways allows for their different inductive roles in the same tissue during development.

  • * These authors contributed equally to this work.

  • Present address: Department of Biophysics, Kyoto University, Sakyo-ku, Kyoto 606, Japan.

  • To whom correspondence should be addressed. E-mail: belmonte{at}salk.edu

  • § The laboratories of these authors contributed equally to this study.

REFERENCES AND NOTES

View Abstract

Navigate This Article