The Drosophila decapentaplegic (dpp) gene encodes a secreted protein homologous to vertebrate Bone Morphogenetic Protein 4 (BMP-4) in the TGF-ß superfamily. dpp expression is restricted to the dorsal region of the blastoderm embryo and is required for development of all dorsal cell types (Fig. 3). Several years ago we isolated another gene required for patterning the dorsal region of the blastoderm embryo called short gastrulation (sog) (François et al., 1994; Fig. 3). Although sog exerts an effect on dorsal cells in a dosage dependent fashion, it is expressed in lateral neuroectodermal cells. sog encodes a predicted secreted protein (Sog) consistent with Sog functioning to influence dorsal cell fates at a distance. Sog appears to be a dedicated Dpp antagonist (François et al., 1994). For example, sog- and dpp- mutants have opposite phenotypes with respect to expression of extreme dorsal markers, and the phenotype of sog-;dpp- double mutant embryos is the same as that of dpp- single mutants (Biehs et al, 1996). Furthermore, sog mutants suppress the haplo-insufficient lethality of dpp null mutants. While dpp-/+ individuals rarely survive to adulthood, sog-/+; dpp-/+ trans-heterozygous animals have nearly wild type viability (François et al., 1994).
Another important function of sog is to antagonize Dpp signaling within the neuroectoderm itself (Biehs et al, 1996). sog- mutants lack ventral cuticle and a subset of neuroblasts derived from the neuroectodermal region of the embryo. These sog- mutant phenotypes are enhanced by increasing the number of dpp copies. In the absence of sog, dorsally produced Dpp diffuses down into the lateral neuroectodermal region of the embryo where it activates its own expression through a positive autoregulatory loop (autoactivation) and suppress neurogenesis (Biehs et al, 1996). Thus, sog normally provides a permissive condition for the default pathway of neurogenesis by preventing Dpp autoactivation.
As mentioned on the Introductory Page, Sog and Dpp have been highly conserved during evolution (François and Bier, 1995). The Xenopus homologue of sog called chordin is expressed in the neurogenic Spemann organizer and has a potent dorsalizing activity which can mimic transplantation of the Spemann organizer by inducing a secondary axis (Schmidt et al., 1995). sog and chordin are expressed in corresponding regions of fly and frog embryos which abut domains of dpp and BMP-4 expression respectively. Most critically, Sog and Chordin or Dpp and BMP-4 can substitute for each other functionally in flies and frogs (Schmidt et al., 1995). A unifying theme in flies, frogs, and fish is that the default cell fate of ectoderm is neural and that this intrinsic tendency is actively suppressed by Dpp/BMP-4 signaling in non-neural regions of the embryo. Sog and Chordin function as anti-neural inhibitors to protect the neuroectoderm from the invasive positive feedback loop created by Dpp/BMP-4 diffusion and autoactivation (Bier, 1997).
sog and dpp also play antagonistic roles during wing vein development (Yu et al., 1996). Reduction of dpp function compromises vein formation whereas ectopic dpp supplied during pupal stages generates excess veins. Conversely, reducing sog activity promotes ectopic vein formation while sog mis-expression leads to vein-loss phenotypes. Sog appears to function by preventing Dpp from autoactivating in vein cells. Our current model is that intervein expression of Sog defines straight channels within which Dpp can diffuse and autoactivate to promote vein continuity (see wing vein section below). Interestingly, sog is only capable of blocking one of the two known actions of Dpp during vein development (i.e. vein promotion but not lateral inhibition).
The Sog product, which is a predicted secreted factor (François et al., 1994), is likely to diffuse from its lateral site of production into the dorsal region and influence cell fates (Biehs et al, 1996). Sog has a large extracellular domain containing four repeats of a 10 cysteine motif and several dibasic amino acids that could serve as potential processing sites for serine proteases. As it has been shown that vertebrate Chordin binds to BMP-4 with high affinity, Sog is likely to bind to Dpp and sequester it in an inactive form. Our data suggest that the story is not quite this simple, however, since mis-expression of sog falls short of producing phenotypes as strong as those in dpp- loss-of-function mutants (Biehs et al, 1996). In addition, there are situations in which sog cannot block Dpp activity (Yu et al., 1996).
We are currently investigating the basis for the specificity of Sog function. One possible explanation for this apparent specificity of Sog function is that Sog can block signaling through the Saxophone type-I Dpp receptor but cannot interfere with signaling through the Thick Veins type-I receptor. We also are testing whether putative secreted Sog peptides might diffuse dorsally to bind to Dpp and/or other components of the Dpp signaling pathway. For this analysis we are raising antisera to various domains of Sog as well as constructing tagged versions of Sog and components of the Dpp signaling pathway to test for direct protein-protein interactions. We also are mutagenizing each of the four Sog CR repeats to determine which in any of these domains is required for Sog activity.
