Interactive Fly, Drosophila

Targets of DPP in mesoderm and endoderm: segmentation of the midgut

Ultrabithorax is expressed in parasegment 7 of the visceral mesoderm. The central constriction of the midgut forms at the junction between parasegments 7 and 8. dpp expression is activated in the visceral mesoderm by UBX. wingless is induced in parasegment 8 by the joint action of dpp (from parasegment 7) and local action of abdominal-A. DPP (PS7) and WG (PS8) diffuse into the endoderm. In this way a segmented mesoderm brings about local differentiation of an initially unsegmented endoderm. Thus Ubx, abd-A, dpp and wg are all required for the formation of the central midgut constriction (Manak, 1994).

DPP receptors in the visceral mesoderm convey dpp inductive signals from the dorsal ectoderm to mesoderm. Midway through embryogenesis, dpp is expressed in the visceral mesoderm, and enhances the expression of labial in the underlying midgut endoderm. Earlier in development, however, dpp expression is limited to the dorsal ectoderm. At this stage in development ectodermal dpp might not only be required for development of dorsal ectoderm, but also act inductively to mediate pattern formation in the underlying mesoderm. Expressing dpp ectopically in the ectoderm and mesoderm and examining dpp null mutant embryos verified that dpp indirectly regulates expression of mesodermal genes (Staehling-Hampton, 1994). Mutations affecting DPP receptors block the expression of two DPP-responsive genes, dpp and labial, in the embryonic midgut (Penton, 1994).

Hox genes have large expression domains, yet these genes control the formation of fine pattern elements at specific locations. The mechanism underlying subdivision of the abdominal-A (abdA) Hox domain in the visceral mesoderm has been examined. AbdA directs formation of an embryonic midgut constriction at a precise location within the broad and uniform abdA expression domain. The constriction divides the abdA domain of the midgut into two chambers, the anterior one producing the Pointed (Pnt) ETS transcription factors and the posterior one the Odd-paired (Opa) zinc finger protein. Transcription of both pnt and opa is activated by abdA. Near the anterior limit of the abdA domain, two signals, Decapentaplegic and Wingless, are produced, in adjacent non-overlapping patterns, under Hox control in mesoderm cells. AbdA is proposed to activate three targets, in distinct subsets of its broad domain of expression: wg at the anterior boundary of Connectin (Con) patch 7; pnt from anterior Con patch 7 to anterior Con patch 8, and opa, from anterior Con patch 8 through Con patch 11. Dpp signaling plays a central role in setting these distinct expression domains. The initial activation of wg by AbdA requires dpp. opa is activated in all abdA-expressing cells that do not receive a Dpp signal, defining the site of the posterior constriction. wg, in collaboration with abdA, activates pnt to generate the appropriate number of cells in the third midgut chamber, positioning the posterior constriction at the proper distance from the central constriction and partitioning the posterior midgut appropriately. Fine patterning of the posterior midgut is achieved by the activity of diffusible signals emanating from the central midgut, a remarkably long-range organizing effect (Bilder, 1998a).

In the embryonic midgut both Decapentaplegic (Dpp) and Wingless (Wg) signaling pathways control the subcellular localization of Extradenticle protein. Exd protein is predominantly nuclear in endoderm cells close to the Dpp-and Wg-secreting cells of the visceral mesoderm, but is found in the cytoplasm in more distant endoderm cells. Both dpp and wg are required for the nuclear localization of Exd in the endoderm; ectopic expression of dpp and wg expands the domain of nuclear Exd (Mann, 1996). The requirement of homothorax for Exd's nuclear localization is apparent in many embryonic tissues, including the ectoderm, visceral mesoderm, and endoderm. This requirement is observed in cells where the signaling molecules Wg and Dpp contribute to Exd's nuclear translocation (e.g., the endoderm). It is suggested that Wg and Dpp both regulate expression of hth in the domains in which Exd nuclear function is required (Rieckhof, 1997).

teashirt is necessary for proper formation of anterior and central midgut structures. Antp activates tsh in anterior midgut mesoderm. In the central midgut mesoderm Ubx, abd-A, dpp, and wg are required for proper tsh expression. The control of tsh by Ubx and abd-A, and probably also by Antp, is mediated by secreted signaling molecules. By responding to signals as well as localized transcription regulators, the TSH transcription factor is produced in a spatial pattern distinct from any of the homeotic genes (Mathies, 1994). odd paired is positively regulated by Antennapedia and Abdominal-A at the location of the first and third midgut constructions respectively. Between these domains opa is negatively regulated by Ultrabithorax and Decapentaplegic (Cimbara, 1995).

In Drosophila, mothers against dpp is specifically required in cells responding to Decapentaplegic (Dpp) signals. The role of Mad in Dpp-mediated signaling was examined by utilizing tkvQ199D, an activated form of the Dpp type I receptor serine-threonine kinase thick veins (tkv). In the midgut, dpp is expressed in the visceral mesoderm of parasegments 3 and 7. In response to Dpp signals, cells expressing dpp in parasegment 3 repress the expression of the homeotic gene Sex combs reduced. Dpp signals are also required to maintain dpp expression in parasegment 3 through an autocrine feedback loop. However, cells in parasegment 4 do not appear to be affected by Dpp signals; Scr is expressed while dpp is not. In ps7, the homeotic gene Ultrabithorax initiates dpp expression. Subsequently, Dpp functions in an autocrine manner to maintain Ubx and thus dpp expression. In ps7, Dpp also signals between germ layers to the underlying endoderm. Within the midgut endoderm, which does not express dpp, expression of the homeotic gene labial is dependent on Dpp signals (Newfeld, 1997b).

In the embryonic midgut, tkvQ199D mimics Dpp-mediated inductive interactions. There is an anterior expansion of labial midgut endoderm expression in response to ubiquitously expressed tkvQ199D. In early stage tkvQ199D embryos, dpp expressin is expanded to include ps4, ps5 and ps6. In late stage tkvQ199D embryos, the expanded domain of expression is maintained at very high levels, while in late stage Mad mull tkvQ199D embryos, this is not observed. Analysis of Scr expression in Mad null embryos, combined with tkvQ199D, reveals an anterior expansion of Scr, showing that Mad and dpp are required for repressing Scr. Mad function is epistatic to tkvQ199D in the repression of Scr. Thus homozygous Mad mutations block signaling by tkvQ199D and appropriate responses to signaling by tkvQ199D are restored by expression of MAD protein in Dpp-target cells (Newfeld, 1997b).

Extracellular signals can act at different threshold levels to elicit distinct transcriptional and cellular responses. The transcriptional regulation of the Wingless target gene Ultrabithorax has been examined in the embryonic midgut of Drosophila. Ubx transcription is stimulated in this tissue by Dpp and by low levels of Wingless signaling. High levels of Wingless signaling can repress Ubx transcription. The response sequence within the Ubx midgut enhancer required for this repression coincides with a motif required for transcriptional stimulation by Dpp, namely a tandem array of binding sites for the Dpp-tranducing protein, Mad. Indeed, Wingless-mediated repression depends on low levels of Dpp, although apparently not on Mad itself. In contrast, high levels of Dpp signaling antagonize Wingless-mediated repression. This suggests that transcriptional activation of Ubx is subject to competition between Dpp-activated Mad and another Smad whose function as a transcriptional repressor depends on high Wg signaling. Wingless can repress its own expression via an autorepressive feedback loop that results in a change of the Wingless signaling profile during development (Yu, 1998).

Dpp and Wg signaling synergize in the visceral mesoderm to stimulate Ubx transcription, targeting distinct, albeit adjacent, response sequences in the Ubx midgut enhancer. Therefore, efficient stimulation of Ubx transcription by Wg depends on dpp. Wg-mediated repression also depends on dpp, but, remarkably in this case, the response sequence for Wg-mediated repression within the Ubx enhancer coincides with that for Dpp-mediated stimulation. Indeed, the WRS-R/DRS (Wingless response sequence mediating repression and Dpp response sequence) functions in two antipodal responses: it mediates efficient transcriptional stimulation when the signaling levels of Dpp are high and those of Wg are low, but it is also required for transcriptional repression when the Wg signaling levels are high and those of Dpp are low. This raises the possibility that the same factor may confer the two antipodal responses. However, this is unlikely to be the case since Mad itself, which binds to the DRS to mediate the positive response to Dpp, is apparently not required for the Wg-mediated repression (Yu, 1998).

Thus it is proposed that the two antipodal responses are conferred by two distinct factors: by Mad and by a hypothetical protein WR. It is further proposed that WR is a Mad-related protein, i.e. a Smad, since WR acts through Mad-binding sites and since its function as a repressor depends on dpp. It is envisaged that WR, like Mad itself and other Smads, is activated by Dpp signaling through phosphorylation by ligand-bound membrane receptors, an event that promotes their subsequent translocation to the nucleus. In this scenario, Dpp enables WR (which also needs to be activated by high Wg signaling) to occupy the Mad-binding sites within the Ubx enhancer. Once bound to this enhancer, WR dominantly represses Ubx transcription, overriding the activating function of Arm-Pangolin and other transcriptional activators bound to the same enhancer (Yu, 1998).

How is WR's repressor function activated by high Wg levels? It is presumed that high Wg signaling regulates, directly or indirectly, the availability of WR as an enhancer-binding protein: either high Wg signaling controls a post-transcriptional event (e.g. it may promote WR's association with Armadillo, or WR's translocation into the nucleus), or it simply activates transcription of the WR gene. The latter possibility of indirect regulation, which involves transcriptional coupling, is favored because it accomodates readily the dependence of Wg-mediated repression on arm and Pangolin. Whatever the case, it is emphasized that high Wg signaling controls the activity of the protein WR (possibly a Smad), which also requires Dpp signaling. Thus, WR is a common target for two signaling pathways and represents a point of convergence between them (Yu, 1998).

This model readily explains how high Dpp levels antagonize WR, namely by promoting maximal levels of nuclear Mad which now competes with WR for binding to the Ubx enhancer. The outcome of this competition is the transcriptional activation or repression of target genes, depending on the prevalence of Mad or WR. This may illustrate a general principle, namely that the response sequence for the positive effect of one signal is also the response sequence for the negative effect of an antagonistic signal. Such a layout provides a sharp flipping of the response from positive to negative in an area where cells are experiencing increasingly more of one signal and increasingly less of the antagonizing one (Yu, 1998).

Medea is the Smad4 homolog that is known to be the common oligomerization partner for pathway-specific Smads. Furthermore, Medea binds to the same DNA sequences as Mad. This raises the possibility that Medea is an oligomerization partner of WR: while Medea, together with Mad, is expected to activate transcription, together with WR it may repress transcription. A precedent for this scenario is the Myc/Mad/Max system, in which Mad (a bHLH protein that happens to have the same name as the Dpp transducer Mad) is a common dimerization partner for either Myc, a transcriptional activator, or Max, a transcriptional repressor. In addition to antagonism, there is also synergy between Wg and Dpp in the embryonic midgut. This synergy apparently results from cooperation between the nuclear target factors activated by the two signals, i.e. between Arm-Pangolin and Mad/CRE-binding proteins. Other examples of apparent synergy between Wg and Dpp are the leg and wing imaginal discs, where these signals act together in central disc regions to stimulate expression of homeobox genes. But the two signals also antagonize each other in leg discs, as well as in eye discs. Although it is conceivable that the developmental context determines the synergy or antagonism between Dpp and Wg, the situation in the midgut suggests that the decisive factor in each case may be the levels of signaling (Yu, 1998 and references).

It is interesting that Wg signaling can repress its own expression when signaling levels reach a critically high level. This indicates a negative feedback loop, which could account for two observations: (1) Wg signaling shifts its own expression towards the anterior over time. It is not known at present whether this shift has any biological significance. (2) Wg has the potential for switching itself off over time. This is actually observed, since Wg expression becomes undetectable by the end of embryogenesis. Clearly, Wg's negative feedback loop is capable of changing the Wg signaling profile as development procedes. There are negative feedback loops for other signaling pathways in Drosophila. For example, the epidermal growth factor (EGF) receptor inhibits itself eventually, after signaling has reached a critical level, by switching on expression of an inhibitory ligand, Argos. In the ovary, this negative feedback loop causes splitting of a single signaling peak into twin peaks. Furthermore, Hedgehog signaling in the eye imaginal disc is repressive at high Hedgehog levels, but stimulatory in cells, further away from the signaling source, which experience lower Hedgehog levels. Perhaps such 'hard-wired' negative feedback loops in signaling pathways are fairly universal, and serve to stop these pathways from escalating out of control. If so, this would be akin to feedback inhibition of metabolic pathways, which provides homeostatic control (Yu, 1998 and references).

Table of contents

The Interactive Fly resides on the
Society for Developmental Biology's Web server.