Dorsal/ventral pattern formation

The Dorsal nuclear concentration gradient. The panel on the left shows a cross section through a blastoderm embryo stained with antibodies against Dorsal to reveal the concentration gradient. The panel on the right shows a fate map of the blastoderm embryo, illustrating the domains of expression of twi, sna, and rho, which are activated by Dorsal, and of dpp and zen, which are repressed by dorsal.

Dorsal functions as both an activator and a repressor of transcription. Early in development, a set of maternally-encoded gene products (the maternal dorsal/ventral patterning system) generates an activity gradient of the Dorsal protein, with high activity on the ventral side and low activity on the dorsal side of the embryo. This gradient results in the spatially regulated expression of genes that are essential for germ layer establishment. For example, since Dorsal is an activator of the twist (twi) gene, twi is only tran-scribed in the ventrally-situated presumptive mesoderm where Dorsal activity is high. In contrast, Dorsal is a repressor of the decapentaplegic (dpp) gene, which is therefore only transcribed in the dorsally-situated presumptive ectoderm where Dorsal activity is low. A major goal of our research is to learn what determines whether Dorsal will act as an activator or a repressor of any given target gene.

     Using a wide variety of approaches, including protein biochemistry, reverse genetic analysis, and the generation and analysis of genetic mosaics, we have identified and extensively analyzed the cis-regulatory elements as well as some of the protein factors that enable Dorsal to regulate the transcription of genes such as twi and dpp in the early Drosophila embryo (1, 3, 5, 7, 9, 10, 14). We have found that the ventral specific activation of twi depends on ventral-specific enhancers in the 5' flanking region of the gene, while the ventral repression of dpp depends on ventral-specific silencers in the second intron of the gene. Both the ventral enhancers and the ventral silencers contain Dorsal binding sites that are essential for the function of these modules.

Dorsal binding sites function in a context-dependent manner. What accounts for the ability of Dorsal to activate some genes and repress others? Our find-ings suggest that it is the context of a Dorsal binding site that determines whether it will mediate activation or repression. Isolated Dorsal sites mediate activation (2). However, when Dorsal binding sites occur in the context of co-repression elements they mediate repression. These co-repression elements are thought to serve as binding sites for proteins that directly or indirectly inter-act with Dorsal to convert it from an activator into a repressor (7, 14).

    The regulatory factors that bind the co-repression elements described above are expected to play critical roles in pattern formation. By allowing Dorsal to function as both an activator and a repressor, these elements should allow the smooth monotonic Dorsal gradient to generate multiple distinct domains of gene activity. Somewhat paradoxically, however, the extensive genetic analysis of Drosophila embryogenesis, which was initiated over 15 years ago by Nüsslein-Volhard and Wieschaus, has failed to reveal the genes that encode these proteins. We have postulated that this is due to genetic pleiotropy as well as genetic redundancy, both of which could obscure the developmental phenotypes of mutations in such genes.

	Gro is required for Dorsal-mediated repression but not Dorsal-mediated activation. Wild-type embryos (A, D, G) or embryos derived from germ line clones of a hypomorphic (B, E, H) or null (C, F, I) groucho allele were stained to reveal the distribution of the dpp, zen, or twi mRNAs. A-F are sagittal views. G-I are ventral views.

We have therefore sought to identify these proteins through biochemical approaches. In the last several years, we have identified a number of co-repression element-interacting proteins. These include NTF-1, which binds to a co-repression element in dpp (the dpprepression element or DRE) (7), as well as Dead Ringer (Dri) and Cut, which bind to a co-repression element in zen(the AT2 site) (14). Our analysis of loss of function mutations in the genes encoding Dri and Cut, demonstrates that these factors are indeed critical for the function of the AT2 co-repression element in zen. As predicted, the roles of the genes encoding these factors in dorsal/ventral pattern formation was obscured by their pleiotropic roles in multiple developmental processes as well as by the existence of multiple partially redundant control regions within the zen gene. Thus, the discovery of the function of these genes in dorsal/ventral patterning was only possible because of our extensive application of both biochemical and genetic approaches.

Dorsal-mediated repression requires Groucho. How do factors like Dri and Cut convert Dorsal from an activator to a repressor? Our recent studies on the role of a factor called Groucho in Dorsal-mediated repression provide the probable answer to this question. Groucho is a pleiotropic nuclear protein that has roles in many developmental processes. This pleiotropy makes it impossible to assay the role of the groucho gene in early development by standard genetic techniques. However, using an approach that involved the creation of genetic mosaic flies carrying homozygous grouchogerm line clones, we found that Groucho protein was essential for Dorsal-mediated repression (10). As far as we can tell, Groucho does not bind directly to DNA. However, we have recently found that Groucho does bind to both Dorsal and Dri. Furthermore, DNA-bound Dorsal and Dri can cooperatively recruit Groucho to DNA in vitro (10, 14). We thus propose that a platform comprised of Dorsal and DNA-bound repressors such as Dri serves to recruit Groucho to the ventral silencer.

Model for Dorsal-mediated repression and activation. When Dorsal binds to DNA in the context of sites (e.g., the AT2 site) that interact with assistant repressors such as Dri and Cut, the resulting protein platform creates a surface for the efficient recruitment of the corepressor Groucho. When Dorsal binds to the DNA in the context of E boxes, which interact with bHLH factors such as Twi), the resulting protein platform creates a surface for the efficient recruitment of coactivators such as CBP or Dip3.

Efficient activation by Dorsal requires synergistic interactions with other factors. In the absence of DNA-bound assistant repressors, Dorsal protein directs transcriptional activation. Efficient activation requires synergistic interactions between Dorsal and other sequence specific transcription factors. For example, genetic studies carried out in the laboratory of Michael Levine at UCSD indicate that certain basic-helix-loop-helix (bHLH) transcription factors (e.g., the products of twi, daughterless, and the achaete scute complex) are able to amplify the ventral specific transcription directed by Dorsal.

    Using in vitro transcription and transient transfection assays, we have found that the Twi bHLH factor and Dorsal do indeed interact in a synergistic manner. Deletion analysis of Twi indicates that the domain of this factor responsible for the synergistic interaction is in the N-terminal portion of the protein, outside of the bHLH domain. Remarkably, deletion of this domain can, under appropriate conditions, results in a marked increase in the efficiency with which Twi activates transcription in the absence of Dorsal protein. Thus, we suspect that this N-terminal domain is serving a negative regulatory role. Dorsal may alleviate the inhibition brought about by the Twi N-terminus, perhaps by induc-ing a conformational change in Twi. In support of this model, we have shown that the proposed negative regulatory region in Twi binds directly to an N-terminal domain in Dorsal (12). In addition, we have discovered that mutations in Dorsal that significantly increase the affinity of Dorsal for Twist actually abolish activation. Thus, our current working model is that a transient interaction between Dorsal and Twi induces a conformational change in Twi and/or Dorsal that leads to transcriptional activation, perhaps by facilitating the recruitment of coactivators.