A highly conserved, cystein rich domain, presumed to bind Wnts, in the absence of a transmembrane domain represents the structural requirement for a dominant negative molecule. It was thus suggested that sFRP-3 may sequester Wnts in the extra-cellular space and prevent binding to the Frizzled membrane receptors. How this may occur in molecular terms is not yet completely understood but the recent elucidation of sFRP-3 crystal structure led to identify a Wnt-binding site in the CRDs exhibiting a conserved dimer interface that may be a feature of Wnt signaling. Indeed, all the initial reports describing the biological effects of sFRP-3 in different developmental processes Mechlorethamine hydrochloride supported this hypothesis. However, it was recently reported that sFRP-3 unexpectedly increased osteoblast differentiation through a b-catenin-independent pathway in addition to its previously known function as a decoy receptor for Wnts. As a matter of fact, EGF protein is expressed in the neuroectoderm and in the mesoderm contiguously to regions where the sFRP-3 messanger is expressed: these include a ventral area of the neural tube, the myotome 
and dermomyotome in somites and a proximal area of the developing limbs. Testing this hypothesis by classic knock out or morpholino loss of function experiments is complicated by the fact that the effects of either sFRP-3 or EGF cannot be analyzed separately from interaction with their primary ligands. In fact, blocking sFRP-3 by Xenopus morpholino injection experiments show a disorder in Wnt��s pathways. The injected embryos present eye and fore brain disorder due to the lacking of antagonism to inhibit the posteriorizing effects of Wnt��s signals, as already demonstrated in Xenopus and other vertebrates for several Wnt��s antagonists. However, in the Xenopus ectoderm, where Wnts are not known to be expressed, ablation of sFRP-3 caused a delay in the cement gland differentiation. This adhesive organ, that allows the Xenopus embryo to attach to objects in the water by secreting mucus, arise from the ectoderm that forms a pseudo-stratified columnar Epimedoside-A epithelium expressing cytokeratins by stage 28/29 NF. Thus, the delay in ectoderm differentiation induced by ablation of sFRP-3 during the cement gland development suggest a role of sFRP-3 in regulating EGFinduced proliferation that maintains the ectoderm in an undifferentiated state. Moreover, in gain of function experiments, we show that in the large majority of sFRP-3 treated embryos co-injection of EGF can restore a normal axis, whose elongation is blocked by sFRP-3. In the mouse embryo, where sFRP-3 affects axis elongation similarly to its effect in Xenopus, the concomitant transplacental delivery of both sFRP-3 and EGF restored a normal embryo morphology in the majority of the embryos and also restored sFRP3-dependent inhibition of myogenesis as it does in vitro. All these data indicate a reciprocal interference between sFRP-3 and EGF in all the assays that we have used, both in Xenopus and in mouse. The most likelyexplanation for thesephenomena is a physical interaction of the two proteins, possibly involving their CRD. Examples of non canonical protein-protein interaction have been reported to play a significant if not a major role in tissue and organ morphogenesis. For example, the Cerberus protein functions as a multivalent growth factor that antagonizes Nodal, BMP and Wnt proteins by direct interaction in the extra-cellular space, via independent binding sites. Also, Chordin antagonizes signaling by bone morphogenetic proteins by blocking binding to their receptors.