Osteoblasts and chondrocytes are involved in building up the vertebrate skeleton and are thought to differentiate from a common mesenchymal precursor, the osteo-chondroprogenitor. Although numerous transcription factors involved in chondrocyte and osteoblast differentiation have been identified, little is known about the signals controlling lineage decisions of the two cell types. Here, we show by conditionally deleting beta-catenin in limb and head mesenchyme that beta-catenin is required for osteoblast lineage differentiation. Osteoblast precursors lacking beta-catenin are blocked in differentiation and develop into chondrocytes instead. In vitro experiments demonstrate that this is a cell-autonomous function of beta-catenin in an osteoblast precursor. Furthermore, detailed in vivo and in vitro loss- and gain-of-function analyses reveal that beta-catenin activity is necessary and sufficient to repress the differentiation of mesenchymal cells into Runx2- and Sox9-positive skeletal precursors. Thus, canonical Wnt/beta-catenin signaling is essential for skeletal lineage differentiation, preventing transdifferentiation of osteoblastic cells into chondrocytes.

Joints, which separate skeleton elements, serve as important signaling centers that regulate the growth of adjacent cartilage elements by controlling proliferation and maturation of chondrocytes. Accurate chondrocyte maturation is crucial for endochondral ossification and for the ultimate size of skeletal elements, as premature or delayed maturation results predominantly in shortened elements. Wnt9a has previously been implicated as being a player in joint induction, based on gain-of function experiments in chicken and mouse. We show that loss of Wnt9a does not affect joint induction, but results to synovial chondroid metaplasia in some joints. This phenotype can be enhanced by removal of an additional Wnt gene, Wnt4, suggesting that Wnts are playing a crucial role in directing bi-potential chondro-synovioprogenitors to become synovial connective tissue, by actively suppressing their chondrogenic potential. Furthermore, we show that Wnt9a is a temporal and spatial regulator of Indian hedgehog (Ihh), a central player of skeletogenesis. Loss of Wnt9a activity results in transient downregulation of Ihh and reduced Ihh-signaling activity at E12.5-E13.5. The canonical Wnt/␤-catenin pathway probably mediates regulation of Ihh expression in prehypertrophic chondrocytes by Wnt9a, because embryos double-heterozygous for Wnt9a and ␤-catenin show reduced Ihh expression, and in vivo chromatin immunoprecipitation demonstrates a direct interaction between the ␤-catenin/Lef1 complex and the Ihh promoter.

Recently canonical Wnt signaling in the ectoderm has been shown to be required for maintenance of the apical ectodermal ridge (AER) and for dorsoventral signaling. Using conditional gain-and loss-of-function ␤-catenin alleles, we have studied the role of mesenchymal ␤-catenin activity during limb development. Here, we show that loss of ␤-catenin results in limb truncations due to a defect in AER maintenance. Stabilization of ␤-catenin also results in truncated limbs, caused by a premature regression of the AER. Concomitantly, in these limbs, the expression of Bmp2, Bmp4 and Bmp7, and of the Bmp target genes Msx1, Msx2 and gremlin, is expanded in the mesenchyme. Furthermore, we found that the expression of Lmx1b, a gene exclusively expressed in the dorsal limb mesenchyme and involved in dorsoventral patterning, is reduced upon loss of ␤-catenin activity and is expanded ventrally in gainof-function limbs. However, the known ectodermal regulators Wnt7a and engrailed 1 are expressed normally. This suggests that Lmx1b is also regulated, in part, by a ␤-catenin-mediated Wnt signal, independent of the non-canoncial Wnt7a signaling pathway. In addition, loss of ␤-catenin results in a severe agenesis of the scapula. Concurrently, the expression of two genes, Pax1 and Emx2, which have been implicated in scapula development, is lost in ␤-catenin loss-of-function limbs; however, only Emx2 is upregulated in gain-of-function limbs. Mesenchymal ␤-catenin activity is therefore required for AER maintenance, and for normal expression of Lmx1b and Emx2.

The individual elements of the vertebrate skeleton are separated by three different types of joints, fibrous, cartilaginous and synovial joints. Synovial joint formation in the limbs is coupled to the formation of the prechondrogenic condensations, which precede the formation of the joint interzone. We are beginning to understand the signals involved in the formation of prechondrogenic condensations and the subsequent differentiation of cells within the condensations into chondrocytes. However, relatively little is known about the molecules and molecular pathways involved in induction of the early joint interzone and the subsequent formation of the synovial joints. Based on gain-of function studies Wnt-signalling, in particular the canonical pathway, has been implicated in the joint induction process. Here we provide genetic evidence from loss-of function analysis of embryos lacking either the central player of the canonical Wnt-pathway, β-catenin, in the limb mesenchyme or the two ligands, Wnt9a and Wnt4, demonstrating that canonical Wnt-signalling plays an important role in suppressing the chondrogenic potential of cells in the joint thereby actively allowing joint formation. Furthermore our data show that the β-catenin activity is not essential for the induction of molecular markers expressed in the joint interzone. Thus, suggesting that canonical Wnt-signalling is not required for the induction, but for the subsequent maintenance of the fate of the joint interzone cells.