Gradients of morphogens determine cell fates by specifying discrete thresholds of gene activities. In the Drosophila embryo, a BMP gradient subdivides the dorsal ectoderm into amnioserosa and dorsal epidermis, and also inhibits neuroectoderm formation. A number of genes are differentially expressed in response to the gradient, but how their borders of expression are established is not well understood. Evidence is presented that the BMP gradient, via the Smads, provides a two-fold input in regulating the amnioserosa-specific target genes such as Race. Peak levels of Smads in the presumptive amnioserosa set the expression domain of zen, and then Smads act in combination with Zen to directly activate Race. This situation resembles a feed-forward mechanism of transcriptional regulation. In addition, ectopically expressed Zen can activate targets like Race in the presence of low level Smads, indicating that the role of the highest activity of the BMP gradient is to activate zen (Xu, 2005).

Specific activation or repression of transcription by a combination of transcription factors is a common theme in the regulation of developmentally important genes. The results from the genetic analysis and the molecular dissection of the Race enhancer clearly show that Race is activated by the combined action of Smads and Zen. Although Smads can single handedly activate Race when overexpressed, under normal circumstances concurrent Zen activity is required. Why are both Smads and Zen necessary (Xu, 2005)?

Zen may act to restrict target gene expression specifically to the presumptive amnioserosa. Since the Dpp pathway is used repeatedly during development, other factors must function in combination with Dpp to ensure tissue specificity. Ectopic expression studies support this idea. In normal embryos, Race is activated only in regions where there are peak levels of PMad and Zen. In embryos where Zen is ubiquitously expressed, Race can now be activated in regions where there are lower levels of PMad, indicating that high level PMad is not the determining factor for amnioserosa tissue specificity. Rather PMad allows expression, and Zen determines the border of expression. The overexpression studies where Dpp can activate Race alone are interpreted to be situations where there are such high levels of Smads that Race and hindsight (hnt/pebbled) become activated promiscuously, and hence differential regulation is lost. In normal embryos, the combination of Smads and Zen ensures that the high level target genes are activated only in the presumptive amnioserosa (Xu, 2005).

In contrast, why the need for Smads? One role for Smads is suggested from the observation that Smads facilitate the binding of Zen to the Race enhancer. It is well established that Hox proteins often require co-factors for DNA binding to target enhancers. For example, composite sites that also bind the co-factor Extradenticle (Exd) ensure a greater selectivity for binding over the higher frequency Hox core site such as TAAT. In other examples, binding sites for signaling pathway effectors lie close to Hox/Selector-binding sites. The closely apposed Zen- and Smad-binding sites in the Race enhancer is one such scenario, since Zen can be thought of as a Selector gene. The current studies add to this idea of Smads and Selector cooperativity by demonstrating enhanced binding of Zen in the presence of Smads. Though the observed enhancement observed in in vitro assays is not dramatic, it is possible that in the embryo a moderate enhancement is functionally significant as is the twofold doubling of the dpp dose (Xu, 2005).

Another potential role of Smads was suggested from previous overexpression studies. Zen was able to activate Race only in the absence of Dpp if fused to a strong activation domain derived from VP16. This suggests that Smads provide a transactivation function different from that of Zen. The Smad MH2 domain has been shown to interact with the transcriptional co-activators CBP and p300. Zen has not yet been analyzed for interaction with transcriptional co-activators, however, the activation domain of Zen lies within the C-terminal 119 amino acids, and does not overlap with the homeodomain or the Mad interaction domain. Mechanistically, the difference in the activation potential between Zen and Smads could be due to their ability to recruit different co-activators to the transcriptional machinery (Xu, 2005).

Gradients of morphogens provide positional information to the cells by activating different genes at different threshold concentrations. In early Drosophila embryos, the transcriptional threshold responses to the Bicoid and Dorsal (Dl) morphogens have been extensively studied. The major mechanisms by which thresholds are established exploits the DNA-binding affinities of Bcd and Dl to their operator sites, as well as synergistic interactions with other transcription factors bound to the cis-regulatory sequences (Xu, 2005).

The BMP morphogen gradient also elicits different threshold responses from its targets, and a combinatorial mechanism is used to activate Race, a high level Dpp target. The genetic results indicate that Race, and also another high level target hnt, are activated only when a specific threshold of Zen and Smad activities are reached. In sog mutant embryos, Zen and Smad concentrations are relatively high, though below peak levels, and there is just enough of their combined activity to weakly activate Race. By contrast, in the double heterozygous embryos dpp^hr4/+; zen^w36/+, Race is not activated because Zen and Smad concentrations are below the threshold levels required for activation (Xu, 2005).

A simple way to explain these results is if the Race enhancer has low affinity to Zen and Smad proteins in vivo. To transcribe Race effectively would then require relatively high concentrations of the proteins, which are indeed reached in the dorsalmost cells. It has been known for some time that the enhancers of the high level Dl targets contain binding sites with lower affinity for Dl compared with genes responding to lower levels of Dl. Recently, it has been shown that increasing the affinities of Smad-binding sites in the Race enhancer broadens the Race expression domain, which argues that the affinities of the Smad-binding sites in this high level Dpp target gene enhancer are low. The results suggest that cooperative binding between Smads and Zen, which is dependent on their physical interaction, should increase their binding to the Race enhancer. It is possible that interacting with Smads at the protein level either increases the binding affinity of Zen or effectively increases the local concentration of Zen when Smads bind the adjacent sites. This in turn leads to a robust transcriptional response of Race. The overexpression results are consistent with such a model. Ectopic Zen can only activate Race if some detectable level of PMad is present, and in addition Zen must contain the Smad interaction domain (Xu, 2005).

How are the lower level target genes activated? u-shaped (ush) and rhomboid are expressed in a broader domain, the border coinciding exactly with that of low level PMad staining. The Zen domains, however, do not; refined zen is not broad enough, while early Zen is too broad encompassing the entire dorsal domain, though Zen could possibly be graded in this region. Thus, it is possible that this class of target genes relies on a mechanism that uses numerous high-affinity Smad sites, and/or synergistic action of Smads with other co-factor(s) besides Zen. Such a mechanism resembles the activation of target genes in the neurogenic ectoderm of the embryo by Dorsal. It has been shown that the threshold responses from these genes depend on high-affinity Dl-binding sites, as well as synergistic interactions of Dl with bHLH transcription factors (i.e., dorsal/twist interactions) (Xu, 2005).

The pnr expression domain, which is about three times broader than ush, may represent a third threshold of Dpp activity. However, pnr is a different type of target gene compared with the prior classes, in that it is repressed by Brk, which is present in a reverse gradient to Dpp. In brk mutants, pnr expands into the ventral region, while Race and ush, for example, are unchanged. It is expected that the pnr gene enhancer contains Brk-binding sites, whereas the Race enhancer does not. Brk binding sites often overlap with GNCN sites, and it is possible that in the embryo, as in the wing disc, a concentration-dependent competition between Smads and Brk establishes the expression domains of the target genes regulated by both inputs. However, whether direct competition for binding can generate threshold responses remains to be seen. In summary, it appears that different classes of Dpp target genes are regulated by different combinations of transcription factors (Xu, 2005).

One of the simple regulatory motifs used in transcriptional networks is the feed-forward or self-enabling mechanism, whereby one regulator controls a second regulator and then both bind a common target gene. It has been shown both in prokaryotes and yeasts that this mode of regulation appears relatively frequently and is favored over others, e.g., autoregulation motifs, single input motifs in which one regulator controls several genes, or regulator chain motifs whereby one gene regulates a second which regulates a third, and so on. Such an over-representation of the feed-forward motif is probably due to its potential to provide enhanced sensitivity and temporal control to the transcriptional response. The feed-forward loop is especially suitable for eliciting precise threshold responses of morphogen targets because it allows a strong response of the target gene to small changes in the activity of the regulator that initiates the loop (Dpp), due to the combined action with the second regulator (Zen). In fact Bcd and Dl use mechanisms that are reminiscent of the feed-forward loop to activate their high level targets. Bcd regulates zygotic hunchback (hb) and together Bcd and Hb activate the downstream target even-skipped (eve) stripe 2, and Dl activates sna with the help of Twi. It is striking that the three morphogen gradients involved in specifying the Drosophila embryonic axes use the feed-forward strategy to regulate downstream target genes (Xu, 2005).

An unexpected implication from these results concerns the role of the high concentration end of the BMP morphogen gradient. In Drosophila embryos, the refined zen domain depends on peak levels of BMP activity, and Zen can activate high level targets as long as there is some level of PMad present to facilitate DNA binding. It can be then concluded that, for the high level targets, the role of Dpp is twofold: to set the domain of zen, which can be referred to as a primary target gene; and then to act in combination with Zen to activate the other, secondary, target genes such as Race and hnt. In addition, with respect to the BMP gradient in the Drosophila embryo, it is further proposed that the sole purpose of the peak of the gradient is to set up the zen domain (Xu, 2005).

Robotic methods and the whole-genome sequence of Drosophila melanogaster were used to facilitate a large-scale expression screen for spatially restricted transcripts in Drosophila embryos. In this screen, scylla (scyl) and charybde (chrb), which code for dorsal transcripts in early Drosophila embryos and are homologous to the human apoptotic gene RTP801, were identified. In Drosophila, both gene products are transcriptionally regulated targets of Dpp/Zen-mediated signal transduction and appear more generally to be downstream targets of homeobox regulation. Gene disruption studies revealed the functional redundancy of scyl and chrb, as well as their requirement for embryonic head involution. From the perspective of functional genomics, these studies demonstrate that global surveys of gene expression can complement traditional genetic screening methods for the identification of genes essential for development: beginning from their spatio-temporal expression profiles and extending to their downstream placement relative to dpp and zen, these studies reveal roles for the scyl and chrb gene products as links between patterning and cell death (Scuderi, 2006).

The foundation for the current study was a survey of RNA expression patterns by automated whole-mount RNA hybridization in situ. This screening protocol led to the identification of scyl as a dorsally restricted transcript in blastoderm stage embryos. Based on its spatial and temporal expression properties, which represent a subset of the dpp expression pattern, it was postulated that scyl is a transcriptionally regulated target of the Dpp signaling cascade that specifies early embryonic dorsal fates. To test this hypothesis, scyl transcript distributions were compared in wild-type and dorsoventral patterning mutant embryos. Fate determining genes functioning downstream of the ventral morphogen Dorsal (Dl) and/or the dorsal morphogen Dpp are expected to exhibit altered expression patterns in Dl-deficient/Dpp-constitutive and Dl-constitutive/Dpp-deficient mutant embryos. Indeed, this was found to be the case for the scyl transcript. Transcription of scyl in wild-type and mutant embryos was similar to that of zen, the best characterized target of Dpp, indicating that Dpp is both necessary and sufficient for scyl transcription. In blastoderm stage embryos, zen is expressed at the posterior pole and in the dorsal-most 40% of the developing embryo. In like fashion, scyl is expressed at the posterior pole and is dorsally restricted. scyl transcripts, however, are confined to the subset of zen-expressing cells that correspond to the dorsal-most 10% of the developing embryo. Both scyl and zen are ubiquitously expressed in dorsalized embryos derived from dl mutant females and in which the dorsal morphogen Dpp is ubiquitously expressed. Neither scyl nor zen is expressed in the abdominal regions of ventralized embryos, which are derived from cactus (cact) mutant females and which lack zygotic Dpp (Scuderi, 2006).

Two observations that led to an examination of the regulation of scyl and chrb by downstream components of the Dpp signaling cascade are: (1) scyl, like zen, is a transcriptionally regulated target of Dpp-mediated signaling and (2) scyl, chrb and zen are expressed in overlapping dorsal fields. The gene encoding the divergent homeobox transcription factor Zen is itself activated by Dpp-mediated signaling in dorsal domains of the blastoderm stage embryo. In zen mutant embryos, dorsally restricted scyl and chrb transcripts are lost, placing both genes downstream of the Zen transcriptional effector of Dpp-mediated signal transduction (Scuderi, 2006).

In addition to the dorsal field of scyl and chrb expression in early embryos, segmental expression along the anteroposterior axis suggests that scyl and chrb may also be sensitive to regulation by homeobox genes other than zen. Both scyl and chrb transcripts are localized in anteroposteriorly segmented patterns in ventral regions of the blastoderm and in the three thoracic segments of stage 13 embryos. In stage 13 embryos, genes of the bithorax complex (BX-C) repress expression of target genes in abdominal segments, restricting their expression to the thorax. The expression of scyl and chrb was examined in BX-C mutant embryos: expansion was observed of thoracic expression into abdominal segments of mutant embryos, thereby placing scyl and chrb downstream of the BX-C homeobox transcription factors, Ubx, Abd-A and/or Abd-B. Consistent with this observation is the finding that Ubx binds to regulatory regions of both scyl and chrb (Scuderi, 2006).

Finally, as an extension of the observation that scyl and chrb are downstream targets of homeobox genes acting in distinct signaling pathways, bioinformatic tools were used to identify conserved elements of the scyl and chrb promoters in D. melanogaster and D. pseudoobscura. In a computational cross-genome comparison utilizing algorithms based on both Gibbs sampling and Artificial Neural Networks, one 24-nucleotide motif and three 16-nucleotide motifs were identified that are conserved in the promoter regions of both genes in both species. Motif 1 was found much more frequently than expected for a random sequence, suggestive of its role as a generic transcription factor binding site or regulatory element. Motifs 2-4 were observed much less frequently, and this statistic is interpreted to be indicative of more specific roles for motifs 2-4 in the co-regulation of scyl and chrb. With respect to the identification of defined binding sites, neither motif 1 nor motif 2 corresponds to any canonical transcription factor binding sites listed in the TRANSFAC transcription factor database. Motif 3, however, is GC-rich and contains canonical binding sites for two widely used transcriptional activators: SP1 (GCCCGCCCCCC) and AP2 (GCCCGCGGC). More notable, however, is the characterization of motif 4, which was found only twice in each of the Drosophila genomes (in the promoter regions of scyl and chrb). In scans of the TRANSFAC database, it was found that motif 4 harbors a 10-bp canonical binding site for Zen (ATTTAAATGA) (Scuderi, 2006).