TGF-2β proteins are synthesized as pre-pro-proteins requiring cleavage of the signal peptide (black) and prodomain (orange) prior to dimerization of active ligand (red diamonds). Cells responsive to Nodal-like ligands display Type I/II Activin receptors on their cell surface with co-receptors of the EGF-CFC family required for certain ligands, such as Nodal and Gdf1. The Lefty proteins (yellow) interfere with mature ligand binding to receptors. Ligand binding triggers trans-phosphorlyation of Type I receptors by the Type II receptor serine-threonine kinase domain. Receptor-regulated R-Smads (green) become activated on the inner surface of the cell membrane through phosphorylation by Type I receptors. Activated Smad complexes include two R-Smads and the common Smad4 (black) that translocate into the nucleus to affect gene expression. Target genes typically contain binding sites for Smads and co-factors such as FoxH1 (red triangle) to achieve tissue-specific gene expression. Human mutations have been previously described in LEFTYA [MIM+601877] and LEFTYB [ref. 57; MIM*603037], ACVR2B [ref. 58; MIM+602730], CFC1 [ref. 36, 38; MIM*605194], and TDGF1 [ref. 37; MIM+187395].

Relative activity of human NODAL alleles. In vitro-synthesized mRNA encoding variant NODAL alleles were individually injected into zebrafish embryos together with a Nodal-reponsive luciferase vector and an internal luciferase standard, and embryos were harvested and measured for luciferase activity several hours later (see methods). Induced luciferase activity from each allele was normalized relative to wild type NODAL activity as described in the methods. The X axis in the graphical display indicates the allele tested (DSJ: deletion splice junction) and the Y axis (WT-adjusted activity [WTAA]) indicates the activity of the allele relative to WT NODAL. The horizontal solid gray line indicates 100% WT activity. Filled circles represent the average WTAA index from multiple experiments and vertical lines indicate the SEM. Values used for the graphical display and the number of independent experiments for each allele are given in the accompanying Suppl. Table 2. One patient was identified by sequencing to have two different changes p.A21S;A55V in cis and this construct was tested with both changes incorporated into the wild-type backbone. Similarly, a single patient had p.A19E; A63E changes in cis and a double mutant construct was studied. The remainder of the missense changes were analyzed individually.

Summary of pre-pro-domain variations in NODAL and GDF1. A), Domain architecture is shown as in Fig. 1, with arrows indicating the sites of cleavage of the signal peptide and prodomains from the canonical TGF-β ligand domain. Positions of the variant amino acids are numbered and color-coded as follows: red, severe (less than 50% activity); orange, mild (50–99% activity); black, normal activity; and blue, not tested. Note that NODAL missense changes p.H165R and p.E203K represent functionally abnormal common polymorphic variants. The leftmost panel shows a Clustal W-generated multiple sequence alignment of a putative novel domain within the NODAL prodomain associated with a cluster of functionally significant human variations between positions p.55–65. This alignment uses the Clustal W biological similarity coloring scheme (orange, identical; red, highly similar; pink, biochemically related; and black, biochemically distinct). The central panel reveals that the amino acid sequences of higher mammals can be aligned in the vicinity of the p.H165 position to allow evolutionary comparisons. Multiple sequence alignments of this region indicate that this region of the prodomain is only conserved among highly related vertebrate Nodals (data not shown). The rightmost panel contains a multiple sequence alignment of TGF-β prodomains (orange) and ligands (red) separated by the canonical RXXR consensus sequence. Numbers in italic precede the cleavage site, while those in normal font denote the amino-terminus of the mature ligand. Note that the construct delSJ (deletion of splice junction) tested in our assays is missing the consensus site. The patient with this variation has a complex in-del, with the indicated amino acids deleted and replaced in frame with heterologous aminoacids. B), The GDF1 p.C227X truncation mutation is depicted by the red triangle without fill to signify a presumed loss-of-function allele.

Homology within the TGF-β domain. A), Unlike the signal peptide and prodomains, there is extensive homology (blue for identity, yellow for similarity) within the ligand domain (red font). Structural features include two anti-parallel β-sheets (arrows) forming wing-like projections (termed finger 1 and finger 2), separated by the α-helical core buried between them. Two of the missense changes occur in finger 1, one in finger 2, and the remaining three occur within or near the core (barrel). Symbols to describe variations are as in legend for Fig. 3. B), The predicted three-dimensional structure of human NODAL. The alpha-helices are indicated by purple barrels, while the beta-sheets are indicated by the yellow, ribbon-like arrows. The ligand domain is modeled after the solved structure of BMP7 [55] and conforms to the typical features of the cystine-knot superfamily [59,60]. The positions of the missense changes are indicated within the model, with the wild type amino acid side chains shown in red.