BIOLOGICAL OVERVIEW

Smads can be grouped into three classes: receptor-regulated (R-) Smads, common (Co-) Smads, and antagonistic (Anti-) Smads. In Drosophila, four Smads have been identified, the prototypic Mad, an R-Smad; Medea, a Co-Smad; Dad, an Anti-Smad and an R-Smad termed Smad on the X (Smox), described here. The activation mechanism of the Smads has been well defined; first the R-Smads are phosphorylated by the type I receptor of the ligand-activated receptor complex; next, they bind to the Co-Smad and enter the nucleus. Finally, this Smad complex induces the transcriptional activation of target genes, including the Anti-Smads, which act as antagonists of signaling. Smad1, 5 and 8 function in the BMP pathways, while Smad2 and 3, mammalian homologs of Smox, mediate activin and TGFbeta signaling. Smad4 acts as a Co-Smad for the three major TGFbeta signaling pathways and forms hetero-oligomers with the R-Smads (Das, 1999 and references therein).

When expressed alone in cultured cells, Baboon (Babo), the Type 1 activin-A-receptor of Drosophila, is unable to bind TGF-beta, activin, or bone morphogenetic protein 2. However, Baboon binds activin efficiently when coexpressed with the distantly related Drosophila activin receptor (Punt), with which it forms a heteromeric complex. Baboon can also bind activin in concert with mammalian activin type II receptors (ActR-II and ActR-IIB). Maternal Baboon transcripts are abundant in the oocyte and widespread during embryo development and in the imaginal discs of the larva (Brummel, 1999). Prior to the identification of Smad on the X (Smox), Mad and Medea, both of which mediate Dpp signaling, were the only activating Smads that had been identified. Because Babo only appears to induce Smad2/3 responsive promoters in mammalian cell culture, the Drosophila ESTs (Berkeley Genome Project) database was searched for new Smad-like genes. One clone with significant homology to vertebrate Smad2/3 was identified. Using this clone as a probe, a Drosophila ovarian cDNA library was screened to obtain the full-length cDNA, which was sequenced and named Smox (Brummel, 1999, Henderson, 1999 and Das, 1999).

The specificity of Smad interactions with receptors and nuclear targets is dictated by the MH2 domain. This suggests that Smox might function in a Drosophila TGF/Activin-like signaling pathway that involves Babo. To investigate this possibility, the ability of Babo to mediate the phosphorylation of Smox was tested. COS-1 cells were transfected with an epitope-tagged Smox either alone or together with wild-type or constitutively active versions of Babo. Analysis of Smox in the absence of signaling shows some basal phosphorylation of the protein. However, on coexpression of activated Babo, a strong increase in phosphorylation of the protein is observed. Previous work has shown that receptor-dependent phosphorylation of Smads occurs on the last two serines of the protein. To confirm that Babo induces phosphorylation of Smox on these residues, the last two serines were mutated to alanines (dSmad2-2SA). Unlike wild-type Smox, this mutant is not phosphorylated by activated Babo. These data suggest that Smox is a downstream target of Babo and is phosphorylated on the last two serine residues in the carboxyl terminus (Brummel, 1999).

One functional consequence of phosphorylating receptor-regulated Smads is the induction of heteromeric complex formation with the common partner Smad4. In Drosophila, the Smad4 homolog Medea similarly associates with Mad and is required for a subset of Dpp signaling. Thus, an investigation was carried to see whether Babo-dependent phosphorylation of Smox might induce association with Medea. In the absence of signaling, Smox and Medea form some heteromeric complexes, however, the level of complex formation is substantially increased on cotransfection with a constitutively active form of Babo. Furthermore, use of dSmad2-2SA abolishes this receptor-dependent increase in heteromeric complex formation. These data indicate that phosphorylation of Smox on the last two serines is necessary for receptor-dependent induction of heteromeric complexes of Smox and Medea (Brummel, 1999).

Because recognition of type I receptors by Smads requires activation by the type II receptor, a test was carried out to see whether Punt, the Drosophila type II receptor can function to activate Babo. In cultured cells, transient transfection of either Punt or Babo alone has no effect on the 3TP promoter. However, overexpression of Punt together with Babo leads to a strong induction of the promoter. This is consistent with previous observations that type II and type I ser/thr kinase receptors have intrinsic affinity for each other and on overexpression can associate and signal in the absence of ligand. To test whether Smox interacts with Punt/Babo complexes, a kinase-deficient Babo, which functions to stabilize Smad-receptor interactions, was used. When Smox is immunoprecipitated in the presence of wild-type receptors, no complexes can be detected. However, when the kinase-deficient form of Babo is utilized, receptor complexes coprecipitating with Smox are readily detected. In addition, when the phosphorylation site mutant of Smox is tested, stable complex formation is detected between the mutant protein and wild-type receptor complexes. Thus, Smox interacts transiently and specifically with Punt-Babo receptor complexes. Taken together, these functional and biochemical analyses strongly suggest that Smox is a Drosophila homolog of Smad2/Smad3 and functions as a downstream signaling component that directly interacts with Babo (Brummel, 1999).

In order to more directly address the question of whether Drosophila has a activin/TGFbeta pathway, functional assays of the Drosophila genes were performed using Xenopus embryos, since these assays have provided an effective method of classifying TGFbeta superfamily ligands and their signal transduction components into either BMP- or activin-type pathways. These two signals have distinct effects during Xenopus development. When introduced as synthetic mRNA, BMP2/4, the constitutively active (CA-) BMP-type I receptors ALK2, CA-ALK3, CA-ALK6, Smad1 and Smad5 all function to ventralize the mesoderm and promote epidermal fates in the ectoderm. Conversely, Activin, CA-ALK4 (an activin/TGFbeta type I receptor), Smad2 and Smad3 convert ectoderm to mesoderm and induce axial duplications. Therefore, molecules suspected of being involved in transducing TGFbeta superfamily signals can be easily placed into either a BMP- or activin-like pathway (Das, 1999).

mRNA encoding Smox was injected into the animal hemisphere of one-cell Xenopus embryos. The embryos were allowed to develop until the blastula stage, at which time the ectodermal cap was removed and cultured in isolation. The ectodermal explants were harvested at either the gastrula stage or late neurula stage and analyzed by RT-PCR for the presence of mesoderm and muscle-specific markers (Das, 1999).

At the gastrula stage, Smox induces the expression of the mesoderm specific marker, Xbra, while at neurula stages Smox is effective at inducing the expression of the muscle specific marker, cardiac actin (c-actin). Previous studies have shown that the activation of c-actin is characteristic of the Smad2, but not the Smad1 pathway. This result demonstrates that Smox is capable of mediating the vertebrate activin-like signal transduction pathway, and suggests that Drosophila possesses activin-like signals (Das, 1999).

Several lines of evidence demonstrate that Smox and Baboon function in an activin-like pathway in Drosophila. (1) Smox shares many sequence characteristics with the vertebrate activin/TGFbeta Smads -- Smad2 and Smad3 -- including the two residues in the L3 loop that are thought to confer pathway specificity on the Smads. (2) Smox can promote activin-like fates, but not BMP-like fates, in Xenopus animal cap assays, suggesting that the normal role of Smox in Drosophila is in an activin-like pathway. (3) Smox becomes phosphorylated and translocates to the nucleus in response to activin-like signaling, and acts to initiate transcription from activin-stimulated promoters. Smox is not phosphorylated by the Dpp receptor Tkv, while Mad, the Dpp R-Smad, is not activated by Baboon, demonstrating the specificity of Smox signal transduction. (4) Medea is required for Smox activation in vitro, a result which would be expected if the BMP and activin pathways had been conserved during evolution (Das, 1999).

While these results strongly suggest the existence of a functional activin-like pathway in Drosophila, the exact nature of its function will have to await the identification of mutants in components of the pathway, including the two newly identified activin-like ligands. However, the prominent expression of Smox and Baboon in the region of LPC mitoses in the brain, and in the morphogenetic furrow of the eye imaginal disk, implies that one of the roles of the Drosophila activin pathway is in transmitting proliferative signals. The expression of dpp in the OPC would suggest that it has a proliferative role in the development of the brain, and since such a role has also been proposed for Mad, it is possible that the dpp and activin pathways have intersecting functions. Additionally, both pathways may also share the patterning roles that have been shown for dpp in the brain. Interestingly both tolloid (tld) and tolkin (tok), BMP1-related metalloproteases that are implicated in the regulation of TGFbeta-like ligands, are expressed in the OPC of the brain. While a direct role for Tld in the regulation of Dpp activity has been shown, no such function is known for Tok. An interesting possibility is that Tok may be involved in the regulation of the Drosophila activin-like ligands. The identification of mutants in components of the pathway will enable the detailed study of these issues (Das, 1999).

R-Smad competition controls activin receptor output in Drosophila

Animals use TGF-β superfamily signal transduction pathways during development and tissue maintenance. The superfamily has traditionally been divided into TGF-β/Activin and BMP branches based on relationships between ligands, receptors, and R-Smads. Several previous reports have shown that, in cell culture systems, 'BMP-specific' Smads can be phosphorylated in response to TGF-β/Activin pathway activation. Using Drosophila cell culture as well as in vivo assays, this study found that Baboon, the Drosophila TGF-β/Activin-specific Type I receptor, can phosphorylate Mad, the BMP-specific R-Smad, in addition to its normal substrate, dSmad2. The Baboon-Mad activation appears direct because it occurs in the absence of canonical BMP Type I receptors. Wing phenotypes generated by Baboon gain-of-function require Mad, and are partially suppressed by over-expression of dSmad2. In the larval wing disc, activated Baboon cell-autonomously causes C-terminal Mad phosphorylation, but only when endogenous dSmad2 protein is depleted. The Baboon-Mad relationship is thus controlled by dSmad2 levels. Elevated P-Mad is seen in several tissues of dSmad2 protein-null mutant larvae, and these levels are normalized in dSmad2; baboon double mutants, indicating that the cross-talk reaction and Smad competition occur with endogenous levels of signaling components in vivo. In addition, it was found that high levels of Activin signaling cause substantial turnover in dSmad2 protein, providing a potential cross-pathway signal-switching mechanism. It is proposed that the dual activity of TGF-β/Activin receptors is an ancient feature, and several ways this activity can modulate TGF-β signaling output are discussed (Peterson, 2012).

This report presents experimental results showing that the Baboon receptor can directly phosphorylate Mad in cell culture and in vivo, and that this cross-talk activity is tightly controlled by the availability of dSmad2. These findings extend the initial report describing canonical signaling between Baboon and dSmad2 where dSmad2, but not Mad, was shown to be a substrate of Baboon in mammalian cells. There are several possible reasons why Baboon-Mad activity is observed in Drosophila cells but was missed in the initial report. First, it is possible that Mad binding to Baboon may be too weak or transient to be detected by immunoprecipitation. Alternatively, endogenous Smad2/3 may have blocked the interaction in heterologous cell systems, or species-specific co-factors may facilitate Mad-Baboon binding (Peterson, 2012).

The strong functional conservation of TGF-β superfamily proteins prompts a comparison of the current results in Drosophila with studies describing cross-pathway signaling in mammalian systems. The limited number of pathway members and the efficacy of RNAi in Drosophila enabled distinguishing between competing models presented for mammalian epithelial cells. With regard to the key mechanistic question of whether TGF-β Type I receptors can directly phosphorylate the BMP R-Smads, the observation of Mad phosphorylation in the absence of BMP Type I receptors is inconsistent with the model of heteromeric TGF-β/BMP Type I receptor complexes proposed in a previous study, but is consistent with the model of direct phosphorylation of BMP R-Smads by TGF-β/Activin receptors. It is possible that both types of mechanisms exist, but are differentially utilized depending on the ligands and receptors present in the particular tissue being examined. Even in Drosophila, it is noteed that direct action of Baboon on Mad does not preclude the possibility that mixed receptors form active signaling complexes in some situations. Mixed receptor complexes have been detected in Drosophila cell culture under over-expression conditions, but their functionality is unknown. A similar point can be made regarding mixed Smad oligomers detected in mammalian epithelial cells. In adult wing assay, this study found that the Babo* phenotype depended primarily on Mad and that dSmad2 was not required. If the wing development defect was caused by the activity of a complex containing both P-Mad and P-dSmad2, then removal of either one should have blocked the Babo* phenotype. Again, this observation does not argue against the formation or activity of mixed R-Smads complexes in some contexts, but shows that productive signaling by cross-pathway phosphorylation can take place independently of such complexes (Peterson, 2012).

One key observation in this report concerning the mechanism of cross-pathway signal regulation is that the degree to which it occurs, both in cell culture and in vivo, appears to be regulated by competition between the R-Smads, likely for receptor binding. Further work is required to determine how general this mechanism might be. Epithelial cell culture models showed that cross-talk is important for the TGF-β-induced migratory switch. The results in the larval wing disc and gut represent the first examples of cross-talk in vivo, and it is expected that additional examples will be found in various animals and tissues. With regard to developmental studies, other systems should be evaluated to see if loss-of-function mutations of Smad2/3 orthologs lead to increased signaling through BMP R-Smads. Additionally, several human diseases have been attributed to mutations in TGF-β components. A cautionary implication of this work is that mutations in the TGF-β branch may have unanticipated loss- or gain-of-function influences on the BMP branch (Peterson, 2012).

Curiously, TGF-β/Activin Type I receptors appear to have gained or retained cross-phosphorylation activity throughout evolution, but BMP Type I receptors do not appear to have reciprocal activity. Phosphorylation of dSmad2 by activated Drosophila Type I BMP receptors, or in response to various ligands, has never been seen. Likewise, no phosphorylation of Smad2/3 by BMP receptors has been reported in vertebrate cells. Why is there this distinction? The growing number of complete genome sequences has allowed phylogenetic reconstruction of the evolution of TGF-β signaling. Apparently all metazoans, even the simple placazoan, have a complex TGF-β network containing both TGF-β/Activin and BMP subfamilies. It is tempting to speculate that a single receptor with dual Smad targets could have played a transitional role in the expansion of the signaling network. The all-or-none nature of the TGF-β-superfamily network (both BMPs and TGF-β/Activins present, or none) in extant organisms, however, does not provide support for this idea. What it does support is the possibility that relationships between core pathway proteins stabilized hundreds of millions of years ago. If the dual-kinase activity of Baboon orthologs has been available to all metazoans during the radiation of animal forms, this activity would be expected to be deployed by different animals and different cell types in diverse ways. Several phenotypes are currently being investigated that differ between baboon and dSmad2 mutants to determine which depend on Babo-Mad cross-talk and which might depend on other forms of Smad-independent signaling (Peterson, 2012).

The different responses of adult wings and larval imaginal tissue upon Baboon activation illustrate that TGF-β pathway wiring and output can vary with developmental context. Given the relationship between Babo and dSmad2, the cross-talk activity can be viewed two ways. From one perspective, the response to loss of dSmad2 depends on the level of Baboon signaling: only cells with Baboon activity can produce P-Mad by cross-talk. From the other perspective, the response of wildtype cells to Baboon stimulation depends on dSmad2 levels: efficient cross-talk will only occur in the absence of dSmad2 (Peterson, 2012).

Given the dominant control that dSmad2 exerts on the ability of Baboon to phosphorylate Mad, the simplest model is that output of TGF-β ligand stimulus in a given cell depends on the expression level of dSmad2. Although dSmad2 is widely expressed, as are Smad2/3 proteins in other animals, different subsets of cells might express different ratios of R-Smads that could influence the signaling output (Peterson, 2012).

The observation that Baboon activity can lower the overall level of dSmad2 protein offers an additional regulatory possibility, where the TGF-β/Activin signal itself can influence its response. In the substrate-switch model, a cell exposed to a prolonged Activin signal would eventually degrade enough of its dSmad2 pool to allow Baboon signaling through Mad. It is not known which tissues, if any, require Baboon-to-Mad signaling in normal developmental contexts. In some cases, dSmad2 over-expression leads to mad loss-of-function phenotypes, which could represent disruption of a sensor incorporating dSmad2 concentration as an input and Mad activity as an output. Proteosomal degradation of activated Smad proteins is a well documented mechanism of signal attenuation, but the current observation of bulk degradation adds a new dimension because it can redirect receptor signaling. The mechanism of signal-dependent bulk dSmad2 degradation is unknown, and preliminary experiments do not support a simple ubiquitylation-proteosomal pathway. Regardless of the molecular details, the observed reduction of total dSmad2 available for receptor competition is the key parameter in the competition model. Since activated R-Smad proteins can also be dephosphorylated to rejoin the pool of would-be substrates, the relative rates of recycling and bulk degradation would be predicted to influence the substrate switch to Mad (Peterson, 2012).

What could be gained by the substrate switch? A general answer is that multiple interactions permit diverse regulatory schemes. For example, cross-talk would permit Mad signaling in a cell that is not exposed to BMP ligands or is otherwise not competent to transduce such signals. Other modes of signal integration are intriguing, such as the conditional formation of mixed R-Smad complexes. Another type of regulatory link that could contribute to the evolutionary maintenance of an integrated dSmad2-Babo-Mad triad considered. If dSmad2 transcription were a target for P-Mad, this would affect the substrate switch. In once case, Baboon signaling to Mad would be triggered by dSmad2 depletion and serve to up-regulate Smad2 to return balance to the system. In contrast, if dSmad2 were down-regulated by P-Mad, this would stabilize the switch to the Baboon-Mad interaction. Additional studies are required to explore these intriguing possibilities (Peterson, 2012).

GENE STRUCTURE

cDNA clone length - 3083

Bases in 5' UTR - 792

Bases in 3' UTR - 849

PROTEIN STRUCTURE

Amino Acids - 486

Structural Domains

Smox sequence alignment with other known Smads revealed ~50% overall identity to Mad or human Smad1 and ~70% identity to human Smad2 or Smad3. Furthermore, the MH2 domain represented the region of highest homology with greater than 90% identity between Smox and either Smad2 or Smad3. In the MH1 domain, Smox lacks the two inserts that are found in the MH1 domain of Smad2. In the linker, Smox contains a PY motif that is present in other Smads, but in addition has a glycine, serine, and glutamine-rich insert (amino acid residues 177-251) that is absent in either hSmad2 or hSmad3 (Brummel, 1999).

Smox has 60% overall identity to human SMAD3, 57.1% identity to human SMAD2, and 51.8% identity to Drosophila MAD. Smad proteins have two highly conserved domains, known as MH1 and MH2, separated by a linker region that is less conserved. Smox MH1 has 59.0% identity to the MH1 domain of SMAD3 and 48.9% identity to the MH1 domain of SMAD2. Smox MH2 has 92.9% identity to the MH2 domains of both SMADs2 and 3. Like SMAD3, Smox does not have the two insertions of nine and thirty residues found in the MH1 domain of SMAD2. Smox has an SSXS motif at the C-terminus that has been found in all receptor-regulated SMADs characterized thus far. The two most C-terminal Ser residues of this motif in other class I SMADs are phosphorylated by activated type I receptors (Henderson, 1999).

To determine where Smox may be functioning, the expression of Smox was examined both by Northerns and by in situ hybridization to whole mount embryos. The Northern analyses revealed two Smox transcripts, both approximately 4.5 kb. The larger of the two transcripts is abundant in adult females and in 0-2 hr embryos suggesting that the larger Smox transcript is provided maternally. To determine the spatial expression pattern of Smox, embryos were hybridized with both sense and antisense Smox RNA probes. Global expression was detected throughout embryogenesis with the antisense probe but not with the negative control sense probe, suggesting that Smox is expressed to equivalent levels in all cells of the embryo (Henderson, 1999).

date revised: 23 June 2023

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

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