Morphogen gradients play fundamental roles in patterning and cell specification during development by eliciting differential transcriptional responses in target cells. In Drosophila, Decapentaplegic (Dpp), the BMP2/4 homolog, downregulates transcription of the nuclear repressor brinker (brk) in a concentration-dependent manner to generate an inverse graded distribution. Both Dpp and Brk are crucial for directing Dpp target gene expression in defined domains and the consequent execution of distinct developmental programs. Thus, determining the mechanism by which the brk promoter interprets the Dpp activity gradient is essential for understanding both Dpp-dependent patterning and how graded signaling activity can generate different responses through transcriptional repression. This study uncovered key features of the brk promoter that suggest it uses a complex enhancer logic not represented in current models. First, it was found that the regulatory region contains multiple compact modules that can independently drive brk-like expression patterns. Second, each module contains binding sites for the Schnurri/Mad/Medea (SMM) complex, which mediates Dpp-dependent repression, linked to regions that direct activation. Third, the SMM repression complex acts through a distance-dependent mechanism that probably uses the canonical co-repressor C-terminal Binding Protein (CtBP). Finally, these data suggest that inputs from multiple regulatory modules are integrated to generate the final pattern. This unusual promoter organization may be necessary for brk to respond to the Dpp gradient in a precise and robust fashion (Yao, 2008).

The brk gene is unique in that eleven SMM sites are present in its regulatory region: no other locus in the genome has more than three sites. The SMM sites are characterized by a GRCGNC(N5)GTCTG motif. The GRCGNC sequence is bound by Mad while Med binds GTCTG, and the five-nucleotide spacer is crucial for recruitment of Shn to the complex. These sites are widely dispersed over 16 kb and separated from each other by 0.35 to 5.5 kb, with the exception of sites 7/8/9, which are clustered in a 183 bp region. For seven of the eleven SMM sites (3, 4, 5, 7/8/9 and 10), sequences that mediate transcriptional activation are located within ~380 bp of the SMM sites. These SMM sites and linked activator sequences can independently generate brk-like expression patterns, suggesting that they function as autonomous modules. The fact that the L1 transgene, which contains a single SMM site (#1), also drives a brk-like pattern, strongly argues for a sixth module in addition to the five that have been demonstrated. Thus, the 11 SMM sites in the brk regulatory region probably correspond to a total of 9 or 10 distinct modules, depending on whether the 7/8/9 cluster represents one or more modules. The evolutionary conservation of this unusual promoter organization provides additional support for its functional importance. Analysis of brk flanking regions in D. pseudoobscura and D. virilis, which are 30 and 40 million years distant from D. melanogaster, identified 12 and 11 SMM sites, respectively, arranged with a similar spacing relative to the basal promoter. Furthermore, 11 sites are found upstream of the brk-coding region in the mosquito Anopheles gambiae, which is separated from Drosophila by ~200 million years (Yao, 2008).

How does the brk promoter read the pMad gradient and generate a complementary graded expression pattern, and what benefit could the presence of multiple modular enhancers confer in generating the Brk gradient? This work, as well as previous studies, indicates that SMM sites act as sensors for Dpp signaling by binding a repressor complex that antagonizes broadly expressed activators in a dose-dependent manner. The data that SMM-mediated repression has a limited range suggests that each module can autonomously generate an output representing the balance between activation and signaling-dependent repression within that module. The patterns produced by individual modules probably reflect variation in SMM site sequence and affinity, the distance between SMM and activator sites, activator site sequence and number, as well as whether sites for additional transcription factors are present (Yao, 2008).

The endogenous brk pattern does not appear to reflect the activity of a single 'dominant' module, but rather is a composite pattern resulting from integration of multiple modular inputs. This can be inferred from the fact that large promoter fragments containing more than one module (e.g. L2 and L6) drive patterns that resemble, but are not identical to, those of their constituent modules. Furthermore, the additive effect of module multimerization on expression levels in regions of low Dpp activity is also consistent with integration across modules. Finally, strong support for this idea comes from the buffering capacity of multimodular promoter fragments (Yao, 2008).

A significant feature of the brk promoter is the remarkable ability of intact modules to override medial activation by mutant modules that are Dpp insensitive. This is apparent from the data that activators uncoupled from Dpp-dependent repression drive strikingly different expression patterns in isolation than they do in the context of larger fragments containing additional wild-type modules. Thus, disruption of the SMM sites in module 3 (L12^M3), module 4 (L14^M4) and module 7/8/9 (L15^M7/8/9) caused derepression throughout the center of the wing disc. However, the same mutations in a larger fragment containing several additional modules (L13^M3, L13^M4 and L13^M7/8/9) resulted in no derepression in the center of the disc. These results are inconsistent with a simple model in which only modules unbound by SMM complexes contribute to the transcriptional output of the promoter. If this were the case, in cells at the AP boundary, high levels of Dpp signaling would repress all intact modules in the L13^M variants, leaving the constitutively active mutant module(s) free to interact with the transcriptional machinery. As a consequence, L13^M variants would be expected to upregulate expression throughout the medial region of the disc. One potential explanation for the ability of wild-type modules to dampen expression from mutant modules could be that activators from SMM-repressed modules may compete disproportionately with activators from unrepressed modules for access to the transcriptional machinery, thus diluting the effect of the mutant modules. Alternatively, the SMM repressor complexes bound at multiple modules could act cooperatively (perhaps by modifying chromatin structure), thus reducing the output from adjacent mutant modules. In both cases, the absence of any expression in the medial region even with two Dpp-insensitive modules present, argues that repressed modules make a significant contribution to the transcriptional output compared with the derepressed modules. Such an integrative mechanism also provides a framework for understanding how poorly resolved patterns like those generated by module 4, could be refined to generate the wild-type brk pattern. An important consequence of this promoter logic is that although individual SMM repression complexes act locally, modules in aggregate can, nevertheless, exert a long-range/global effect on promoter activity (Yao, 2008).

The specialized architecture of the brk promoter may provide a mechanism to respond to Dpp signaling in a uniquely precise and robust fashion. Multiple modules allow simultaneous parallel reads of the pMad gradient, thus increasing the precision with which the brk promoter detects Dpp morphogen levels. Integration would also be predicted to increase the fidelity of the brk promoter response by making it less sensitive to fluctuations at any individual module. This fidelity would be further enhanced by a disproportionate contribution from repressed as opposed to active modules. This buffering ability of the brk promoter is likely to be important in preventing stochastic fluctuations or transcriptional noise in wild-type animals, as well as in rendering brk transcription more resistant to mutational insults (Yao, 2008).

Several developmentally important genes have modular promoters consisting of multiple non-overlapping enhancers that function autonomously to generate a composite expression pattern. The segmentation gene eve provides an archetypal example, with five enhancers that drive expression in seven discrete stripes in the embryonic blastoderm. Although brk resembles eve in its modular promoter organization and the ability of individual modules to function independently, three key differences make brk unique. First, individual eve elements are bound by different combinations of activators and repressors, and thus drive expression in distinct stripes in the embryo. By contrast, individual brk modules respond to a common set of repressive cues and drive expression in largely overlapping domains. Second, in any given region of the embryo, the eve pattern represents the output of a single enhancer. By contrast, multiple brk modules are active in each cell and contribute collectively to the final expression pattern. A final crucial difference is that in eve short-range repression prevents crosstalk between enhancers that drive expression in different stripes, while in brk the outputs of modules that appear to respond autonomously to the Dpp gradient are integrated. Why do brk and eve cis-regulatory elements display different properties, even though both use the CtBP co-repressor? One potential explanation arises from the fact that CtBP functions as part of a complex that includes histone deacetylases, histone methylase/demethylases and SUMO E2/E3 ligases. CtBP complexed with SMM on the brk promoter may recruit a different subset of activities from a CtBP-gap gene complex on eve enhancers. In addition, the SMM complex itself may recruit unique activities to the brk promoter. Furthermore, as the activators that mediate brk and eve expression are likely to be distinct, they may be affected by CtBP differentially (Yao, 2008).

Two lines of evidence argue that SMM activity is distance dependent: Shn interacts directly with the co-repressor dCtBP, and there is a functional requirement for close linkage of SMM sites and activator sequences. Short-range repression appears to be a property of the SMM complex in other contexts as well, as an SMM site located ~89 bp from a germ cell-specific enhancer in the bag of marbles (bam) gene fails to mediate repression when this spacing is increased. Furthermore, an SMM site and activator sequences are closely linked in a compact 514 bp Dpp-dependent enhancer in the gooseberry (gsb) promoter. Loss of dCtBP binding strongly reduces repression by ShnCT^M, demonstrating that this interaction is relevant in vivo. However, ShnCT^M still retains residual ability to repress brk-LacZ, and brk is not ectopically expressed in dCtBP clones in the wing disc. This could indicate that the dCtBP interaction motif actually has a different function in vivo. Alternatively, Shn may employ redundant repression strategies, consistent with the current view that Shn proteins act as scaffolds for co-repressors, and indeed co-activators and other modulators, enabling the Smad complex to elicit different transcriptional responses dependent on cellular context (Yao, 2008).

The identity of the activator(s) targeted by the SMM repression complex remain to be determined, as do the precise sequences to which it binds. It is possible that different brk modules incorporate inputs from distinct activators, and that some of these activators are spatially or temporally restricted. In addition to inputs from the SMM complex and the activator, there is genetic evidence that brk negatively autoregulates its own expression, most prominently in the mediolateral regions of the wing disc. Consistent with this, the brk promoter contains multiple sites that match the Brk consensus and may mediate autoregulation (Yao, 2008).

Bioinformatics methods have identified enhancers that mediate restricted expression in the Drosophila embryo. However, only a small fraction of the predicted enhancers actually work when tested in vivo. In the present study, co-regulated neurogenic enhancers that are activated by intermediate levels of the Dorsal regulatory gradient are shown to contain several shared sequence motifs. These motifs permit the identification of new neurogenic enhancers with high precision: five out of seven predicted enhancers direct restricted expression within ventral regions of the neurogenic ectoderm. Mutations in some of the shared motifs disrupt enhancer function, and evidence is presented that the Twist and Su(H) regulatory proteins are essential for the specification of the ventral neurogenic ectoderm prior to gastrulation. The regulatory model of neurogenic gene expression defined in this study permitted the identification of a neurogenic enhancer in the distant Anopheles genome. The prospects for deciphering regulatory codes that link primary DNA sequence information with predicted patterns of gene expression are discussed (Markstein, 2004).

Previous studies identified two enhancers, from the rho and vnd genes, that are activated by intermediate levels of the Dorsal gradient in ventral regions of the neurogenic ectoderm. The present study identified a third such enhancer from the brk gene. This newly identified brk enhancer corresponds to one of the 15 optimal Dorsal-binding clusters described in a previous survey of the Drosophila genome. Although one of these 15 clusters has been shown to define an intronic enhancer in the short gastrulation (sog) gene, the activities of the remaining 14 clusters were not tested. Genomic DNA fragments corresponding to these 14 clusters were placed 5' of a minimal eve-lacZ reporter gene, and separately expressed in transgenic embryos using P-element germline transformation. Four of the 14 genomic DNA fragments were found to direct restricted patterns of lacZ expression across the dorsoventral axis that are similar to the expression patterns seen for the associated endogenous genes (Markstein, 2004).

The four enhancers respond to different levels of the Dorsal nuclear gradient. Two direct expression within the presumptive mesoderm where there are high levels of the gradient. These are associated with the Phm and Ady43A genes. The third enhancer maps ~10 kb 5' of brk, and is activated by intermediate levels of the Dorsal gradient, similar to the vnd and rho enhancers. Finally, the fourth enhancer maps over 15 kb 5' of the predicted start site of the CG12443 gene, and directs broad lateral stripes throughout the neurogenic ectoderm in response to low levels of the Dorsal gradient. In terms of the dorsoventral limits, this staining pattern is similar to that produced by the sog intronic enhancer (Markstein, 2004).

The remaining ten clusters failed to direct robust patterns of expression and are thus referred to as 'false-positives'. Since analysis of spacing and orientation of the Dorsal sites alone did not reveal features that could discriminate between the false positives and the enhancers, whether additional sequence motifs could aid in this distinction was examined. A program called MERmaid was developed that identifies motifs over-represented in specified sets of sequences. MERmaid analysis identified a group of motifs, which was largely specific to the brk, vnd and rho enhancers, suggesting that the regulation of these coordinately expressed genes is distinct from the regulation of genes that respond to different levels of nuclear Dorsal (Markstein, 2004).

The rho, vnd and brk enhancers direct similar patterns of gene expression. The rho and vnd enhancers were previously shown to contain multiple copies of two different sequence motifs: CTGNCCY and CACATGT. A three-way comparison of minimal rho, vnd and brk enhancers permitted a more refined definition of the CTGNCCY motif (CTGWCCY), and also allowed for the identification of a third motif, YGTGDGAA. The CACATGT and YGTGDGAA motifs bind the known transcription factors, Twist and Suppressor of Hairless [Su(H)], respectively. All three motifs are over-represented in authentic Dorsal target enhancers directing expression in the ventral neurogenic ectoderm, as compared with the 10 false-positive Dorsal-binding clusters. Some of the false-positive clusters contain motifs matching either Twist or CTGWCCY; however, none of the false-positive clusters contain representatives of both of these motifs. The rho enhancer is repressed in the ventral mesoderm by the zinc-finger Snail protein. The four Snail-binding sites contained in the rho enhancer share the consensus sequence, MMMCWTGY; the vnd and brk enhancers contain multiple copies of this motif and are probably repressed by Snail as well (Markstein, 2004).

The functional significance of the shared sequence motifs was assessed by mutagenizing the sites in the context of otherwise normal lacZ transgenes. Previous studies have suggested that bHLH activators are important for the activation of rho expression, since rho-lacZ fusion genes containing point mutations in several different E-box motifs (CANNTG) exhibited severely impaired expression in transgenic embryos. However, it was not obvious that the CACATGT motif was particularly significant since it represents only one of five E-boxes contained in the rho enhancer. Yet, only this particular E-box motif is significantly over-represented in the rho, vnd and brk enhancers. vnd-lacZ and brk-lacZ fusion genes were mutagenized to eliminate each CACATGT motif, and analyzed in transgenic embryos. The loss of these sites causes a narrowing in the expression pattern of an otherwise normal vnd-lacZ fusion gene. By contrast, the brk pattern is narrower in central and posterior regions, but relatively unaffected in anterior regions. The brk enhancer contains two copies of an optimal Bicoid-binding site, and it is possible that the Bicoid activator can compensate for the loss of the CACATGT motifs in anterior regions (Markstein, 2004).

Similar experiments were performed to assess the activities of the Su(H)-binding sites (YGTGDGAA) and the CTGWCCY motif. Mutations in the latter sequence cause only a slight reduction and irregularity in the activity of the vnd enhancer, whereas similar mutations nearly abolish expression from the brk enhancer. Thus, CTGWCCY appears to be an essential regulatory element in the brk enhancer, but not in the vnd enhancer. Mutations in both Su(H) sites in the brk enhancer caused reduced staining of the lacZ reporter gene, suggesting that Su(H) normally activates expression. Further evidence that Su(H) mediates transcriptional activation was obtained by analyzing the endogenous rho expression pattern in transgenic embryos carrying an eve stripe 2 transgene with a constitutively activated form of the Notch receptor (Notch^IC). rho expression is augmented and slightly expanded in the vicinity of the stripe2-Notch^IC transgene. A similar expansion is observed for the sim expression pattern (Markstein, 2004).

To determine whether the shared motifs would help identify additional ventral neurogenic enhancers, the genome was surveyed for 250 bp regions containing an average density of one site per 50 bp and at least one occurrence of each of the four motifs for Dorsal, Twist, Su(H) and CTGWCCY. In total, only seven clusters were identified. Three of the seven clusters correspond to the rho, vnd and brk enhancers. Two of the remaining clusters are associated with genes that are known to be expressed in ventral regions of the neurogenic ectoderm: vein and sim. Both clusters were tested for enhancer activity by attaching appropriate genomic DNA fragments to a lacZ reporter gene and then analyzing lacZ expression in transgenic embryos. The cluster associated with vein is located in the first intron, about 7 kb downstream of the transcription start site. The vein cluster (497 bp) directs robust expression in the neurogenic ectoderm, similar to the pattern of the endogenous gene. The cluster located in the 5' flanking region of the sim gene (631 bp) directs expression in single lines of cells in the mesectoderm (the ventral-most region of the neurogenic ectoderm), just like the endogenous expression pattern. These results indicate that the computational methods define an accurate regulatory model for gene expression in ventral regions of the neurogenic ectoderm of D. melanogaster (Markstein, 2004).

To assay the generality of these findings, genomic regions encompassing putative sim orthologs from the distantly related dipteran Anopheles gambiae were scanned for clustering of Dorsal, Twist, Su(H), CTGWCCY and Snail motifs. One cluster located 865 bp 5' of a putative sim ortholog contains one putative Dorsal binding site, two Su(H) sites, three CTGWCCY motifs (or close matches to this motif), a CACATG E-box and several copies of the Snail repressor sequence MMMCWTGY. A genomic DNA fragment encompassing these sites (976 bp) was attached to a minimal eve-lacZ reporter gene and expressed in transgenic Drosophila embryos. The Anopheles enhancer directs weak lateral lines of lacZ expression that are similar to those obtained with the Drosophila sim enhancer. These results suggest that the clustering of Dorsal, Twist, Su(H) and CTGWCCY motifs constitutes an ancient and conserved code for neurogenic gene expression (Markstein, 2004).

This study defines a specific and predictive model for the activation of gene expression by intermediate levels of the Dorsal gradient in ventral regions of the neurogenic ectoderm. The model identified new enhancers for sim and vein in the Drosophila genome, as well as a sim enhancer in the distant Anopheles genome. Five of the seven composite Dorsal-Twist-Su(H)-CTGWCCY clusters in the Drosophila genome correspond to authentic enhancers that direct similar patterns of gene expression. This hit rate represents the highest precision so far obtained for the computational identification of Drosophila enhancers based on the clustering of regulatory elements. Nevertheless, it is still not a perfect code (Markstein, 2004).

Two of the seven composite clusters are likely to be false-positives: they are associated with genes that are not known to exhibit localized expression across the dorsoventral axis. It is possible that the order, spacing and/or orientation of the identified binding sites accounts for the distinction between authentic enhancers and false-positive clusters. For example, there is tight linkage of Dorsal and Twist sites in each of the five neurogenic enhancers. This linkage might reflect Dorsal-Twist protein-protein interactions that promote their cooperative binding and synergistic activities. Previous studies identified particularly strong interactions between Dorsal and Twist-Daughterless (Da) heterodimers. Da is ubiquitously expressed in the early embryo and is related to the E12/E47 bHLH proteins in mammals. Dorsal-Twist linkage is not seen in one of the two false-positive binding clusters (Markstein, 2004).

The regulatory model defined by this study probably fails to identify all enhancers responsive to intermediate levels of the Dorsal gradient. There are at least 30 Dorsal target enhancers in the Drosophila genome, and it is possible that 10 respond to intermediate levels of the Dorsal gradient. Thus, half of all such target enhancers might have been missed. Perhaps the present study defined just one of several 'codes' for neurogenic gene expression (Markstein, 2004).

The possibility of multiple codes is suggested by the different contributions of the same regulatory elements to the activities of the vnd and brk enhancers. Mutations in the CTGWCCY motifs nearly abolish the activity of the brk enhancer, but have virtually no effect on the vnd enhancer. Future studies will determine whether there are distinct codes for Dorsal target enhancers that respond to either high or low levels of the Dorsal gradient. Indeed, it is somewhat surprising that the sog and CG12443 enhancers essentially lack Twist, Su(H) and CTGWCCY motifs, even though they direct lateral stripes of gene expression that are quite similar (albeit broader) to those seen for the rho, vnd and brk enhancers (Markstein, 2004).

This study provides direct evidence that Twist and Su(H) are essential for the specification of the neurogenic ectoderm in early embryos. The Twist protein is transiently expressed at low levels in ventral regions of the neurogenic ectoderm. SELEX assays indicate that Twist binds the CACATGT motif quite well. The presence of this motif in the vnd, brk and sim enhancers, and the fact that it functions as an essential element in the vnd and brk enhancers, strongly suggests that Twist is not a dedicated mesoderm determinant, but that it is also required for the differentiation of the neurogenic ectoderm. However, it is currently unclear whether the CACATGT motif binds Twist-Twist homodimers, Twist-Da heterodimers or additional bHLH complexes in vivo. Su(H) is the sequence-specific transcriptional effector of Notch signaling. The restricted activation of sim expression within the mesectoderm depends on Notch signaling; however, the rho, vnd and brk enhancers direct expression in more lateral regions where Notch signaling has not been demonstrated. Nonetheless, mutations in the two Su(H) sites contained in the brk enhancer cause a severe impairment in its activity. This observation raises the possibility that Su(H) can function as an activator, at least in certain contexts, in the absence of an obvious Notch signal (Markstein, 2004).

The Dorsal gradient produces three distinct patterns of gene expression within the presumptive neurogenic ectoderm. It is proposed that these patterns arise from the differential usage of the Su(H) and Dorsal activators. Enhancers that direct progressively broader patterns of expression become increasingly more dependent on Dorsal and less dependent on Su(H). The sog and CG12443 enhancers mediate expression in both ventral and dorsal regions of the neurogenic ectoderm, and contain several optimal Dorsal sites but no Su(H) sites. By contrast, the sim enhancer is active only in the ventral-most regions of the neurogenic ectoderm, and contains just one high-affinity Dorsal site but five optimal Su(H) sites. The reliance of sim on Dorsal might be atypical for genes expressed in the mesectoderm. For example, the m8 gene within the Enhancer of split complex may be regulated solely by Su(H). The Anopheles sim enhancer might represent an intermediate between the Drosophila sim and m8 enhancers, since it contains optimal Su(H) sites but only one weak Dorsal site. This trend may reflect an evolutionary conversion of Su(H) sites to Dorsal sites, and the concomitant use of the Dorsal gradient to specify different neurogenic cell types. A testable prediction of this model is that basal arthropods use Dorsal solely for the specification of the mesoderm and Su(H) for the patterning of the ventral neurogenic ectoderm (Markstein, 2004).

The spatial and temporal control of gene expression during the development of multicellular organisms is regulated to a large degree by cell-cell signaling. A simple mechanism has been uncovered through which Dpp, a TGFß/BMP superfamily member in Drosophila, represses many key developmental genes in different tissues. A short DNA sequence, a Dpp-dependent silencer element, is sufficient to confer repression of gene transcription upon Dpp receptor activation and nuclear translocation of Mad and Medea. Transcriptional repression does not require the cooperative action of cell type-specific transcription factors but relies solely on the capacity of the silencer element to interact with Mad and Medea and to subsequently recruit the zinc finger-containing repressor protein Schnurri. These findings demonstrate how the Dpp pathway can repress key targets in a simple and tissue-unrestricted manner in vivo and hence provide a paradigm for the inherent capacity of a signaling system to repress transcription upon pathway activation (Pyrowolakis, 2004).

One of the primary events controlled by the Dpp morphogen gradient during growth and patterning of imaginal discs is the establishment of an inverse gradient of brk expression. brk expression is controlled by two opposing activities, a ubiquitous enhancer and a Dpp-dependent silencer. The minimal requirements for a functional silencer complex, both at the DNA and at the protein level, have been determined. Importantly, it has been demonstrated that the minimal element functions in vivo when assayed in the vicinity of a strong enhancer (the brk enhancer) or when present in a single copy in chimeric transgenes (brk enhancer-bamSE fusions) or from within an endogenous gene (gsb-enhancer lacZ fusions). The minimal functional silencer contains a distinct, single binding site for each of the two signal mediators, Mad and Med. Med binds to a GTCTG site, previously recognized as a high-affinity site for Smad binding. Mad binds to a different, GC-rich sequence. Upon binding of Mad and Med, the zinc finger protein Shn is recruited to the protein-DNA complex, bringing along a highly effective repression domain. Although ShnCT contains three essential zinc fingers, it does not bind the silencer element in the absence of Mad and Med. These data suggest that even in the triple protein complex, Shn might bind DNA with moderate sequence specificity, since only a single nucleotide position was identified that is essential for Shn recruitment. However, a number of other cis-regulatory elements that bind Mad and Med (derived from the vestigial, labial tinman, and ubx genes failed to recruit Shn, demonstrating the exquisite selectivity of the element defined in this study (Pyrowolakis, 2004).

Part of this selectivity is accounted for by the specific spacing and orientation of the Mad and Med binding sites in the silencer. Deletion and insertion of single base pairs between the two sites abolish Shn recruitment in vitro and Dpp-dependent repression in vivo, although such alterations still allow the efficient formation of a Mad/Med complex. These findings suggest that Shn recruitment requires a specific steric positioning of amino acid residues in the Smad signal mediators. Strikingly, GTCTG- and GC-rich elements were also found to be crucial for the activation of the Id gene by BMP signaling, but in this case the spacing between the GTCTG- and the GC-rich sites is much larger, and additional factors might be involved in the signal-dependent activation of the Id gene. A more recent study also links these two elements to transcriptional activation of the BMP4 synexpression group in Xenopus. It is tempting to speculate that simple sequence elements similar to the one identified here in several Drosophila genes might be involved in the repression of genes by BMP signaling. Interestingly, human Smad1/5 and Smad4 do form a complex with ShnCT on the Drosophila silencer element from brk; however, a mammalian protein sharing clear homology with Shn in the C-terminal three zinc fingers has not been identified (Pyrowolakis, 2004).

The Dpp-dependent SE allows cells in the developing organism to read out the state of the Dpp signaling pathway. This readout is relatively straightforward because the SE participates in a single switch decision, that is, either to repress (bind Mad/Med and recruit Shn along with its repression domain) or not to repress (not bind Mad/Med, thus failing to recruit Shn). This decision is critically dependent upon one major parameter: the amount of available nuclear Smad complex. For the SE to be functional in vivo, it only needs to interact with a Mad/Med heteromer in those regions of the genome that are actively transcribed; genes that are not active in a given tissue do not need to be repressed by Dpp signaling. This might be one of the main characteristics explaining why such a simple sequence element can have operator-like function in vivo; the element only needs to be recognized by the relevant trans-acting factors in open and active chromatin regions (Pyrowolakis, 2004).

A minimal Dpp-dependent silencer element derived from the brk gene has been identified and demonstrated to function in vivo in a single copy. Its interaction with relevant trans-acting factors have been identified. Based on the results of this analysis, it was possible to derive a consensus sequence, GRCGNCN(5)GTCTG, which allowed scanning of the entire Drosophila genome for potential additional elements. Approximately 350 sites were identified, that, when assayed using transgenic approaches in vivo or in cell culture, should function in a manner analogous to the SEs isolated from the brk regulatory region. Strikingly, and likely significantly, in silico search revealed that the brk gene contains a total of ten SEs, three of them in regions that have been shown to respond to Dpp-dependent repression. Since brk transcription responds to (or can respond to) Dpp signaling in all tissues examined so far, brk might require a SE in the vicinity of each of the different enhancers driving expression in distinct tissues. Alternatively, the readout of the Dpp morphogen gradient might require several SEs, each contributing to the graded repression by Dpp signaling (Pyrowolakis, 2004).

Interestingly, subsequent analysis of two genes containing such Dpp-dependent SEs has demonstrated that these elements function in these transcription units the same way as they do in the brk regulatory region. Therefore, the same molecular principle underlies morphogen readout (brk repression), germline stem cell maintenance (bam repression), and restriction of gene expression to the ventral side of the developing embryo (gsb repression). When the SEs from these three genes are aligned, all the parameters determined to be important for complex formation and for repression are conserved; at all other positions, different base pairs were found in different SEs. In addition, several genes harboring silencer elements are expressed in the wing imaginal disc in a pattern similar to brk or are known to be repressed by Dpp signaling. In contrast, SEs were not found in the vicinity of enhancers known to be activated by Dpp signaling (Pyrowolakis, 2004).

Clearly, these findings implicate that Dpp-induced, Shn-dependent repression via SE elements is a key aspect of development. The readout of the brk gradient contributes to growth and patterning of appendages, and the repression of bam in the germline is essential for the maintenance of germline stem cells. To what extent the repression of gsb contributes to proper cell fate determination along the dorsoventral axis will have to be determined by rescuing the gsb phenotype with a transgene lacking the gsbSE. However, it has been observed that wingless (wg) expression expands from ventral positions to the dorsal side in shn mutant embryos. Since gsb activates wg transcription, the expansion of gsb (in the absence of the gsbSE) possibly leads to the expansion of wg and subsequently to the alteration of dorsoventral cell fate assignments (Pyrowolakis, 2004).

It is important to note that genes repressed by a signaling pathway will not easily be identified in genetic screens because the loss-of-signaling phenotype does not correspond to the loss-of-function phenotype of a repressed gene; in the absence of the signal, such genes are ectopically expressed, leading to a locally restricted gain-of-function phenotype of the corresponding gene. Moreover, since these specific, local patterns of misexpression are likely to result in different phenotypes than widespread overexpression would, simple gain-of-function screens for candidate targets of signal-mediated repression are unlikely to offer straightforward results. Since the target sequence of Dpp/Shn-mediated repression have been identified, the genome can now be scanned and potential target genes can be identified by expression studies and enhancer dissection. It is likely that additional Dpp-repressed genes will be identified using this approach, and this will allow the painting of a much clearer picture of the gene network controlled by Dpp signaling (Pyrowolakis, 2004).

Only a few cases of signal-induced repression have been studied at the molecular level. In most of these cases, repression relies on cooperative action of cell type-specific transcription factors with nuclear signal mediators. The DNA elements that have been demonstrated to mediate repression of particular genes have not been demonstrated to be important for the regulation of other genes, and genome-wide identification of potential target genes using a bioinformatic approach might therefore be difficult, if not impossible (Pyrowolakis, 2004).

Patterning along developing body axes is regulated by gradients of transcription factors, which activate or repress different genes above distinct thresholds. Understanding differential threshold responses requires knowledge of how these factors regulate transcription. In the Drosophila wing, expression of genes such as omb and sal along the anteroposterior axis is restricted by lateral-to-medial gradients of the transcriptional repressor Brinker (Brk). omb is less sensitive to repression by Brk than sal and is consequently expressed more laterally. Contrary to previous suggestions, it has been shown that Brk cannot repress simply by competing with activators, but requires specific repression domains along with its DNA-binding domain. Brk possesses at least three repression domains, but these are not equivalent; one, 3R, is sufficient to repress omb but not sal. Thus, although sal and omb show quantitative differences in their response to Brk, there are qualitative differences in the mechanisms that Brk uses to repress them (Winter, 2004).

The simplest method of transcriptional repression involves competition with an activator, and can operate at the level of DNA if the activator and the repressor have the same, or overlapping, binding sites in an enhancer. In theory, assuming a transcription factor is nuclear, it should only require a DNA-binding domain to act in this fashion. Brk has been shown to possess an N-terminal sequence-specific DNA-binding domain (DBD), and this study has identified several mutations in this domain that either completely inactivate or reduce the activity of the protein, indicating that this region is essential for Brk activity (Winter, 2004).

Previous studies suggested that Brk could function by competition, more specifically, by competing with Mad for overlapping binding sites in vitro. However, a nuclear localized Brk protein consisting primarily of the DBD, Brk^NLS, cannot repress any Brk target in vivo, including the embryonic UbxB reporter, which has been shown to possess overlapping Brk and Mad binding sites that Brk and Mad can compete for in vitro. It is possible that Brk^NLS cannot bind to DNA in vivo. However, a modified protein, Brk^NLSW, which is identical to Brk^NLS apart from the addition of the four amino acids WRPW that recruit the co-repressor Gro, can repress targets, indicating that Brk^NLS should also be capable of binding to these targets in vivo (Winter, 2004).

Competition has been proposed as a mechanism for many transcriptional repressors. However, direct in vivo support for or against such proposals is rare, at least of the sort presented here, i.e., testing, in vivo, the ability of a protein consisting largely of a functional DBD, which has access to the nucleus, to repress a target for which there is in vitro evidence for overlapping binding sites with an activator. There is some in vivo evidence that the Drosophila embryonic repressor Kruppel can repress a synthetic enhancer containing overlapping binding sites with the activators Dorsal and Bicoid. However, although this repression is CtBP-independent, and further studies are required to rule out additional domains outside of the DBD being required in a similar fashion to the 3R domain in Brk. The paucity of good examples of binding-site competition in vivo in eukaryotes is in stark contrast to that in prokaryotes, and raises the question of how common this phenomenon really is in eukaryotes (Winter, 2004).

If Brk cannot repress by competition it must possess repression domains/motifs, and previous studies identified interaction motifs for the co-repressors CtBP and Gro (CiM and GiM). However, repression of at least one Brk target, omb, has been shown shown not to require CtBP or Gro. This is consistent with the demonstration that the protein produced by the endogenous mutant brk^F138, which is truncated before the CiM and GiM, can still repress omb. Truncated proteins that lack the CiM and GiM, Brk^Stop1, Brk^EC and Brk^A, can also repress omb, but only if they contain a specific region between the DBD and CiM that has been classified as a third repression domain, 3R. Further studies are required to determine if 3R is a true autonomous repression domain, i.e., if it can function outside of Brk, or if it is more specific (for example, antagonizing activators such as Mad), and to determine what its specific properties are (for example, how close do Brk sites have to be to activator sites for 3R to be effective?) (Winter, 2004).

The three repression domains/motifs of Brk are not equivalent. Wild-type Brk and proteins containing only a GiM, Brk^NLSW, or only a CiM, Brk^Stop1NAC, can repress both sal and omb, and they are more effective at repressing sal than omb. Analysis of gro and CtBP single and double mutant clones reveals that Gro is required for normal repression of sal in wing discs, and that CtBP can provide some, but not always complete, activity for the repression of sal in the absence of Gro. By contrast, Gro and CtBP are not required for repression of omb (Winter, 2004).

The 3R domain (the region between the DBD and the CiM) is sufficient for Brk to repress omb and the UbxB enhancer in embryos, but is deficient for the repression of sal. Furthermore, misexpression of proteins possessing only the 3R domain (plus the DBD) are much more effective at repressing omb than sal, i.e., the converse of wild-type Brk or Brk possessing only a GiM or a CiM. Although some results suggested that 3R may confer a limited ability to repress sal, this is probably indirect, because a previous study demonstrated that sal requires Omb to be expressed, and if omb is repressed directly, sal will be lost also. However, the possibility cannot be ruled out that high levels of proteins possessing only the 3R domain can repress sal directly (Winter, 2004).

Contradictory results were obtained regarding the ability of 3R to repress the vg-QE. Expression of the vg-QE did show expansion in some brk^F138 clones, indicating that the truncated protein produced in this mutant (which only has the 3R repression domain) cannot efficiently repress this enhancer. However, similar in vitro truncated proteins, such as Brk^Stop1, could efficiently repress vg-QE expression when misexpressed using the UAS/Gal4 system. Such a difference could simply be a reflection of the high levels of expression achieved with the Gal4/UAS system, and that, at physiological levels, the 3R domain is not sufficient for complete repression of the vg-QE (Winter, 2004).

Whether a single repression domain is sufficient for Brk to repress a particular target may depend upon the positioning of Brk sites in relation to activator sites (or possibly the promoter) at that target. The UbxB reporter has overlapping Brk and activator (Mad) sites. Analysis of an omb enhancer revealed that an important Brk site may also overlap with an activator. Conversely, analysis of the cis-regulatory elements of the sal gene indicate that activator and Brk sites are separated. Proteins possessing only 3R can repress UbxB and omb, but not sal, suggesting that 3R may only be sufficient for the repression of genes in which the Brk sites are situated very close to activator sites (Winter, 2004).

Why does Brk possess at least three, probably four, independent repression domains/motifs? There are two obvious answers: qualitative, different repression domains/motifs are required for the repression of different targets; quantitative, more domains/motifs provide greater repressor activity. Other transcription factors have multiple repression domains and there is evidence that they have these for either qualitative or quantitative reasons, and, in some cases, both. For example, in the Drosophila embryo, the pair-rule protein Runt requires Gro for the repression of one stripe of the pair-rule genes, even skipped (eve) and hairy, but not for the repression of engrailed. The gap protein Knirps represses different stripes of eve; for stripes 4 and 6 it requires CtBP, but for stripes 3 and 7, it does not. However, this appears to be a quantitative difference, because increasing the levels of Knirps allows it to repress stripes 4 and 6 even in the absence of CtBP. Similarly, Gro appears to increase the repressor activity of the Eve protein (Winter, 2004).

There is some difference in the ability of the three domains/motifs in Brk to repress different targets. For example, the 3R domain is sufficient for the normal repression of omb but not sal. However, either the CiM or GiM appear to be sufficient for the repression of both sal and omb, so why does Brk need the 3R domain? In the absence of Gro and CtBP, the Brk protein appears fully active in its ability to repress omb, and recruiting Gro and CtBP does not seem to increase its activity towards omb; otherwise, the width of the omb domain would be expected to shift in brk^F138 mutant cells, which have no CiM or GiM, or in CtBP gro double mutant cells, but Brk protein does not. It is possible that, in regard to omb, the 3R domain is more efficient than either of the other two and provides Brk with sufficient activity to establish the omb domain in the correct position (Winter, 2004).

Brk needs to recruit either CtBP or Gro for the repression of some targets, including sal and brk itself, or just Gro for some others, including the vg-QE. Consequently, why does Brk need to recruit CtBP? Mutation of the CiM alone, in common with mutation or deletion of just the GiM and 3R, does reduce activity of Brk, as judged by its effect when misexpressed. However, there is no evidence that CtBP is required specifically for the repression of any Brk target in the wing, because CtBP mutant clones have no effect on the expression of any known Brk target in the wing. The CtBP and Gro motifs in Brk have been conserved over millions of years, and thus, recruiting CtBP is presumably important for Brk activity. It is possible that CtBP is required outside of the wing -- for example in the embryo -- or for some other, as yet, uncharacterized targets in the wing (Winter, 2004).

Recruiting both CtBP and Gro does appear to be a little illogical from what is known about their basic properties, CtBP acting only over a short range, while Gro acts over much longer ranges. It might be assumed that different transcription factors would use either Gro or CtBP, because the primary advantage of recruiting CtBP is that it would allow a transcription factor to repress one enhancer without disrupting the activity of one nearby, which would be repressed if Gro was recruited, although this simple model does not always hold. Consequently, most transcription factors do recruit only one of these co-repressors. However, there are two other exceptions, Hairy and Hairless. In Hairy it appears that CtBP may actually be functioning to antagonize Gro activity and not in its standard role as a co-repressor. There is no evidence that it does this in Brk, where it can provide repressor activity. For Hairless, there is genetic evidence that both CtBP and Gro provide repressor activity to the protein, although it is not clear if CtBP is required to increase the general activity of Hairless, or for repression of specific targets that cannot be repressed adequately by Gro (Winter, 2004).

With the exception of the brk^F138 mutant, this analysis has been limited to analyzing the effects of misexpressing modified Brk proteins in positions where the endogenous protein is not found. Consequently, further insights into the precise roles of individual repression domains will require replacing the endogenous gene with one in which only one or two repressions domain have been mutated or deleted (Winter, 2004).

To conclude, it is often assumed that the sensitivity of one enhancer to a transcription factor compared with that of another enhancer is based largely upon the number or the affinity of the binding sites for that transcription factor in each enhancer. However, other factors are also important; for example, the ability of the Giant transcription factor to repress a promoter is related to how closely it binds. This study has shown that the two best characterized outputs of the Dpp morphogen gradient, sal and omb, appear to be regulated differently by Brk. Consequently, simply counting binding sites and measuring their affinity will not reveal why one is more sensitive to Brk than the other, and it is necessary to factor in what specific repressive mechanisms are being used, and the relative efficiencies of each (Winter, 2004).

Smad proteins regulate transcription in response to transforming growth factor-β signaling pathways by binding to two distinct types of DNA sites. The sequence GTCT is recognized by all receptor-activated Smads and by Smad4. The subset of Smads that responds to bone morphogenetic protein signaling recognizes a distinct class of GC-rich sites in addition to GTCT. Recent work has shown that Drosophila Mad protein, the homologue of bone morphogenetic protein rSmads, binds to GRCGNC sites through the same MH1 domain β-hairpin interface used to contact GTCT sites. However, binding to GRCGNC requires base-specific contact by two Mad proteins, and this study provides evidence that this is achieved by contact of the two Mad subunits that overlap across the two central base pairs of the site. This topology is supported by results indicating that His-93, which is located at the tip of the Mad β-hairpin, is in close proximity to base pairs 2 and 5. Also consistent with the model is disruption of binding by mutation of Glu-39 and Glu-40, which are predicted to lie at the interface of the two overlapping Mad MH1 domains. As predicted from the overlapping model, binding is disrupted by insertion of 1 bp in the middle of the site, whereas insertion of 2 bp creates abutting sites that can be bound by the Mad-Medea heterotrimer without requiring Glu-39 and Glu-40. Overlapping Mad sites predominate in Decapentaplegic response elements, consistent with a high degree of specificity in response to signaling (Gao, 2006).

DNA contact by Smad proteins has been shown to play an important role in many instances of target regulation by TGFβ pathways. For the consensus Smad3/Smad4-binding site, GTCT can also be bound by Mad- and BMP-specific Smad1, but Smad3 does not bind to GC-rich Mad/Smad1-binding sites, leaving open the question of whether such sites are contacted by a different mechanism. Recent work had shown that Mad-binding sites within the brk and bam silencers are bound by two Mad subunits and that in each case two Mad MH1 domains simultaneously contact a single 6-bp site using the same three β-hairpin residues that are responsible for base-specific contact by Smad3. By using mutational analysis and directly measuring binding, this study provides evidence that two Mad MH1 domains bind to the 6-bp site by overlapping across the two central base pairs. Smad1-binding sites match this 6-bp motif, a likely indication that overlap is also a feature of BMP-response elements (Gao, 2006).

The overlapping structure of Mad sites explains the seeming discrepancy between the Mad consensus and that of Smad3/Smad4. Smad3 differs from Mad at two positions that influence binding to brkS. Arg-58 at the C terminus of helix 2 is absolutely essential for binding to brkS; in Smad3 this position is a threonine, whereas the adjacent Lys-59 of Mad is absent in Smad3. Glu-39 at the N terminus of helix 2 contributes substantially to binding affinity for brkS, and Smad3 has instead a glutamine at this position. In addition, the loop between helices 1 and 2 is three residues shorter in Mad than in Smad3, a difference that modeling suggests will affect the structure of the alpha-carbon backbone and side chains near the N terminus of helix 2. Each of these differences is conserved between Smad3 and Smad2 and between Mad and the vertebrate BMP-specific rSmads (Gao, 2006).

Mutational analysis indicated the optimal sequence for an overlapping Mad site is GGCGCC, meaning each Mad MH1 prefers GGCG in the context of overlap. However, even when the two Mad sites are spaced such that they do not overlap, GGCG is still bound by Mad with about the same affinity as GTCT. The structural basis for this compatibility with two distinct sites remains to be determined, but the differential effects of helix 2 alanine substitutions suggest distinct docking geometries. Individual GGCG motifs occur in Dpp and BMP-response elements, and the results indicate that these are likely sites for contact by Mad and Smad1 (Gao, 2006).

Although the natural brkS element was specific for the Mad-Medea heterotrimer, changing the Mad site to abutting SBEs allowed binding by Mad alone or by Medea alone. The ability of such a site to be bound by Medea oligomers (putatively homotrimers) without Dpp signaling seemingly would make it ill-suited to function as a Dpp-response element, although signaling-induced activation was observed by reporter analysis (perhaps an indication that Medea alone is a poor activator). However, the brkS derivative with abutting GGCG sites (i.e. GGCGCGCC) shows little or no Medea binding in the absence of active Mad, is able to recruit Shn, and causes repression in response to signaling. Similar sites in the Dpp-response element of Race (GACGCGAC), which does not respond to repression by Brk protein, and in a BMP-response element of Smad7 (GGCGCGCC) appear to be examples of functional nonoverlapping Mad/Smad1 sites. In Drosophila a potentially significant difference between overlapping and nonoverlapping Mad sites is that the overlapping motif allows for competitive binding by the Brinker protein and thus dual control of Dpp targets, whereas the nonoverlapping motif does not. This may account for the predominance of overlapping Mad sites in Drosophila. The predominance of overlapping sites in BMP-response elements may reflect specificity for Smad1 but not Smad3 (Gao, 2006).