BIOLOGICAL OVERVIEW

Drosophila Smurf1 (Lethal with a checkpoint kinase) is a negative regulator of signaling by the BMP2/4 ortholog Decapentaplegic during embryonic dorsal-ventral patterning. Smurf1 encodes a HECT domain ubiquitin-protein ligase, homologous to vertebrate Smurf1 and Smurf2, that binds the Smad1/5 ortholog in Drosophila Mothers against dpp (Mad) and likely promotes its proteolysis. The essential function of Drosophila Smurf1 is restricted to its action on the Dpp pathway. Smurf1 has two distinct, possibly mechanistically separate, functions in controlling Dpp signaling. Prior to gastrulation, Smurf1 mutations cause a spatial increase in the Dpp gradient, as evidenced by ventrolateral expansion in expression domains of target genes representing all known signaling thresholds. After gastrulation, Smurf1 mutations cause a temporal delay in downregulation of earlier Dpp signals, resulting in a lethal defect in hindgut organogenesis. The results suggest that Smurf1 provides an important mechanism to maintain the available pool of Mad at limiting concentrations, and may have additional functions in regulating the levels of Dpp receptors (Podos, 2001).

How do ubiquitin ligases work? The process of regulated degradation has been implicated in a variety of cellular responses such as the heat shock response, cell cycle progression, DNA repair, signal transduction, and transcription. It is now understood that protein ubiquitination is carried out by a sequence of three enzymes, an E1 ubiquitin-activating enzyme, E2 ubiquitin-conjugating enzymes, and E3 ubiquitin ligases. Among these, E3 ubiquitin ligases play a crucial role in defining substrate specificity and subsequent protein degradation by the 26S proteasomes. Smurfs (Smad ubiquitination regulatory factor), members of the Hect family of E3 ubiquitin ligases, interact with the BMP-activated Smads, thereby triggering their ubiquitination and degradation. Hect domain proteins represent a major subclass of E3 ligases and contain a conserved cysteine, located toward the carboxyl end of the Hect domain, which is capable of forming a thioester bond with ubiquitin. Ubiquitin is first transferred from an appropriate E2 enzyme to this cysteine residue of the E3 ligase. This E3-ubiquitin thioester then acts as the donor for amide bond formation with the protein substrate. Another motif often found in the Hect family of E3 ligases is the WW domain, which derives its name from the presence of two highly conserved tryptophans and a conserved proline in an approximately 30-amino acid region. The WW domains have a preference for binding to small proline-rich sequences, PPXY motifs, and different WW domains possess differential substrate specificity. The WW domains of Smurf1 have been shown to interact with Smads through a PPXY motif in the linker region of the Smads (Zhang, 2001 and references therein).

To identify novel negative regulators of BMP signaling in Drosophila, a genetic selection was conducted for mutations that result in elevated Dpp activity during embryonic D-V pattern formation. From 65,000 mutagenized genomes, six extragenic mutations were recovered that suppressed the lethal, partially ventralized embryonic phenotype caused by the hypomorphic maternal-effect Medea mutation, Med¹⁵. Both the molecular and genetic characterization of two of these mutations, 11R and 15C, which disrupt the previously unrecognized DSmurf locus, are presented in this study. Both mutations act as largely recessive maternal-effect suppressors that restore viability and wild-type pattern to Med¹⁵ embryonic progeny, either as homozygotes or in trans-heterozygous combination. Because these mutations act in the same fashion to suppress the partially ventralized embryonic phenotype caused by dpp haploinsufficiency, they effect a general elevation of Dpp signaling activity during embryonic D-V pattern formation (Podos, 2001).

The suppressor activities of 11R and 15C were mapped to a ~1 Mbp interval on the second chromosome. Following the reported identification of human and Xenopus Smurf1 (Zhu, 1999), a Drosophila gene with strong sequence similarity was identified within the defined genetic interval, at position 54D. The 11R and 15C mutants each contain a defective form of a Hobo element inserted into a separate site within the coding region of this gene. In addition, intragenic revertants of the lethal 15C phenotype have been obtained that lack the Hobo insertion. The Smurf1 gene therefore corresponds to the locus identified genetically, and 11R and 15C represent probable loss-of-function Smurf1 mutations (Podos, 2001).

The activities of vertebrate Smurf1 and Smurf2 in opposition to BMP and TGF-ß signals are mediated in part by direct binding interactions with their R-Smad substrates. Whether Drosophila Smurf1 interacts physically with the R-Smad encoded by Mad was examined. In a yeast two-hybrid assay, it was found that Smurf1 binds Mad but not its co-Smad Medea. Similar to the vertebrate Smurf-Smad interactions, the interaction between Smurf1 and Mad was disrupted by deletion of the PY motif from Mad. Smurf1 therefore shares substrate binding properties with its vertebrate homologs, likely reflecting a common function in restricting Dpp/BMP signals by promoting the proteolysis of the BMP-specific R-Smad proteins (Podos, 2001).

The cuticles of the Smurf1 mutant embryos revealed no overt defects in D-V pattern; there was no evident expansion of dorsal epidermal markers or loss of ventrolateral fates. However, most Smurf1^15C embryos, and many dead Smurf1^11R embryos, display a novel phenotype, a hole in the dorsal posterior region of the cuticle extending from the position of the spiracles to the posterior pole. Smurf1^15C embryos also have variable head defects, and half show a more extensive dorsal-open phenotype (Podos, 2001).

Despite the high penetrance of the Smurf1^15C lethal phenotype, elimination of one copy of the dpp gene restores these Smurf1 mutant embryos to wild-type embryonic appearance and nearly full embryonic viability (83% of expected). These embryos ultimately give rise to adults with normal morphology and significant fertility. Since dpp itself is haplolethal, the viability of this genotype is evidence of mutual suppression between two lethal conditions. These results demonstrate that the only essential function of Smurf1 is to restrict Dpp signals during embryogenesis. Smurf1 might act upon other substrates, yet such functions must be redundant with a second E3 ubiquitin-protein ligase or otherwise be dispensable for viability. In contrast, Smurf1^15C does not suppress the fully ventralized embryonic phenotype of a dpp null homozygote, indicating that loss of Smurf1 function does not elicit ligand-independent signaling. This result supports previous observations that elevation of the cytoplasmic pool of Mad in the context of embryonic D-V pattern formation is not sufficient to elicit ligand-independent signaling (Podos, 2001).

While mutation of Smurf1 does not cause overt alterations in D-V cuticular pattern, the cuticle presents a snapshot of embryonic development that does not necessarily reflect initial D-V pattern and does not incorporate the dorsal-most tissue, the amnioserosa. To obtain a direct readout of the Dpp activity gradient that is sensitive to subtle changes in its strength and spatial parameters, wild-type and mutant embryos were examined for changes in the spatial extent of staining with the phosphorylated form of MAD (P-Mad) antibody and changes in the expression domains of direct Dpp target genes (Podos, 2001).

In wild-type embryos at the onset of gastrulation, a stripe of P-Mad staining is visible in a dorsal subset of dpp-expressing cells and in the cells at either pole of the embryo. In Smurf1^15C mutant embryos, there is a small but statistically significant increase (28%, P < 0.001) in the width of the dorsal P-Mad stripe as well as a nonquantitated increase in the intensity of staining. In wild-type embryos at this stage, the Dpp target genes zen and Race are activated by high levels of Dpp signaling in the presumptive amnioserosa, while the intermediate threshold target gene u-shaped (ush) is activated in a broader domain by lower levels of Dpp activity. All three transcriptional domains showed significant lateral expansion in Smurf1^15C mutant embryos; a lesser but significant expansion of zen was also observed in Smurf1^11R mutant embryos. Later, Smurf1^15C mutant embryos differentiate a nearly 2-fold excess of amnioserosa cells compared to wild-type. A 2-fold increase in dpp gene dosage effects a similar expansion of zen transcription and a comparable increase in amnioserosa cell number. These observations indicate that disruption of Smurf1 gene activity elicits an expansion of multiple Dpp signaling thresholds in the early embryonic ectoderm comparable to the phenotype caused by a doubling of dpp gene dosage (Podos, 2001).

At its lowest threshold, Dpp signaling, acting to promote its own transcription, defines the boundary between the dorsal epidermis and neurogenic ectoderm. This positive feedback of Dpp on its own transcription is opposed by the action of the negative regulators Short gastrulation (Sog) and Brinker (Brk) in the neurogenic ectoderm. Although this boundary was positioned normally in Smurf1 mutants, it was important to determine whether sog activity masked an effect of a Smurf1 mutation on this Dpp threshold (Podos, 2001).

In both Smurf1^15C and sog single mutant embryos, dpp is transcribed approximately within its normal dorsal domain at the onset of gastrulation. However, in sog; Smurf1^15C double mutant embryos, dpp transcription expands significantly, although variably, into the ventrolateral neurogenic ectoderm. Strikingly, these double mutant embryos ultimately differentiate a fully dorsalized cuticle, in which ventral denticles are replaced by dorsal hairs. These results indicate that Smurf1 and sog are genetically redundant, yet functionally distinct, in limiting the spatial extent of dpp transcription and the consequent specification of dorsal epidermis. It is concluded from this set of results that Smurf1 contributes quantitatively to the establishment of multiple Dpp signaling thresholds across the entire range of the Dpp activity gradient (Podos, 2001).

Although the biochemical activities of Smurf family members have been described, this genetic analysis has allowed several conclusions to be drawn about wild-type Smurf1 function: (1) Smurf1 is required for normal embryogenesis, and its essential function is restricted to action on the Dpp pathway and not derived from its separate action on other possible substrates; (2) Smurf1 is required for the proper spatial control of Dpp signaling prior to gastrulation, since Smurf1 mutant embryos show evidence of lowered Dpp signaling thresholds for all target genes examined; (3) Smurf1 is required for the downregulation of Dpp signaling in multiple tissues after gastrulation. It is proposed that the pre- and post-gastrulation functions of Smurf1 arise from distinct activities of the Smurf1 enzyme on Dpp signal transduction components (Podos, 2001).

Despite the intensive study of Dpp-dependent developmental events, there are multiple reasons why the role of Smurf1 eluded prior notice: (1) both the maternal and zygotic components of Smurf1 must be mutated to uncover a lethal phenotype; (2) Smurf1 mutants are relatively dosage insensitive, as Smurf1 mutations do not exert significant dominant phenotypes even in sensitized backgrounds such as Med¹⁵, precluding their isolation in most genetic screens. Such dosage insensitivity might be a general feature of enzymes, as opposed to stoichiometric components of signaling pathways such as Mad and MED. (3) Smurf1 mutations do not cause overt defects in dorsal-ventral patterning. Possibly, the activity of Smurf1 is partially redundant with another ubiquitin-protein ligase. In support of this hypothesis, Dpp signals are eventually downregulated in Smurf1 mutants. Moreover, an analysis of human Smad2 turnover (Lo, 1999) has implicated a ubiquitination activity that does not require the Smad2 linker domain and therefore is likely independent of the Smurf proteins (Podos, 2001).

The spatial modulation by Smurf1 of graded Dpp signaling is evident prior to the onset of gastrulation. The abrogation of Smurf1 activity causes a sensitization to Dpp signals at all positions in the D-V activity gradient, as indicated by expanded domains of P-Mad staining, target gene transcription, and tissue differentiation. Similar global expansions of Dpp-dependent territories have been observed in embryos with elevated dpp gene dosage. Since an increase in ligand concentration is likely to result in the phosphorylation of additional cytoplasmic Mad, this spatial control over Dpp target gene expression is likely to derive from the unregulated ubiquitin-mediated proteolysis of Mad throughout the embryo. Similar properties have been established for human Smurf1 (Zhu, 1999), from demonstrations that BMP receptor activation does not alter the rate of Smad1 ubiquitination and degradation mediated by Smurf1 (Podos, 2001).

The results suggest that Smurf1 provides an important mechanism to maintain the available pool of Mad at limiting concentrations, the necessity of which has been supported by previous genetic observations. Although not normally haploinsufficient, the Mad gene is rendered so when the activities of other components of the Dpp pathway, including dpp, zen, and sog, are reduced. More generally, limiting amounts of Smad protein might be an essential feature of all graded TGF-ß superfamily signaling systems. Cytoplasmic Smad pools are similarly limiting in Xenopus embryos, according to quantitative studies of activin signaling. Experimental elevations in Smad2 concentration cause proportionate increases in Smad activation, as represented by both nuclear Smad2 import and transcriptional readout. Therefore, it is predicted that Smurf enzymes will prove to be essential to maintain Smad proteins at limiting concentrations to ensure appropriate responses to all graded BMP and activin/TGF-ß signals (Podos, 2001).

The Med¹⁵ mutation is a missense lesion in the MH2 domain, within the L3 structural loop that has been implicated in the signal-dependent interaction between trimers of Mad and MED. Because an experimental elevation of wild-type Mad levels is sufficient to restore the specification of amnioserosa to embryos derived from Med¹⁵ mothers (Hudson, 1998), it is hypothesized that Smurf1 mutations suppress Med¹⁵ because the resulting elevation of Mad is sufficient to overcome its reduced affinity for the mutant MED protein (Podos, 2001).

Phenotypic analysis has identified a second requirement for Smurf1 in the temporal downregulation of Dpp signaling. With the exception of the amnioserosal cells, all of the descendants of cells with high levels of P-Mad at the onset of gastrulation downregulate P-Mad staining by stage 10. In contrast, Smurf1 embryos of the same stage retain P-Mad staining in all these cell types, leading to deleterious consequences in hindgut morphogenesis. A causal link has been established between the prolonged Dpp signaling in the dorsal hindgut primordium, ectopic zen transcription, and the subsequent breakdown in the epithelial integrity of the hindgut cells (Podos, 2001).

This second function of Smurf1 is distinct, and possibly mechanistically separable, from its ability to target Mad for destruction. Although the incremental spatial expansion of P-Mad staining along the dorsal-ventral axis at the blastoderm stage is consistent with an overall increase in the amount of Mad protein in Smurf1 mutants, the complete lack of temporal downregulation of P-Mad staining in stage 8-10 Smurf1 embryos is more consistent with a specific effect of Smurf1 on P-Mad. Because the level of P-Mad is a readout of both the amount of Mad protein within a cell and the intensity of Dpp signaling that the cell receives, one attractive hypothesis is that Smurf1 downregulates the level of P-Mad by antagonizing Dpp signaling, independent of Mad degradation. One possible mechanism is suggested by demonstrations that vertebrate Smurf1 and Smurf2 can use the I-Smad, Smad7, as an adaptor to promote the degradation of activated TGF-ß receptors (Ebisawa, 2001; Kavsak, 2000; Zhu, 1999). It is proposed that Smurf1 might similarly target activated Dpp receptors for degradation, using Mad or, more likely, the I-Smad protein DAD as an adaptor. Such an activity would serve as a feedback mechanism of attenuation, whereby receptors are targeted for degradation only upon activation by ligand. Feedback mechanisms are integral to many developmental signaling processes; the proposed feedback activity of Smurf1 on the Dpp receptors would be one of the few instances where temporal downregulation of a signaling system has been shown to be necessary for the proper differentiation of cells previously exposed to the signal (Podos, 2001).

Involvement of Smurf1 have been demonstrated in the control of multiple aspects of Dpp signaling. Further analysis will be required to determine whether Smurf1 acts bifunctionally to mediate the degradation of Mad and of activated Dpp receptors. Smurf1 might also have additional activities that have not been uncovered by these mutants. For example, although homozygous Smurf1 adults have normal cuticular morphology, an examination of Dpp target gene expression might reveal Smurf1 function in imaginal disc patterning. By analogy to the vertebrate Smurf proteins, Smurf1 also might modulate the activin/Smad2 signaling pathway, which has been implicated in the control of cell proliferation. Lastly, Smurf1 might promote the ubiquitin-mediated degradation of other substrates, independently of the Smads. More generally, further characterization of the relationship between Smurf1 and other modulators of Dpp signaling will yield insights into the precise regulation that underlies intercellular signaling systems during normal development (Podos, 2001).

GENE STRUCTURE

cDNA clone length - 4886

Bases in 5' UTR - 881

Bases in 3' UTR - 819

PROTEIN STRUCTURE

Amino Acids - 1061

Structural Domains

Ubiquitin-mediated proteolysis regulates the activity of diverse receptor systems. Smurf2, a C2-WW-HECT domain ubiquitin ligase, associates constitutively with Smad7. Smurf2 is nuclear, but binding to Smad7 induces export and recruitment to the activated TGF beta receptor, where it causes degradation of receptors and Smad7 via proteasomal and lysosomal pathways. IFN gamma, which stimulates expression of Smad7, induces Smad7-Smurf2 complex formation and increases TGF beta receptor turnover, which is stabilized by blocking Smad7 or Smurf2 expression. Furthermore, Smad7 mutants that interfere with recruitment of Smurf2 to the receptors are compromised in their inhibitory activity. These studies thus define Smad7 as an adaptor in an E3 ubiquitin-ligase complex that targets the TGF beta receptor for degradation (Kavsak, 2000).

The domain structure of the predicted Drosophila Smurf1 protein is typical for HECT domain ubiquitin-protein ligases, and closely resembles that of the vertebrate Smurf1 and Smurf2 proteins (Kavsak, 2000; Zhu, 1999). Smurf1 contains an amino-terminal Ca²⁺/phospholipid binding C2 domain of unknown function, three protein binding WW domains, and a carboxy-terminal catalytic HECT domain with the catalytic cysteine that covalently binds to ubiquitin moieties. The second and third WW domains (WW2 and WW3) of Drosophila Smurf1 correspond to WW domains within vertebrate Smurf1 and Smurf2 that have been implicated in Smad substrate recognition, binding a short peptide motif, PPXY (PY), that is present within the linker domain of their Smad substrates. Within both the WW2-WW3 region and the HECT domain, the Drosophila Smurf1 sequence is 70%-72% identical to the vertebrate Smurf1 and Smurf2 proteins but less than 50% identical to related domains of any other proteins in the sequenced human and Drosophila genomes. Two additional regions of similarity, which have been designated S1 and S2, are unique to the Drosophila and human Smurf proteins and thus might participate in Smad substrate recognition. It is concluded from this analysis that Smurf1 encodes the only ortholog of vertebrate Smurf1 and Smurf2 in the fully sequenced Drosophila genome (Podos, 2001).

date revised: 25 March 2015

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.