InteractiveFly: GeneBrief

Liquid facets : Biological Overview | Regulation | Developmental Biology | Effects of Mutation | Evolutionary Homologs | References

BIOLOGICAL OVERVIEW

Epsin is part of a protein complex that performs endocytosis in eukaryotes. Drosophila epsin, Liquid facets (Lqf), was identified because it is essential for patterning the eye and other imaginal disc derivatives. Epsins are endocytic proteins that bind phosphatidylinositol (4,5)-bisphosphate [PtdIns(4,5)P2] in the plasma membrane as well as Clathrin, AP-2 Adapter Complex, and other accessory proteins in coated pits (reviewed by Wendland, 2002). Epsins were initially thought to be core components of the endocytic machinery because of the dominant-negative effects of truncated Epsin proteins on endocytosis in mammalian cells (Chen, 1998; Ford, 2002), their essential role in yeast endocytosis (Wendland, 1999), and their inherent capacity to induce membrane curvature (Ford, 2002) and bind other core components such as Clathrin and AP-2 (Chen, 1998; Owen, 1999; Rosenthal, 1999; Wendland, 1999; Drake, 2000). More recently, however, the identification of Ubiquitin-interacting motifs (UIMs) in Epsins, as well as in other proteins involved in membrane trafficking (Hofmann, 2001), have led to the suggestion (Wendland, 2002) that Epsins belong to a family of cargo-selective adapters that link mono-ubiquitinated cell-surface proteins with the endocytic machinery (Wang, 2004 and references therein).

Ubiquitination of Delta appears necessary for endocyosis of Delta. Two members of the Notch pathway, neuralized and mind bomb, encode E3-ubiquitin-ligases, ubiquitinate Delta and appear to be required for Delta internalization; loss of neuralized or mind bomb function causes excessively high levels of Delta at the cell membrane and blocks Notch signaling. The Drosophila homolog of epsin, liquid facets (lqf), has also been shown to promote Delta internalization (Overstreet, 2003), suggesting lqf may regulate Notch signaling even though the functional ramifications of lqf on Notch pathway activity have not been investigated (Tian, 2004 and references therein).

Delta/Serrate/Lag2 (DSL) ligands must normally be endocytosed in signal-sending cells via the action of Liquid facets to activate Notch on the surface of signal-receiving cells. Surprisingly, however, bulk endocytosis of DSL ligands appears normal in the absence of Lqf. This apparent paradox is resolved by providing evidence that Lqf is unique among adapters that target mono-ubiquitinated cargo proteins for internalization: it allows them to enter a special endocytic pathway that DSL ligands must enter to acquire signaling activity. This requirement can be bypassed by introducing the internalization signal that normally mediates internalization and recycling of the Low Density Lipoprotein (LDL) receptor. On the basis of these results, it is hypothesized that Epsin-mediated endocytosis might be required to allow DSL proteins to be recycled rather than degraded following internalization, possibly to convert them from inactive pro-ligands into active ligands (Wang, 2004).

Cardioblasts are the contractile cells of the heart and coalesce to form the heart tube. liquid facets functions as an inhibitor of cardioblast development in Drosophila. lqf inhibits cardioblast development and promotes the development of fusion-competent myoblasts, suggesting a model in which lqf acts on or in fusion-competent myoblasts to prevent their acquisition of the cardioblast fate. lqf and Notch exhibit essentially identical heart phenotypes, and lqf genetically interacts with the Notch pathway during multiple Notch-dependent events in Drosophila. The link between the Notch pathway and epsin function has been extended to C. elegans, where the C. elegans Lqf ortholog acts in the signaling cell to promote the glp-1/Notch pathway activity during germline development. These results suggest that epsins play a specific, evolutionarily conserved role to promote Notch signaling during animal development and support the idea that they do so by targeting ligands of the Notch pathway for endocytosis (Tian, 2004).

Epsins, including the sole Drosophila Epsin Lqf, contain a series of discrete functional domains that implicate them in Clathrin-mediated endocytosis. These include the N-terminal ENTH domain, which can induce membrane curvature in response to PtdIns(4,5)P₂ binding, as well as binding sites for Clathrin, AP-2 and accessory proteins (e.g. Eps15), and multiple Ubiquitin Interactions Motifs (UIMs). Recent evidence (Mishra, 2002; Wendland, 2002) has suggested that Epsins are members of a class of structurally related proteins that function as cargo selective adapters that target substrate proteins for Clathrin-mediated endocytosis (Wang, 2004).

An absolute, cell-autonomous requirement for Lqf in generating functional DSL ligands has been demonstrated. However, no other role has been detected for Lqf in cell-cell signaling, such as in receiving DSL ligands, or in sending, receiving, or controlling the distribution of other extracellular signals, notably Hg, Wg and Dpp. Furthermore, cells devoid of Lqf activity appear to grow, proliferate and interdigitate in a manner that is indistinguishable from cells devoid of Dl and Ser, the two DSL ligands in Drosophila. These results suggest that Epsin function in Drosophila may be essential solely for the production of active DSL ligands (Wang, 2004).

Surprisingly, no effect of removing Lqf was detected on the steady state accumulation of Dl in endocytic compartments. However, a modest effect on Dl internalization was detected in a sensitized background in which Dl endocytosis was greatly enhance by overexpressing Neuralized, an E3-Ubiquitin ligase that ubiquitinates Dl. Strikingly, high levels of Dl accumulate in endocytic vesicles of such Neur overexpressing cells whether or not they have Lqf, but only cells that have Lqf can signal. It is therefore inferred that Epsin is required for a discrete and apparently small subset of the endocytic events that normally internalize DSL ligands; however, it is this subset that is crucial for generating active DSL signals (Wang, 2004).

The selective requirement for Epsin in sending DSL ligands is reminiscent of that for the Presenilin/gamma-secretase complex in transmembrane cleavage and signal transduction by Notch. Selectivity in the case of the Presenilin/gamma-secretase complex does not reflect a dedicated role in Notch proteolysis, but rather an unusual property of the Notch transduction mechanism, namely that ectodomain shedding activates the pathway by inducing transmembrane proteolysis. Similarly, selectivity in the case of Epsin may reflect an unusual requirement for DSL ligands to signal, and not a dedicated role of Epsin in confering their signaling activity (Wang, 2004).

It is generally thought that Epsins target cargo proteins for endocytosis via mono-ubiquitin internalization signals (Wendland, 2002). The cytosolic domain of Dl was replaced with a random peptide (R+) that contains two Lysines, and the presence of at least one Lysine was shown to be essential for both the endocytosis and signaling activity of the chimeric Dl^R+ ligand. To test the possibility that the presence of Lysine targets Dl^R+, as well as wild-type Dl, for endocytosis by serving as an Ubiquitin acceptor, the cytosolic domain of Dl was replaced with a non-Lysine containing form of Ubiquitin. The resulting Dl^Ubi+ ligand could be endocytosed and had at least partial signaling activity, but not if the Ubiquitin domain contained an additional mutation that blocks its ability to be targeted for endocytosis. Finally, and critically, it was demonstrated that the signaling activity of Dl^R+, like that of wild-type Dl, depends on Lqf. Collectively, these findings implicate mono-ubiquitination as the internalization signal required to target DSL ligands for endocytosis by Epsin (Wang, 2004).

Significantly, bulk endocytosis of the chimeric Dl^R+ ligand, like that of wild-type Dl, appears to be unaffected in cells devoid of Lqf, even though signaling activity is abolished. Hence, it is inferred that Lqf is not the only adapter protein that can target mono-ubiquitinated substrate proteins for endocytosis. Nevertheless, Lqf appears to be unique among all such adapter proteins in its ability to direct internalization of mono-ubiquitinated DSL ligands in a manner that confers signaling activity. It is therefore suggested that Epsin has a dedicated role in directing mono-ubiquitinated cargo proteins into a particular endocytic pathway, one that DSL ligands must enter in order to acquire signaling activity. As is detailed below, it is suggest that Epsin might direct DSL ligands specificially into a recycling pathway (Wang, 2004).

It is notable that substitution of the cytosolic domain of Dl with a peptide carrying the FDNPVY internalization signal from the LDL receptor yields a chimeric Dl^LDL+ ligand that is endocytosed and has signaling activity. However, in this case, Lqf is not essential for signaling. One interpretation of this result is that mono-ubiquitinated DSL ligands are normally targeted for endocytic pathways that preclude their signaling activity, unless they are diverted from entering these pathways by association with Lqf, or by the presence of a heterologous internalization signal such as FDNPVY. In both cases, endocytosis would take place via an alternate pathway compatible with signaling activity (Wang, 2004).

Why must DSL ligands on the surface of signal-sending cells be endocytosed in order to activate Notch on the surface of signal-receiving cells? Two general classes of explanation can be distinguished. In the first, activation of Notch is triggered by early events in the process of DSL endocytosis that occurs while the ligands are still on the cell surface, prior to the pinching off of coated vesicles. In the second, internalization of DSL proteins is a necessary prerequisite for endocytic recycling, which is required for subsequent signaling activity (Wang, 2004).

Most previous models fall into the first class. One such internalization model proposed that DSL/Notch binding creates a physical bridge between the sending and receiving cell that is mechanically stressed by endocytosis of the ligand, causing conformational changes in Notch that elicit either S2 or S3 cleavage. Another model proposed that recruitment of DSL ligands into coated pits increases their local abundance on the cell surface. Both of these models are difficult to reconcile with the finding that Lqf is essential for Dl signaling but not for bulk endocytosis of Dl. This result indicates that DSL endocytosis in signal-sending cells is not sufficient, per se, to activate Notch in signal-receiving cells. Instead, as suggested above, it appears that DSL ligands have to enter, or traffic through, a special Lqf-dependent endocytic pathway to activate Notch (Wang, 2004).

For such internalization models to accommodate these results, it seems necessary to posit that productive interactions between DSL ligands and Notch require a special micro-environment that is associated only with a particular subclass of coated pits or other specializations. Mono-ubiquitinated cargo proteins might be excluded from such structures, unless chaperoned there by Lqf. Thus, only DSL ligands that gain entry, whether via Lqf, or by the targeting mediated by the LDL receptor signal, would be able to activate Notch on the abutting surface of the receiving cell. Furthermore, one would have to posit the existence of accessory molecules that are provided by the sending cell, sequestered in these specializations, and essential for DSL-dependent activation of Notch on the receiving cell, whether by mechanical stress, DSL clustering, or some other means (Wang, 2004).

The second general class of explanation suggests, by the ability of the internalization signal from the LDL receptor to bypasses the requirement for Lqf, that recycling is the key element. In general, mono-ubiquitination acts as a sorting signal in the endosomal system that leads to delivery of membrane proteins to late endosomes and eventually lysosomes. Hence, Epsin-binding to mono-ubiquitinated DSL proteins during endocytosis might allow those DSL proteins to escape degradation by altering their sorting, thus allowing them to enter a recycling pathway. Passage through this pathway would be essential to confer signaling activity (Wang, 2004).

Why might recycling be necessary for DSL ligands to acquire signaling activity? One possibility is that recycling allows DSL ligands to be stripped of the bound ectodomain of Notch so that they can be re-used. Multiple rounds of recycling might then enhance the level of active DSL ligands on the surface of signal-sending cells above a critical threshold necessary to activate Notch transduction in the signal-receiving cell. According to this view, one might expect that massive overexpression of DSL ligands would be able to bypass the requirement for Lqf. However, the results suggest that this is not the case: it is estimated that, in these experiments, overexpressed Dl accumulates on the cell surface at levels up to tenfold higher than peak accumulation of endogenous Dl, yet is unable to rescue DSL signaling activity in cells devoid of Lqf (Wang, 2004).

Alternatively, recycling of nascent DSL proteins may be important to convert inactive 'pro-ligands' into active ligands. Conversion might entail recruitment of DSL proteins into signaling exosomes. However, Dl signaling appears to be unaffected in cells devoid of Hrs, despite impairment in the maturation of early to late endosomes, and in the formation of multi-vesicular bodies from which exosomes might derive. Another possibility is that DSL proteins need to be processed in order to be converted to active ligands, a hypothesis that is consistent with the evidence that Lqf-dependent endocytosis of Dl correlates with a specific proteolytic cleavage of the ligand. Lqf would be required in this scenario to allow DSL ligands to enter a recycling pathway in which the required processing event can occur. The only specificity one needs to invoke in this model is that of Epsin to allow mono-ubiquitinated cargo proteins to gain access to a recycling pathway. The conditions necessary to convert DSL pro-ligands into active signals (e.g., low pH) might exist generally in early endosomes or recycling endosomes (Wang, 2004).

REGULATION

Protein Interactions

A specific protein substrate for a deubiquitinating enzyme: Liquid facets is the substrate of Fat facets

Eukaryotic genomes encode large families of deubiquitinating enzymes (DUBs). Genetic data suggest that Fat facets (Faf), a Drosophila DUB essential for patterning the compound eye, might have a novel regulatory function; Faf might reverse the ubiquitination of a specific substrate, thereby preventing proteasomal degradation of that protein. Additional genetic data implicate Liquid facets (Lqf), a homolog of the vertebrate endocytic protein epsin, as a candidate for the key substrate of Faf. Here, biochemical experiments critical to testing this model were performed. The results show definitively that Lqf is the key substrate of Faf in the eye; Lqf concentration is Faf-dependent, Lqf is ubiquitinated in vivo and deubiquitinated by Faf, and Lqf and Faf interact physically (Chen, 2002).

To detect Lqf protein levels in developing eyes, an antibody was generated to Lqf. Eye discs were double-labeled with anti-Lqf and antibodies to the endocytic protein Shibire. This shows that Lqf and Shi colocalize at cell membranes: Lqf and Shi are concentrated apically in cells within the morphogenetic furrow, an indentation that marks the onset of differentiation, and also in developing photoreceptors where their membranes meet. Similar results were obtained with antibodies to two other endocytic proteins (Dap160 and alpha-Adaptin [alpha-Ada]), and with phalloidin, which labels f-actin at cell membranes (Chen, 2002).

One prediction of the hypothesis that Faf activity prevents the degradation of Lqf is that in the developing eyes (larval eye discs) of faf null mutant flies, there should be less Lqf protein than in wild-type eyes. It was expected there would be less Lqf protein, as opposed to no Lqf protein, because the lqf null mutant eye phenotype is much more severe than the faf null mutant eye phenotype (Chen, 2002).

To test whether the level of Lqf is affected by faf⁺ gene activity, first the levels of Lqf were compared in adjacent groups of faf⁺ and faf^- cells in the eye disc, using confocal microscopy. Clones of homozygous faf^- cells were generated in faf⁺/faf^- heterozygous eye discs, marked by the absence of ß-galactosidase (ß-gal) expression. The eye discs containing clones were triple-labeled with antibodies to ß-gal (to outline the clones), to Lqf (to detect the level of Lqf protein), and to Shi (as a negative control). It was found that throughout the eye disc, the level of Lqf protein, reflected in the strength of the signal from antibody labeling, is lower within the faf^- clones than in the faf⁺/faf^- heterozygous cells surrounding them. In contrast, the levels of Shi protein are the same within and outside the clone boundaries (Chen, 2002).

To quantify the difference in Lqf protein levels in faf⁺ and faf^- cells, the levels of Lqf were assayed in eye disc protein extracts prepared from wild-type and faf^- flies in Western blot experiments. Homozygotes for two different mutant faf alleles that behave genetically as strong loss-of-function mutations were used: faf^BX4 is an inversion that makes no functional Faf protein, and faf^FO8 encodes an Faf protein with histidine residue 1986, which is critical for DUB catalytic activity, changed to tyrosine. There is two- to three-fold less Lqf in eye disc protein extracts of faf^- homozygotes than in wild-type extracts. faf⁺ transgene restores function back to faf^- flies. The transgene containing faf⁺ genomic DNA, which when introduced into faf^- homozygotes complements the mutant eye phenotype, results in a two- to three-fold increase in Lqf protein level in eye disc extracts. A nearly identical transgene that fails to complement the faf^- mutant phenotype because it has a point mutation in the codon for cysteine 1677, which is critical to the DUB activity of Faf, fails also to increase the level of Lqf protein in eye disc extracts. It is concluded that faf⁺ activity results in an increase in the level of Lqf protein (Chen, 2002).

A second prediction of the model wherein Faf prevents proteolysis of Lqf by deubiquitinating it, is that there should be Lqf protein linked to Ub chains present in eye discs. Ubiquitinated proteins are usually detected on Western blots as ladders of protein bands of higher molecular weight than the protein in question, in increments of ~8 kD; each 'rung' on the ladder represents a protein species with a Ub chain that is one Ub residue longer than the previous rung. Proteins with Ub chains are rapidly degraded, and thus difficult to detect; usually, inhibition of proteasome and/or DUB activity is required to detect them. It this study, inhibition of the DUB activity of Faf, genetically, stabilizes ubiquitinated forms of Lqf (Chen, 2002).

It is concluded that in eye discs, Lqf is ubiquitinated, and subsequently either deubiquitinated by Faf or degraded. The observation that considerable amounts of nonubiquitinated Lqf protein remain in faf^- eye discs indicates either that only a fraction of the Lqf protein in the eye disc is ubiquitinated, and/or that DUBs other than Faf also deubiquitinate some Lqf protein (Chen, 2002).

A third prediction of the model wherein Lqf is the substrate of Faf is that the proteins should, either directly or indirectly, interact. Anti-Lqf was used to immunoprecipitate Lqf from protein extracts prepared from embryos, and tested for the presence of Faf in the immunoprecipitates on Western blots. Embryos were used because sufficient protein could not be obtained from eye discs. In addition, to facilitate detection of Faf, the embryos were transformed with a P{hs-myc-faf⁺} transgene, which expresses a fully functional, myc-tagged Faf protein upon heat shock, that can be detected on Western blots with anti-myc. myc-Faf was detected in the anti-Lqf immunoprecipitate of the protein extract from heat-shocked transformant embryos (Chen, 2002).

It is concluded that myc-Faf and endogenous Lqf proteins interact physically in Drosophila embryos. Bacterially produced or in vitro translated partial Faf and full-length Lqf proteins do not bind to each other in GST pull-down assays. One possible explanation is that only full-length Faf can bind to Lqf in these assays. Alternatively, Faf and Lqf may require other proteins for their interaction (Chen, 2002).

These experiments provide critical biochemical evidence for a model in which a DUB called Faf specifically deubiquitinates Lqf protein, thereby preventing its proteolysis. There is less Lqf protein in the developing eye in the absence of catalytically functional Faf protein, that Lqf is ubiquitinated and subsequently deubiquitinated by Faf, and that Faf and Lqf interact physically. Taken together with previous genetic evidence that provides strong support for the model, it is concluded that Faf is a substrate-specific regulator of ubiquitination, a novel function for a DUB (Chen, 2002).

The eyes of faf null or lqf hypomorphic mutants have more than the normal complement of eight photoreceptor cells in each facet, owing to the failure of a cell communication pathway early in eye development. The Faf/Lqf interaction is essential in only a small number of cells in the eye disc, which must be particularly sensitive to the levels of Lqf, and in these cells, Lqf presumably controls the frequency or specificity of endocytosis. Although the precise mechanism of epsin function is unknown, vertebrate epsin binds to the endocytosis complex and also to PIP₂ at the cell membrane, and is required for endocytosis (Chen, 1998; Itoh, 2001). Apparently, appropriate endocytosis in this small group of cells is essential for successful communication with their neighbors; increased Lqf levels either enables these cells to send a signal to their neighbors that inhibits neural determination, or else prevents them from sending their neighbors a positive differentiation signal (Chen, 2002).

Through a variety of mechanisms, endocytosis is proposed to regulate ligand/receptor interactions during development. How Lqf and endocytosis regulate faf⁺-dependent cell signaling remains to be determined. Since faf has vertebrate homologs, this mode of regulation is likely to be conserved. The finding that Lqf is the key substrate of Faf in the Drosophila eye shows not only that a DUB can regulate ubiquitination and thus proteolysis, but also that an endocytosis complex protein can be a target for the control of a cell communication event critical to cell determination (Chen, 2002).

Distinct roles for Mind-bomb, Neuralized and Epsin in mediating DSL endocytos and signaling in Drosophila

Ligands of the Delta/Serrate/Lag2 (DSL) family must normally be endocytosed in signal-sending cells to activate Notch in signal-receiving cells. DSL internalization and signaling are promoted in zebrafish and Drosophila, respectively, by the ubiquitin ligases Mind-bomb (Mib) and Neuralized (Neur). DSL signaling activity also depends on Epsin (Liquid facets), a conserved endocytic adaptor thought to target mono-ubiquitinated membrane proteins for internalization. Evidence is presented that the Drosophila ortholog of Mib (Dmib) is required for ubiquitination and signaling activity of DSL ligands in cells that normally do not express Neur, and can be functionally replaced by ectopically expressed Neur. Furthermore, both Dmib and Epsin are required in these cells for some of the endocytic events that internalize DSL ligands, and the two Drosophila DSL ligands Delta and Serrate differ in their utilization of these Dmib- and Epsin-dependent pathways: most Serrate is endocytosed via the actions of Dmib and Epsin, whereas most Delta enters by other pathways. Nevertheless, only those Serrate and Delta proteins that are internalized via the action of Dmib and Epsin can signal. These results support and extend the proposal that mono-ubiquitination of DSL ligands allows them to gain access to a select, Epsin-dependent, endocytic pathway that they must normally enter to activate Notch (Wang, 2005).

To date, two E3 ubiquitin ligases have been implicated in DSL signaling: zebrafish Mind-bomb (Mib) and Drosophila Neuralized (Neur). Both proteins have been shown to promote DSL ubiquitination, endocytosis and signaling. Moreover, loss-of-function mutations in zebrafish mib and Drosophila neur cause essentially the same hallmark phenotype exemplifying a failure of DSL-signaling in their respective organisms; namely, a dramatic hyperplasia of the embryonic nervous system at the expense of the epidermis. These observations have led to the suggestion that zebrafish Mib and Drosophila Neur are functional homologs. Yet, the two proteins show only limited sequence homology; moreover they appear to be members of distinct Mib and Neur ubiquitin ligase families, each having true orthologs in both vertebrate and invertebrate genomes. As a consequence, the relative roles of Mib and Neur are not known in any animal system and this uncertainty complicates the use of mutations in these genes to assay the role of ubiquitination in DSL endocytosis and signaling (Wang, 2005).

Using newly isolated mutations in dmib, evidence has been found that Dmib and Neur constitute functionally related ubiquitin ligases that are normally required for DSL signaling in different developmental contexts. In the developing Drosophila wing disc, Dmib is required for inductive signaling across the D-V compartment boundary, as well as for the refinement of wing vein primordia, both contexts in which Neur is normally not required or expressed. However, Dmib plays only a modest role in specifying sensory organ precursor (SOP) cells, and little or no role in the subsequent segregation of distinct cell types that form each sensory organ. Instead, Neur appears to provide the essential ubiquitin ligase activity required for DSL signaling in these latter two contexts. Similarly, in the embryo, where Neur is required for most DSL signaling events, Dmib has little or no apparent role. Indeed, embryos devoid of Dmib activity hatch as viable first instar larvae; moreover they develop into pharate adults that show only a limited subset of Notch-related mutant phenotypes, each of which appears to reflect the failure of a particular DSL signaling event that does not, normally, depend on Neur. Thus, it is inferred that Dmib and Neur share a common ubiquitin ligase activity that is essential for DSL ligands to signal (Wang, 2005).

Le Borne (2005) has published similar findings indicating a role for Dmib in DSL signaling, and the capacity of ectopic Neur to substitute for Dmib during wing development. The results differ, however, in that this analysis of clones of dmib^- cells that express Ser, Dl, or the Dl^R+ chimera appears to indicate an absolute requirement for Dmib in sending both Drosophila DSL signals, Dl and Ser. By contrast, Le Borgne interprets his data as evidence for a regulatory rather than an obligatory role of Dmib in sending DSL signals, as well as for a lesser role of Dmib in Dl signaling compared with Ser signaling. Differences in experimental design, particularly in the means used to define or infer the identity of DSL signaling and Dmib-deficient cells, could account for the different conclusions reached (Wang, 2005).

It is noted that if Mib and Neur ligases have overlapping molecular functions in all animal systems, as they have in Drosophila, there is no compelling reason why the ligases would need to be deployed in the same way in different animal species. Instead, any given DSL-signaling process might depend on Mib in one animal system but on Neur in another, as appears to be the case for neurogenesis in zebrafish and Drosophila (Wang, 2005).

In contrast to the selective requirement for Dmib and Neur in overlapping subsets of DSL signaling contexts, Epsin is required for most or all DSL signaling events. This difference is expected if ubiquitination of DSL ligands by either Dmib or Neur is normally a prerequisite for Epsin-mediated endocytosis, and hence for signaling activity (Wang, 2005).

Do Dmib and Neur directly bind and ubiquitinate DSL ligands and thereby confer signaling activity by targeting them for Epsin-mediated endocytosis? Although ectopic Dmib and Neur activity are both associated with enhanced DSL ubiquitination, endocytosis and signaling, there is still no compelling evidence that either ligase directly binds and ubiquitinates DSL proteins, or that Dmib/Neur-dependent ubiquitination of DSL ligands confers signalng activity. However, this study shows that the obligate requirement for Dmib for Dl-signaling by wing cells can be bypassed by replacing the cytosolic domain of Dl with a random peptide, R⁺, that may serve as the substrate for ubiquitination by an unrelated ubiquitin ligase. This result provides in vivo evidence that Dmib/Neur-dependent ubiquitination of DSL ligands is normally essential to confer signaling activity. Moreover, the failure of the chimeric Dl^R+ ligand to bypass the requirment for Epsin, supports the interpretation that ubiquitination of DSL ligands confers signaling activity because it targets them for Epsin-mediated endocytosis (Wang, 2005).

During wing development, Ser and Dl both serve as unidirectional signals that specify the 'border' cell fate in cells across the D-V compartment boundary, and both are found to be equally dependent on Dmib and Epsin function for signaling activity. However, the two ligands differ in the extent to which they accumulate on the cell surface, and to which they are cleared from the surface as a consequence of Dmib and Epsin activity. Specifically, most Ser accumulates in cytosolic puncta rather than on the cell surface, whereas the reverse is the case for Dl. Furthermore, removing either Dmib or Epsin activity results in a dramatic and abnormal retention of Ser on the cell surface, whereas it has no detectable effect on the surface accumulation of Dl. Similar results for Dmib were also obtained by Le Borgne (2005). Thus, it appears that most Ser is efficiently cleared from the cell surface by the actions of Dmib and Epsin, whereas most Dl remains on the cell surface, irrespective of Dmib and Epsin activity. This unexpected difference provides two insights (Wang, 2005).

(1) In a previous analysis of the role of Epsin, focus was placed almost exclusively on Dl endocytosis and signaling and failed to obtain direct evidence that Epsin is required for normal DSL endocytosis, despite the obligate role for Epsin in sending both Dl and Ser signals. Instead, such a requirement could be detected only in experiments in which surface clearance of over-expressed Dl was abnormally enhanced by ectopically co-expressing Neur, or could only be inferred from experiments in which the requirement for Epsin was bypassed by replacing the cytosolic domain of Dl with the well-characterized endocytic recycling signal from the mammalian low density lipoprotein (LDL) receptor. By contrast, the different endocytic behavior of Ser has now allowed direct evidence to be obtained that Dmib and Epsin are both required for normal DSL endocytosis (Wang, 2005).

(2) It was found that even though bulk endocytosis of Ser depends on both Dmib and Epsin activity, neither requirement appears absolute. Instead, the accumulation of Ser can still be detected in cytosolic puncta in both Dmib- and Epsin-deficient cells. Moreover, a difference is detected in the abnormal cell surface accumulation of Ser in Dmib-deficient versus Epsin-deficient cells; significantly more Ser appears to accumulate in the absence of Dmib than in the absence of Epsin. It is inferred that both Dl and Ser are normally internalized by multiple endocytic pathways, only some of which depend on ubiquitination of the ligand, and only a subset of these that depends on Epsin. However, the two ligands normally utilize these pathways to different extents, most Ser being internalized by ubiquitin- and Epsin-dependent pathways, and most Dl being internalized by alternative pathways. It is presumed that this difference reflects the presence of different constellations of internalization signals in the two ligands, especially the presence of signals in Delta, but not Ser, that target the great majority of the protein for internalization pathways that do not depend on ubiquitination or Epsin. Nevertheless, only those molecules of Ser and Dl that are targeted by ubiquitination to enter the Epsin-dependent pathway have the capacity to activate Notch; all other routes of entry that are normally available appear to be non-productive in terms of signaling. These results reinforce previous evidence that endocytosis of DSL ligands, per se, is not sufficient to confer signaling activity; instead, DSL ligands must normally be internalized via the action of Epsin to signal (Wang, 2005).

Why must DSL ligands normally be internalized by an Epsin-dependent endocytic mechanism to activate Notch? Two general classes of explanation are considered. In the first, Epsin confers signaling activity by regulating an early event in DSL endocytosis that occurs before internalization. For example, Epsin might cluster DSL ligands in a particular way or recruit them to a select subset of coated pits or other endocytic specializations. Alternatively, Epsin-mediated invagination of these structures might control the physical tension across the ligand/receptor bridge linking the sending and receiving cell, creating a sufficiently strong or special mechanical stress necessary to induce Notch cleavage or ectodomain shedding. In the second class of models, Epsin acts by regulating a later event in DSL endocytosis that occurs after internalization. For example, Epsin might direct, or accompany, DSL proteins into a particular recycling pathway that is essential to convert or repackage them into ligands that can activate Notch upon return to the cell surface. In both cases, internalization of DSL ligands via the other endocytic routes normally available to them would not provide the necessary conditions, even in the extreme case of Dl, which appears to be internalized primarily by these other pathways (Wang, 2005).

The present results do not distinguish between these models. However, recent studies of Epsin-dependent endocytosis in mammalian tissue culture cells suggest that Epsin may direct cargo proteins to different endocytic specializations or pathways, depending on their state of ubiquitination. They also suggest that interactions between Epsin and components of the core Clathrin endocytic machinery normally regulate where and how Epsin internalizes target proteins. Both properties might govern how DSL proteins are internalized, allowing the ligands to gain access to the select endocytic pathway they need to enter to activate Notch (Wang, 2005).

Auxilin is essential for Delta signaling

Endocytosis regulates Notch signaling in both signaling and receiving cells. A puzzling observation is that endocytosis of transmembrane ligand by the signaling cells is required for Notch activation in adjacent receiving cells. A key to understanding why signaling depends on ligand endocytosis lies in identifying and understanding the functions of crucial endocytic proteins. One such protein is Epsin (Drosophila Liquid facets), an endocytic factor first identified in vertebrate cells. This study shows in Drosophila that Auxilin, an endocytic factor that regulates Clathrin dynamics, is also essential for Notch signaling. Auxilin, a co-factor for the ATPase Hsc70, brings Hsc70 to Clathrin cages. Hsc70/Auxilin functions in vesicle scission and also in uncoating Clathrin-coated vesicles. Like Epsin, Auxilin is required in Notch signaling cells for ligand internalization and signaling. Results of several experiments suggest that the crucial role of Auxilin in signaling is, at least in part, the generation of free Clathrin. These observations in the light of current models for the role of Epsin in ligand endocytosis and the role of ligand endocytosis in Notch signaling (Eun, 2008).

A role for Clathrin in Notch signaling cells was originally inferred from the observation that Chc mutants are strong dominant enhancers of lqf hypomorphs. Since Epsin has both Ubiquitin- and Clathrin-binding motifs, and also binds the plasma membrane, the simplest scenario imaginable for Clathrin and Epsin function in Delta internalization is for Epsin to act as a Clathrin adapter that recognizes ubiquitinated Delta, and brings Clathrin to the membrane for CCV formation. However, in light of evidence that Epsin-dependent endocytosis of ubiquitinated transmembrane proteins such as Delta may not occur through formation of CCVs, it has become unclear how to interpret the Chc/lqf genetic interaction. The results presented in this study point to a crucial role for Clathrin in Notch signaling cells. One intriguing possibility is that Delta internalization depends on Clathrin not because Delta is endocytosed in CCVs, but because Clathrin is a positive regulator of Epsin function. More experiments are required to test this idea (Eun, 2008).

Why do tissues that lack Epsin or Auxilin display Delta-like phenotypes, rather than phenotypes indicating failure of many signaling pathways or even cell death? One possibility is that the apparent specificity of both Epsin and Auxilin might simply reflect the usual redundancy of endocytic protein functions, and an unusual dependence of Notch signaling on efficient endocytosis. Alternatively, a special function of Epsin might be crucial to Notch signaling cells. Two kinds of models have been proposed to explain why Notch signaling requires ligand endocytosis by the signaling cells. One idea (the 'pulling model') is that after receptor binding, ligand endocytosis generates mechanical forces that result in cleavage of the Notch intracellular domain (Notch activation), either by exposing the proteolytic cleavage site on the Notch extracellular domain, or by causing the heterodimeric Notch receptor to dissociate. Alternatively, ligand internalization prior to receptor binding might be required to process the ligand endosomally, and recycle it back to the plasma membrane in an activated form (the 'recycling model'). Epsin might generate an environment particularly conducive to either pulling or recycling, and Auxilin might be required specifically by Notch signaling cells because it activates Epsin, perhaps by providing free Clathrin. Alternatively, Auxilin might be needed to provide free Clathrin because Delta is internalized through CCVs. In this case, if Auxilin is required in Notch signaling solely to provide free Clathrin, the implication would be that efficient CCV uncoating is not important for generating uncoated Delta-containing vesicles per se, which are prerequisite for travel through an endosomal recycling pathway. Further understanding of the role of Auxilin in Notch signaling cells might be key to understanding the role of ligand endocytosis (Eun, 2008).

Drosophila Epsin's role in Notch ligand cells requires three Epsin protein functions: the lipid binding function of the ENTH domain, a single Ubiquitin interaction motif, and a subset of the C-terminal protein binding modules

Epsin is an endocytic protein that binds Clathrin, the plasma membrane, Ubiquitin, and also a variety of other endocytic proteins through well-characterized motifs. Although Epsin is a general endocytic factor, genetic analysis in Drosophila and mice revealed that Epsin is essential specifically for internalization of ubiquitinated transmembrane ligands of the Notch receptor, a process required for Notch activation. Epsin's mechanism of function is complex and context-dependent. Consequently, how Epsin promotes ligand endocytosis and thus Notch signaling is unclear, as is why Notch signaling is uniquely dependent on Epsin. By generating Drosophila lines containing transgenes that express a variety of different Epsin deletion and substitution variants, tests were performed of each of the five protein or lipid interaction modules for a role in Notch activation by each of the two ligands, Serrate and Delta. There are five main results of this work that impact present thinking about the role of Epsin in ligand-expressing cells. First, it was discovered that deletion or mutation of both Ubiquitin interaction motifs (UIM) destroyed Epsin's function in Notch signaling and had a greater negative impact on Epsin activity than removal of any other module type. Second, only one of Epsin's two UIMs was essential. Third, the lipid-binding function of the Epsin-N-terminal homology (ENTH domain) was required only for maximal Epsin activity. Fourth, although the C-terminal Epsin modules that interact with Clathrin, the adapter protein complex AP-2, or endocytic accessory proteins were necessary collectively for Epsin activity, their functions were highly redundant; most unexpected was the finding that Epsin's Clathrin binding motifs were dispensable. Finally, it was found that signaling from either ligand, Serrate or Delta, required the same Epsin modules. All of these observations are consistent with a model where Epsin's essential function in ligand-expressing cells is to link ubiquitinated Notch ligands to Clathrin-coated vesicles through other Clathrin adapter proteins. It is proposed that Epsin's specificity for Notch signaling simply reflects its unique ability to interact with the plasma membrane, Ubiquitin, and proteins that bind Clathrin (Xie, 2012).

Epsin is a complex multi-modular protein that functions differently in different contexts. Each Lqf isoform has two UIMs, two Clathrin binding motifs (CBMs), seven DPW motifs that bind the AP-2 endocytic adapter complex, and two NPF motifs that bind EH-domain-containing endocytic factors such as Eps15. In C. elegans, Drosophila, and mice, Epsin is needed specifically in Notch ligand cells. The structure/function analysis of Epsin performed in this study shows that modules of Epsin associate with the internalization step of endocytosis - the lipid binding function of the ENTH domain and the C-terminal modules that bind proteins present in Clathrin-coated vesicles - are required for Epsin's function in Notch ligand cells. In addition, it was shown that a UIM is necessary (Xie, 2012).

The dispensability of the Cdc42 GAP binding function of the ENTH domain suggests that in ligand cells the primary role of Drosophila Epsin, unlike yeast Ent1, is not regulation of actin dynamics. The other known function of the ENTH domain is the endocytic function, and the results suggest that the ability of the ENTH domain to interact with PIP2 explains why it is needed for maximal Epsin function in Notch ligand cells. These observations are consistent with the lack of typical Notch signaling defects in Drosophila cdc42 mutants. In contrast, flies with mutations in genes for either of two actin regulators, the Arp2/3 complex and WASp, do have notal bristle defects indicative of Notch signaling failure. The notal bristle phenotype described in this study is not due to failure of the Epsin-dependent endocytosis of ligand that activates Notch in all cell types, but instead to failure of ligand transcytosis required in only some cell types to relocalize ligand prior to signaling. The absence of the Arp2/3 complex or WASp in mutants inhibits signaling by blocking traffic of endocytosed Delta to apical microvilli of sensory organ precursors. Whether or not Delta transcytosis in sensory organ precursors also depends on Epsin is unknown. If Epsin is involved, it may be interesting to use the Epsin variant transgenes generated in this study to determine whether or not the Cdc42 GAP interaction function of the ENTH domain is required (Xie, 2012).

There are two types of UIMs: single-sided UIMs that bind one Ubiquitin, and double-sided UIMs that bind two Ubiquitins simultaneously. As the affinity between a UIM and Ubiquitin is low, successful interaction between a mono-ubiquitinated protein and a UIM-containing protein is thought to require either one double-sided UIM, or two single-sided UIMs. Epsins have single-sided UIMs, and so the observation that only one single-sided UIM is required for Drosophila Epsin function in Notch signaling is unexpected. The simplest explanation is that Notch ligands use multiple mono-Ubiquitins or Ubiquitin chains as a signal for Epsin-mediated internalization (Xie, 2012).

Two distinct Lysine residues in the intracellular domains of both Delta and Serrate have been implicated as important for the function of each ligand. In the case of Serrate, simultaneous mutation of both of these Lysines results in a Serrate ligand that can neither activate Notch nor be endocytosed in wing discs. These observations identify two particular Lysines as candidates for the critical Ub attachments, but do not distinguish whether one or both Lysines are required. In the case of Delta, single mutation of either of two specific Lysines results in accumulation of Delta at the cell surface of eye discs and failure to signal. Although Delta is thought to be mono-ubiquitinated, these results suggest the possibility that Delta is multiply mono-ubiquitinated. An alternative explanation for Epsin's ability to promote ligand endocytosis with a single UIM is that mono-ubiquitinated ligands cluster to generate an environment where multiple Ubiquitins attract Epsin to ligand at the plasma membrane (Xie, 2012).

There is compelling evidence that in somatic cells, Notch ligand endocytosis associated with signaling is Clathrin-dependent. First, there are exceedingly strong genetic interactions between the Clathrin heavy chain (Chc) gene and lqf, the gene for Epsin. Flies with only one Chc+ gene copy are wild-type, but this condition is lethal in homozygotes for a normally viable hypomorphic allele of lqf. Second, the Clathrin-coated vesicle uncoating protein Auxilin is, like Epsin, required specifically for Notch signaling in Drosophila and in ligand cells. Given the clear involvement of Clathrin and the lack of strong genetic interaction between α-Adaptin (the gene for an AP-2 subunit) and lqf, the simplest model for Epsin function in Notch signaling was as an adapter protein that links Clathrin and the plasma membrane, independent of AP-2. This model predicted that direct interaction between Epsin and Clathrin would be necessary, and thus the most surprising result of this work is that deletion of the CBMs had no detectable effect on Epsin activity. The dispensability of the CBMs rules out models where Epsin acts as a monomeric Clathrin adapter that links ligand to Clathrin cages (Xie, 2012).

In the Drosophila female germline, Notch signaling requires Epsin but neither Clathrin nor Auxilin. Although this is surprising, Epsin has been shown to function in Clathrin-independent internalization of ubiquitinated transmembrane cargos in vertebrate cell culture. Epsin must therefore function differently in Notch signaling in the female germline than in somatic cells. It is speculated that the ENTH domain and UIMs may be required in germline cells to guide the ubiquitinated proteins into Q6 an endocytic vesicle. However, it is not clear how any of the characterized modules within Epsin's C-terminus might be involved in Clathrin-independent endocytosis. It would be of interest to use the transgenes that were generated in this study to determine which motifs are required in the female germline. Additional experiments could potentially identify unknown C-terminal interaction motifs used in Clathrin-independent endocytosis (Xie, 2012).

Does Epsin function in the same way in the embryo, eye, and wing? The experiments began with the assumption that Epsin functions through the same mechanism in all signaling contexts, and thus it was expected the same Epsin modules would be required for Epsin function in all contexts. Epsin appears to be required in every Notch signaling event and thus could be regarded as a core component of Notch signaling. It therefore seems reasonable to expect that Epsin would function in the same manner in all tissues. The female germline is apparently an exception. Nevertheless, in the three assays used for Epsin activity - rescue of lethality and eye morphology defects due to lqf mutations and rescue of the ability of lqf null cells to activate Cut expression in cells at the D/V boundary in the wing disc - only subtle differences were detected between the eye and the wing in the activity of two Epsin variants, δENTH and δUIM. (The only major difference was with the highly artificial Epsin variant, 4XNPF.) Despite these differences, it is thought that Epsin likely functions the same way in the eye and wing, as well as during embryogenesis. For one, the differences in activity that were observed be explained easily without invoking different mechanisms for Epsin in the eye and wing. Importantly, no even one case was observed where modules were essential in one context (embryogenesis, eye, or wing development) and dispensable in another one. In fact, it is possible to observe all-or-none differences in requirements for Epsin modules. Epsin was found to function outside of Notch ligand cells and modules were found that were dispensable completely in this context yet absolutely essential for Epsin's function in ligand cells (Xie, 2012).

Notch ligands require ubiquitination and (usually) Clathrin-dependent endocytosis, and formation of Clathrin-coated vesicles requires adapter proteins that link the plasma membrane with Clathrin. The absolute necessity of at least one UIM and the observation that the lipid-binding function of the ENTH domain plays a role in ligand cells suggests that Epsin indeed binds ubiquitinated Notch ligands at the plasma membrane.However, as an Epsin derivative lacking CBMs functions as well as wild-type Epsin in ligand cells, the essential role of Epsin in Notch signaling cannot be as a monomeric Clathrin adapter that links Clathrin directly to ligand at the plasma membrane. As any pair of the three types of modules is sufficient for Epsin function (CBMs+DPWs, CBMs+NPFs, or DPWs+NPFs), Epsin must be able to support Notch activation by linking ligand to Clathrin in a variety of different ways. It is speculated that Eps15, the second Drosophila Epsin, is involved because of the three EH-domain proteins in Drosophila (Eps15, Dap160, Past1), none have Clathrin binding motifs, and Eps15 is the only one with motifs for a known Clathrin-binding protein (AP-2). From analysis of mutant phenotypes and genetic interaction studies, there is no evidence for the involvement of Eps15 nor AP-2 in Notch signaling (Xie, 2012). The results presented in this study suggest that Eps15 and AP-2 may play redundant roles in the presence of intact Epsin and this idea could be tested with additional genetic experiments. In light of the evidence indicating a requirement for Clathrin in ligand cells (outside of the germline), the results suggest that Epsin is required absolutely for Notch signaling not because it generates a special endocytic environment, but simply because it is the only UIM-containing endocytic protein with the appropriate complement of interaction modules to target ubiquitinated cargo to Clathrin-coated vesicles (Xie, 2012).

DEVELOPMENTAL BIOLOGY

Embryonic

Liquid facets acts on fusion-competent myoblasts to prevent their acquisition of the cardioblast fate

Endocytosis and trafficking within the endocytosis pathway are known to modulate the activity of different signaling pathways. Epsins promote endocytosis and are postulated to target specific-proteins for regulated endocytosis. A functional link is presented between the Notch pathway and epsins. The Drosophila epsin liquid facets has been identified as an inhibitor of cardioblast development in a genetic screen for mutants that affect heart development. lqf inhibits cardioblast development and promotes the development of fusion-competent myoblasts, suggesting a model in which lqf acts on or in fusion-competent myoblasts to prevent their acquisition of the cardioblast fate. lqf and Notch exhibit essentially identical heart phenotypes, and lqf genetically interacts with the Notch pathway during multiple-Notch-dependent events in Drosophila. The link between the Notch pathway and epsin function is extended to C. elegans, where the C. elegans lqf ortholog acts in the signaling cell to promote the glp-1/Notch pathway activity during germline development. These results suggest that epsins play a specific, evolutionarily conserved role to promote Notch signaling during animal development and support the idea that they do so by targeting ligands of the Notch pathway for endocytosis (Tian, 2003).

Phenotypic studies of lqf embryos are consistent with the model that lqf acts in a subpopulation of fusion competant myoblasts (FCMs) to inhibit their acquisition of the cardioblast fate. How might lqf and Notch inhibit cardioblast development? In lateral and ventral regions of the mesoderm, Notch-mediated lateral inhibition helps select individual somatic muscle progenitor cells from clusters of equipotential cells. Cells in these clusters express lethal of scute (l'sc) and can adopt either the muscle progenitor or FCM fate. Cells that retain l'sc become progenitor cells while cells that lose l'sc expression become FCMs. In these clusters, Notch inhibits l'sc expression and the progenitor fate, thereby promoting the FCM fate. It is speculated that lqf and Notch may act similarly to regulate the cardioblast progenitor/FCM decision in the dorsal mesoderm with Notch functioning to inhibit tinman expression and the cardioblast progenitor fate and, in so doing, promoting the FCM fate. In this model, loss of lqf/Notch activity would lead to excess cardioblast progenitors at the expense of FCMs. Consistent with this, clusters of Tin-expressing cells in the dorsal mesoderm resolve to individual heart cells during the stages when lqf and Notch inhibit cardioblast development (Tian, 2003).

If lqf plays a general role in the promotion of Notch activity, why are defects observed only during heart development in lqf embryos? One explanation is that maternal lqf product masks earlier requirements for lqf during Notch-dependent events. Consistent with this, temperature shift experiments indicate lqf acts during late stage 12 to restrict cardioblast development. Nearly all well-characterized Notch-dependent events in the embryo occur before stage 12. Thus, the apparent specificity of the lqf phenotype for the heart may simply arise due to the late stage at which this Notch-dependent event occurs, combined with the masking effect of lqf maternal product. Unfortunately, it was not possible to assay embryos devoid of maternal and zygotic lqf function since lqf germline clones failed to produce eggs (Tian, 2003).

Regarding the function of C. elegans epsin during postembryonic development, upon Ce-epn-1 RNAi treatment, a weak glp-1 loss-of-function germline phenotype was observed, which was significantly enhanced in a glp-1 temperature-sensitive background at the permissive temperature. Strong lin-12 loss-of-function phenotypes were not observed in these experiments. The weak glp-1 loss-of-function and absence of lin-12 phenotypes in a wild-type background is very likely because the postembryonic feeding RNAi treatment only partially depletes Ce-epn-1 mRNA. The isolation and characterization of a null mutation will greatly facilitate uncovering all roles of Ce-epn-1 in C. elegans (Tian, 2003).

Since epsins appear to regulate endocytosis, and endocytosis regulates the activity of most signaling pathways (reviewed by Wendland, 2002), lqf might act broadly to regulate the output of many signaling pathways rather than acting specifically on the Notch pathway. However, existing data support a specific interaction between lqf and Notch activity. For example, no genetic interactions were observed between lqf and the EGF or FGF pathways during heart development, and lqf does not appear to interact with the dominant Egfr[ellipse] allele during eye development. Furthermore, lqf mutant clones in the Drosophila eye exhibit phenotypes consistent with the specific loss of Notch activity. Thus, lqf appears to display specific interaction with the Notch pathway. Although it is important to assay whether lqf alters the activity of other signaling pathways regulated by endocytosis, such as the TGFß, Wingless, and Hedgehog pathways, these data support a model in which Lqf plays a relatively specific role to target a component of the Notch pathway for endocytosis and in so doing promotes Notch signaling (Tian, 2003).

This work shows that lqf/epsins promote Notch pathway activity in Drosophila and C. elegans. Notably, epsins participate in Notch-mediated lateral inhibition signaling during bristle and perhaps heart development, as well as Notch-mediated inductive signaling in the C. elegans germline. These data argue that epsins are essential evolutionarily conserved components of the Notch pathway, potentially required for most if not all Notch-mediated processes (Tian, 2003).

Genetic studies have the potential to identify all components of a signaling process, however, they do not necessarily differentiate between the roles different genes play in a signaling process. A distinction can be made between core components of a signaling pathway -- those factors that actively take part in the signal transduction event -- and factors that set the stage for signal transduction but do not actively transmit the signal. For example, Notch, DSL ligands, presenillins and CSL effectors are core components of the Notch pathway as they actively transmit the signal -- DSL ligands bind Notch, induce the metalloprotease-mediated S2 cleavage followed by the presenilin dependent intramembrane (S3) cleavage of Notch that releases Notch[intra], which translocates to the nucleus and complexes with CSL-class proteins to activate Notch target genes. However, many other proteins set the stage for signaling by ensuring each core member of a pathway is present in the right location and correct form, such that signal transduction will occur given the proper stimulus. For example, presentation of a functional Notch receptor on the cell membrane appears to require S1-mediated cleavage of Notch by furin-type proteins. Although furins do not actively take part in the signaling event, furin activity and its requirement for presentation of Notch is a prerequisite for Notch signaling. Similarly, ras signaling requires Ras localization to the cell membrane and prenylation of Ras constitutively targets it to the cell membrane (Tian, 2003).

In support of lqf/epsins as core components of the Notch pathway, Delta endocytosis appears essential for Notch signaling and lqf appears essential for Delta endocytosis. Furthermore, epsins are thought to target ubiquitinated membrane proteins for regulated endocytosis via their ubiquitin-interacting motif (UIM) and ubiquitination of Delta appears necessary for Delta endocytosis and active Notch signaling. Thus, Lqf/epsins may act as part of a complex that specifically targets Delta for internalization after ubiquitination and as such be core members of the Notch pathway (Tian, 2003).

It remains possible, however, that lqf/epsin function is a prerequisite for Notch signaling. For example, some endocytic proteins function in protein transport in the secretory pathway and epsin1 family members could in principle enable transport of Delta to the membrane. In such a capacity, epsins would not be considered core components of the Notch pathway. Clearly, future experiments that test the requirement of specific domains of epsins, such as the UIM, for Notch signalling, as well as those that identify the protein complexes within which epsins act, should help elucidate the molecular basis through which lqf/epsins potentiate Notch signaling (Tian, 2003).

EFFECTS OF MUTATION

Genetic interaction studies show that Liquid facets is a candidate for the critical substrate of Fat facets in the eye

Fat facets is a deubiquitinating enzyme required in a cell communication pathway that limits to eight the number of photoreceptor cells in each facet of the Drososphila compound eye. Genetic data support a model whereby Faf removes ubiquitin, a polypeptide tag for protein degradation, from a specific ubiquitinated protein, thus preventing its degradation. Mutations in the liquid facets gene have been identified as dominant enhancers of the fat facets mutant eye phenotype. The liquid facets locus encodes epsin, a vertebrate protein associated with the clathrin endocytosis complex. Genetic experiments reveal that fat facets and liquid facets facilitate endocytosis and function in common cells to generate an inhibitory signal that prevents ectopic photoreceptor determination. The fat facets mutant phenotype is extraordinarily sensitive to the level of liquid facets expression. It is proposed that Liquid facets is a candidate for the critical substrate of Fat facets in the eye (Cadavid, 2000).

There are three key components of the endocytosis complex: (1) clathrin, which forms a cage structure engulfing the cell membrane; (2) AP-2, the core adaptor complex, which binds to clathrin and brings it to the cell surface, and (3) dynamin, a GTPase required for vesicle formation. Additional proteins associated with AP-2 have been identified, many of which contain protein-protein interaction domains called EH-domains and EH-domain-binding motifs. Epsin is an EH-domain-binding protein identified as a partner for Eps15, an EH-domain protein that also binds AP-2. The large number of AP-2-binding proteins identified suggests that many of them may have temporal and/or tissue-specific functions (Marsh and McMahon, 1999). The precise roles of Eps15 and epsin in endocytosis are unknown (Cadavid, 2000 and references therein).

The lqf gene itself is essential in Drosophila; in an otherwise wild-type background, lqf null mutants die as embryos. Clones of cells in the eye in which there is little or no lqf gene function have severely disrupted eye morphology, indicating that lqf is required also after embryogenesis for eye development. The mutant phenotypes associated with two weak lqf mutant alleles reveal specific roles for lqf in eye, wing and leg development (Cadavid, 2000).

The eye defects in homozygous adults for a weak allele of lqf resemble those in faf null mutants. As in faf mutants, the additional photoreceptors in lqf mutants arise from specific precursor cells (M-cells) present early during eye development. In contrast to lqf null mutants, faf null mutants are viable, have normal wings and legs and have less severe eye defects. Thus, lqf functions more broadly than faf, but both the lqf and faf genes are required during eye development in order to prevent the M-cells from becoming photoreceptors (Cadavid, 2000).

The faf mutant eye phenotype is unusually sensitive to a decrease in the dose of the lqf gene, suggesting strongly that the two genes function in a common pathway. Genetic interactions with endocytosis and Ub pathway mutants show that faf and lqf facilitate endocytosis and antagonize ubiquitination. In addition, although lqf is more broadly required than faf in the eye and elsewhere in the fly, weak lqf mutations reveal that like faf, lqf is required to prevent the misdetermination of M-cells as photoreceptors. Moreover, when expressed only in the rough positive cells surrounding the facet preclusters, both faf and lqf genes rescue completely to wild-type their respective M-cell misdetermination mutant phenotypes in the eye. Finally, given the relationship between Faf and its substrate protein, it would be expected that increasing the dose of the substrate should suppress the faf mutant phenotype. Slight overexpression of lqf completely obviates the need for faf in eye development. The simplest model consistent with all of this genetic data is that Lqf is the substrate of Faf. Other more complicated explanations are, of course, possible (Cadavid, 2000).

There is biochemical evidence that AF-6, a scaffolding protein thought to modulate cell-cell junctions in response to Ras activation may be an in vivo substrate of Fam, the mouse homolog of Faf; AF-6 and Fam bind each other in vitro and ubiquitinated AF-6 can be detected and deubiquitinated by Fam in cultured cells. Like lqf, the Drosophila Af6 homolog, canoe, is required pleiotropically for Drosophila eye development. In contrast to lqf mutations, however, canoe mutations do not act as strong dominant enhancers of the faf mutant eye phenotype. Given the striking genetic interactions between faf and lqf, it seems that canoe is unlikely to play a significant role in the essential faf pathway in the eye (Cadavid, 2000 and references therein).

While only one Faf/substrate interaction may be essential to normal eye development in Drosophila, Faf and Fam may have several substrates in vivo. Normally non-essential roles for faf later in eye development have been revealed in particular mutant backgrounds and Faf could have different substrates for its critical role in M-cell fate determination than in its redundant roles. Moreover, in addition to its essential role in eye development, faf is required maternally for cellularization of embryos and the critical maternal substrate of Faf is unknown. Because faf has mouse and human homologs, the modes of regulation by Faf are likely to be conserved. However, it is possible that the critical substrate(s) of Faf in Drosophila may differ from those in vertebrates (Cadavid, 2000 and references therein).

If Lqf is the substrate of Faf, then epsin levels, determined by the balance between its ubiquitination and deubiquitination, could regulate endocytosis. Mono-ubiquitination, however, has been shown previously to regulate endocytosis in two different ways. (1) Mono-ubiquitination of cell surface receptors can act as a signal for receptor endocytosis, which leads to lysosomal degradation. Here, the Ub moiety is somehow recognized by the endocytosis machinery; this process has nothing to do with the proteasome. (2) Eps15, an endocytosis complex component in mammalian cells, is mono-ubiquitinated in response to EGF receptor activation and Eps15 may require this modification to stimulate receptor endocytosis. In addition, Pan1p, a yeast protein similar to Eps15, is required for endocytosis in yeast. Although it is unknown whether Pan1p is mono-ubiquitinated in yeast, there is evidence that ubiquitination of an endocytic complex component is required for endocytosis in yeast; Rsp5p, a component of the ubiquitination machinery called a ubiquitin-ligase, may bind to Pan1p and is required generally for endocytosis in yeast, even for endocytosis of proteins with non-Ub endocytosis signals (Cadavid, 2000 and references therein).

Since Eps15 binds to epsin, could a mono-ubiquitinated Drosophila Eps15 homolog be the substrate of Faf? Two experimental results are inconsistent with this model. (1) It has been shown previously that the activity of Faf antagonizes proteolysis, not just ubiquitination; mutations in a gene encoding a proteasome subunit act as strong suppressors of the faf mutant eye phenotype. This result strongly suggests that Faf activity antagonizes proteolysis and thus that Faf deubiquitinates a protein containing a Ub chain targeting it for degradation, rather than a mono-ubiquitinated protein. (2) If mono-ubiquitination of Eps15 activates it, as the available data suggests, then deubiquitination of Eps15 by Faf would render Eps15 inactive and thus the function of Faf would antagonize endocytosis. The data presented here clearly indicate the opposite; mutations in endocytosis complex genes (particularly lqf and Clathrin heavy chain) act as strong dominant enhancers of faf, suggesting that the normal function of Faf is to facilitate endocytosis (Cadavid, 2000 and references therein).

Elevated levels of Lqf obviate the need for Faf, presumably by stimulating epsin-dependent endocytosis generally or stimulating endocytosis of a specific cell surface protein. How can the observation that Lqf and Faf function outside the M-cells to determine M-cell fate be reconciled with a role for Lqf in endocytosis? Endocytosis is known to modulate ligand/receptor interactions by a variety of mechanisms. One possibility is that M-cell fate is affected by a diffusible ligand that, like Wingless, travels via endocytosis through several cell distances. Alternatively, regulation of a membrane-bound receptor by endocytosis in cells adjacent to the M-cells could affect M-cell fate indirectly. For example, EGF receptor activity is downregulated by endocytosis following ligand binding. By contrast, activity of the Notch receptor may be up-regulated by endocytosis of activated receptors whose intracellular domains have been cleaved off prior to their translocation into the nucleus. Membrane-bound Notch receptors lacking their intracellular domains display dominant negative activity and endocytosis of cleaved Notch receptors may be required normally for precise modulation of Notch activity. Patterning of the photoreceptor preclusters in the developing eye may require that both Notch and the EGF receptor are activated in the rough-expressing cells surrounding the facet preclusters. Thus, Faf could regulate the activity of one or both of these receptors (Cadavid, 2000 and references therein).

Liquid facet functions in endocytosis of Delta in the developing eye

Epsin is part of a protein complex that performs endocytosis in eukaryotes. Drosophila epsin, Liquid facets (Lqf), was identified because it is essential for patterning the eye and other imaginal disc derivatives. Previous work has provided only indirect evidence that Lqf is required for endocytosis in Drosophila. Epsins are modular and have an N-terminal ENTH (epsin N-terminal homology) domain that binds PIP2 at the cell membrane and four different classes of protein-protein interaction motifs. The current model for epsin function in higher eukaryotes is that epsin bridges the cell membrane, a transmembrane protein to be internalized, and the core endocytic complex. This study shows directly that Drosophila epsin (Lqf) is required for endocytosis. Specifically, Lqf is essential for internalization of the Delta (Dl) transmembrane ligand in the developing eye. Using this endocytic defect in lqf mutants, a transgene rescue assay has been developed and a structure/function analysis of Lqf has been performed. When Lqf is divided into two pieces, an ENTH domain and an ENTH-less protein, each part retains significant ability to function in Dl internalization and eye patterning. These results challenge the model for epsin function that requires an intact protein (Overstreet, 2003).

To test for endocytosis defects in lqf^- mutants, the localization of the transmembrane receptor Dl was monitored in developing eyes. Dl normally undergoes endocytosis in the eye, and as the internalized protein is not degraded rapidly, internalized Dl can be detected in vesicles (Overstreet, 2003).

The Drosophila eye, composed of 800 identical 22-cell ommatidia, or facets, develops in the larval and pupal stages in a monolayer epithelium called the eye imaginal disc. Eye development occurs as a wave, where the morphogenetic furrow forms at the posterior of the disc, and moves anteriorly into the monolayer of undifferentiated cells. Rows of ommatidia assemble stepwise posterior to the furrow one or two cells at a time, starting with the eight photoreceptors (R1-R8) (Overstreet, 2003).

In wild-type, Dl is detected exclusively as intracellular dots within developing ommatidial clusters throughout the eye disc. In larval eye discs homozygous for lqf^FDD9, a weak, viable mutant allele, Dl is detected mainly at the membranes of cells just posterior to the furrow. In clones of cells homozygous for lqf^ARI, a strong, lethal mutant allele, similar membrane localization of Dl is observed. It is concluded that lqf⁺ is required for Dl internalization (Overstreet, 2003).

All epsins have an amino-terminal ENTH domain that binds PIP₂ at the cell membrane and three or four types of protein-protein interaction motifs, whose copy numbers vary among different epsins. The ubiquitin interaction motifs (UIMs) bind ubiquitin (Ub) noncovalently. There are also clathrin binding motifs (CBMs), DPW motifs that bind the core endocytic adaptor complex, AP-2, and NPF motifs that bind Eps15, another accessory factor (Overstreet, 2003).

A step toward understanding the role of Lqf in endocytosis is the identification of the modules of Lqf protein that are required. In yeast, there are straightforward assays for the function of the two epsins (Ent1 and Ent2). Structure/function analyses have demonstrated that the ENTH domain of Ent1 is necessary and sufficient to rescue the lethality of ent1Δent2Δ double mutants. Moreover, the ENTH domain and to a lesser extent the UIMs have been shown to be required for endocytosis. Because there are mechanistic differences between endocytosis in yeast and higher eukaryotes, the yeast epsins might function somewhat differently from vertebrate epsins and Drosophila Lqf. The major difference between these systems is that the AP-2 core adaptor complex in yeast has no known function in endocytosis, and, accordingly, the yeast epsins lack DPW motifs. As in yeast, structure/function analyses of epsins in vertebrate cell culture have pointed to the importance of the ENTH domain. These assays, however, rely on dominant-negative effects of mutant epsin proteins on endocytosis, and their interpretation is difficult (Overstreet, 2003).

Either the ENTH domain alone, or an ENTH-less Lqf protein, rescues the patterning and Dl endocytosis defects in lqf^FDD9 homozygous eyes. Since experimental results in yeast and in vertebrates have emphasized the importance of the ENTH domain, the most remarkable result is that an ENTH-less Lqf protein can function. The simplest interpretation of the rescue results is that Lqf^ΔENTH can function independently of the ENTH domain (Overstreet, 2003).

Transgenes that express Rat epsin1 or human epsin 2b in Drosophila with pRO, each as full-length proteins or without the ENTH domain, rescue the eye defects in lqf^FDD9 homozygotes. Thus, there is unlikely to be a significant functional difference between the Drosophila and vertebrate epsins in the region C-terminal to the ENTH domain. In addition, the ENTH domains of Lqf and yeast epsin are functionally similar. It was shown previously that expression of the ENTH domain of Ent1, but not the complementary portion of the protein, restores viability to ent1Δent2Δ yeast. Similarly, expression of full-length Lqf or Lqf^ENTH rescues ent1Δent2Δ lethality but Lqf^ΔENTH expression does not (Overstreet, 2003).

Thus Drosophila epsin, Lqf, is essential for endocytosis of Dl in the developing eye. Moreover, the ENTH domain alone and an ENTH-less Lqf protein each retain significant function. The prevailing model in vertebrates is that epsin functions like a bridge, where the ENTH domain links the membrane to clathrin, a cell surface protein to be internalized, and to AP-2. Since this model requires an intact epsin protein, the results presented here suggest that the prevailing model cannot be the whole story. Moreover, the observation that either the ENTH domain or the remainder of the protein, which are functionally distinct, can be deleted without destroying Lqf function completely suggests that each fragment of Lqf may be partially redundant with another Drosophila endocytic protein. Candidates for the other endycotgic protein include the other ENTH domain protein in Drosophila, Epsin-2 and the Drosophila homolog of AP180, which, like the ENTH-less Lqf protein, binds clathrin and AP-2 (Overstreet, 2003).

Fat facets and Liquid facets promote Delta endocytosis and Delta signaling in the signaling cells

Endocytosis modulates the Notch signaling pathway in both the signaling and receiving cells. One recent hypothesis is that endocytosis of the ligand Delta by the signaling cells is essential for Notch activation in the receiving cells. Evidence is presented in strong support of this model. In the developing Drosophila eye Fat facets (Faf), a deubiquitinating enzyme, and its substrate Liquid facets (Lqf), an endocytic epsin, promote Delta internalization and Delta signaling in the signaling cells. While Lqf is necessary for three different Notch/Delta signaling events at the morphogenetic furrow, Faf is essential only for one: Delta signaling by photoreceptor precluster cells, which prevents recruitment of ectopic neurons. In addition, the ubiquitin-ligase Neuralized (Neur), which ubiquitinates Delta, is shown to function in the signaling cells with Faf and Lqf. The results presented bolster one model for Neur function in which Neur enhances Delta signaling by stimulating Delta internalization in the signaling cells. It is proposed that Faf plays a role similar to that of Neur in the Delta signaling cells. By deubiquitinating Lqf, which enhances the efficiency of Delta internalization, Faf stimulates Delta signaling (Overstreet, 2004).

Cells with decreased lqf⁺ activity accumulate Delta on apical membranes and fail to signal to neighboring cells. Three Notch/Delta signaling events were examined in the eye: proneural enhancement, lateral inhibition and R-cell restriction. Loss of lqf⁺-dependent endocytosis during all three events has identical consequences to loss of Delta function in the signaling cells. It is concluded that lqf⁺-dependent endocytosis of Delta is required for signaling, supporting the notion that endocytosis in the signaling cells activates Notch in the receiving cells. However, Lqf is not required absolutely for all Delta internalization in the eye. Even in lqf-null cells, which are incapable of Delta signaling, some vesicular Delta is present. Perhaps not all of the vesicular Delta present in wild-type discs results from signaling (Overstreet, 2004).

Genetic studies in Drosophila indicate clearly that deubiquitination of Lqf by Faf activates Lqf activity. Moreover, genetic and biochemical evidence in Drosophila suggests that Faf prevents proteasomal degradation of Lqf. In vertebrates, however, it is thought that epsin is mono-ubiquitinated to modulate its activity rather than poly-ubiquitinated to target it for degradation. If Lqf regulation by ubiquitin also occurs this way in the Drosophila eye, the removal of mono-ubiquitin from Lqf by Faf would activate Lqf activity (Overstreet, 2004).

Whatever the precise mechanism, given that both Faf and Lqf are expressed ubiquitously in the eye, two related questions arise. First, why is Lqf ubiquitinated at all if Faf simply deubiquitinates it everywhere? One possibility is that Faf is one of many deubiquitinating enzymes that regulate Lqf, and expression of the others is restricted spatially. This could also explain why Faf is required only for R-cell restriction. Another possibility is that Faf activity is itself regulated in a spatial-specific manner in the eye disc. Alternatively, Lqf ubiquitination may be so efficient that Faf is needed to provide a pool of non-ubiquitinated, active Lqf. Similarly, Faf could be part of a subtle mechanism for timing Lqf activation. Second, why is Faf essential only for R-cell restriction? One possibility is that there is a graded requirement for Lqf in the eye disc, such that proneural enhancement requires the least Lqf, lateral inhibition somewhat more and neural inhibitory signaling by R2/3/4/5 the most. The mutant phenotype of homozygotes for the weak allele lqf^FDD9 supports this idea, as R-cell restriction is most severely affected. Alternatively, Lqf may be expressed or ubiquitinated with dissimilar efficiencies in different regions of the eye disc. More experiments are needed to understand the precise mechanism by which the Faf/Lqf interaction enhances Delta signaling (Overstreet, 2004).

In neur mutants, Delta accumulates on the membranes of signaling cells and Notch activation in neighboring cells is reduced. These results support a role for Neur in endocytosis of Delta in the signaling cells to achieve Notch activation in the neighboring receiving cells, rather than in downregulation of Delta in the receiving cells. Because neur shows strong genetic interactions with lqf and both function in R-cells, Neur and Lqf might work together to stimulate Delta endocytosis. Lqf has ubiquitin interaction motifs (UIMs) that bind ubiquitin. One explanation for how Neur and Faf/Lqf could function together is that Lqf facilitates Delta endocytosis by binding to Delta after its ubiquitination by Neur. This is anattractive model that will stimulate further experiments (Overstreet, 2004).

One exciting observation is that the endocytic adapter Lqf may be essential specifically for Delta internalization. Although, hedgehog, decapentaplegic and wingless signaling pathways have not been examined directly, they appear to be functioning in the absence of Lqf. These three signaling pathways regulate movement of the morphogenetic furrow and are thought to require endocytosis. The furrow moves through lqf-null clones and at the same pace as the surrounding wild-type cells. Moreover, all mutant phenotypes of lqf-null clones can be accounted for by loss of Delta function. Further experiments will clarify whether this apparent specificity means that Lqf functions only in internalization of Delta, or if the process of Delta endocytosis is particularly sensitive to the levels of Lqf (Overstreet, 2004).

Lqf expands the small repertoire of endocytic proteins that are known targets for regulation of cell signaling. In addition to Lqf, the endocytic proteins Numb and Eps15 (EGFR phosphorylated substrate 15) are objects of regulation. In vertebrates, asymmetrical distribution into daughter cells of the alpha-adaptin binding protein Numb may be achieved through ubiquitination of Numb by the ubiquitin-ligase LNX (Ligand of Numb-protein X) and subsequent Numb degradation. In addition, in vertebrate cells, Eps15 is phosphorylated and recruited to the membrane in response to EGFR activation and is required for ligand-induced EGFR internalization. Given that endocytosis is so widely used in cell signaling, endocytic proteins are likely to provide an abundance of targets for its regulation (Overstreet, 2004).

Drosophila Epsin mediates a select endocytic pathway that DSL ligands must enter to activate Notch

In screens for mutations affecting wing pattern, six alleles of a single complementation group were obtained that cause phenotypes similar to those caused by the loss of Notch signaling, namely severe wing notching, wing vein thickening and bristle tufts. All six alleles fail to complement existing alleles of liquid facets (lqf), and are associated with nonsense or missense mutations in the lqf-coding sequence (Overstreet, 2003). One new allele, lqf¹²²⁷, truncates the coding sequence after amino acid 119 in the middle of the ENTH domain, the most N-terminal conserved domain, and abolishes Lqf protein expression in vivo. This allele is referred to as lqf^-, and it was used for all experiments described in this study. A transgene containing the intact lqf gene (Cadavid, 2000) rescues the lethality of lqf^- homozygotes, as well as all of the mutant phenotypes associated with lqf^- clones. Lqf encodes the sole ortholog of vertebrate Epsin1 (Cadavid, 2000); a second Drosophila protein, sometimes referred to as Dm Epsin2 (Overstreet, 2003), lacks several conserved domains found in Lqf and vertebrate Epsin1, and appears instead to be the Drosophila ortholog of vertebrate EpsinR (see Mills, 2003), a functionally distinct Epsin-related protein (Wang, 2004).

In imaginal wing discs, signaling by the DSL ligands Delta (Dl) and Serrate (Ser) specifies the wing margin at the dorsoventral (DV) compartment boundary, and can be assayed by boundary-specific expression of wing margin genes (or their protein products), such as cut, wingless (wg) and vestigial (vg). lqf^- clones resemble Dl^- Ser^- clones or N^- clones in that they cause the loss of cut, wg and vg boundary-specific expression when they abut or cross the DV compartment boundary, corroborating the Notch-related phenotypes of lqf^- clones observed in the adult wing (Wang, 2004).

The loss of margin gene expression in lqf^- clones is not cell autonomous. Instead, wild-type cells can rescue the expression of margin specific genes in adjacent lqf^- cells (e.g., cut). Similarly, non-autonomous rescue of lqf^- clones was observed in the adult, where the presence of wild-type cells can rescue the ability of neighboring lqf^- cells to form single bristles. In both respects, as well as in others, lqf^- clones resemble Dl^- Ser^- clones, but differ from N^- clones, which show a strictly cell-autonomous loss of Notch target gene expression (Wang, 2004).

Collectively, these data establish an obligate role for Lqf in Notch signaling, and implicate Lqf in sending, rather than receiving, DSL signals (Wang, 2004).

To determine whether Lqf is required in signal-sending cells, the MARCM technique was used to generate lqf^- clones that express either Dl or Ser under Gal4 control. Notch is normally expressed in both the D and V compartments of the wing primordium, but is modified in D cells by the action of the glycosyltransferase Fringe (Fng) so that it responds preferentially to Dl signaling from V cells. Ser is expressed predominantly in D compartment cells, and signals in the opposite direction, activating unmodified Notch in V cells. Clones of cells that express Dl under Gal4 control activate Notch strongly in adjacent wild-type cells only when located in the D compartment, as monitored by the expression of margin-specific genes like cut. Conversely, Ser-expressing clones activate Notch strongly only when located in the V compartment. In both cases, the levels of exogenous Dl and Ser expression are several fold higher than the peak levels of endogenous Dl and Ser generated along the DV boundary, and this overexpression autonomously inhibits the activation of Notch in cells within the clones (Wang, 2004).

Clones of lqf^- cells that overexpress either Dl or Ser fail to induce margin gene expression, irrespective of where they are located within the wing primordium. Indeed they behave like simple lqf^- clones in blocking normal margin gene expression when they abut, or cross, the DV compartment boundary. Thus, Lqf is required in DSL signal-sending cells to activate Notch in adjacent, signal-receiving cells (Wang, 2004).

Intact Dl and Ser normally accumulate in intracellular puncta, some of which co-localize with the endosomal marker Hepatocyte growth factor-regulated tyrosine kinase substrate (Hrs), as well as at the apical cell surface. By contrast, C-terminally truncated forms of Dl that lack the intracellular domain (Dl^DeltaC) accumulate predominantly at the cell surface and, like C-terminally truncated forms of Ser (Ser^DeltaC), cannot activate Notch on the surface of neighboring cells. If such truncated DSL ligands fail to signal because they cannot be endocytosed, replacement of the missing Dl cytosolic domain with heterologous domains that contain other internalization signals should rescue both endocytosis and signaling activity. Moreover, mutations in the internalization signals of these domains should eliminate their rescuing activity. These predictions have been tested and confirmed with two heterologous domains, each containing a different internalization signal (Wang, 2004).

(1) The missing intracellular domain of Dl^DeltaC was replaced with a 21 amino acid peptide from the Low Density Lipoprotein (LDL) receptor that contains either the wild-type internalization signal FDNPVY, or a mutant signal, ADAAVA. The LDL peptide contains two Lysines; these were replaced by Arginine to avoid their serving as possible acceptors for ubiquitination. Both the wild-type (Dl^LDL+) and mutant (Dl^LDLm) chimeric proteins were labeled by the insertion of six copies of the Myc epitope tag in the juxtamembrane portion of the extracellular domain. When expressed in the wing disc, Dl^LDL+ shows a similar subcellular distribution to wild-type Dl, accumulating on both the apical cell surface and in intracellular puncta. Dl^LDL+-expressing clones, like wild-type Dl-expressing clones, can induce cut activity in surrounding cells, indicating that the chimeric protein has signaling activity. However, they differ from wild-type Dl-expressing clones in that they only induce cut when located close to the DV boundary. Hence, it is inferred that Dl^LDL+-expressing clones have reduced signaling activity relative to wild-type Dl-expressing clones, and require the additional boost provided by endogenous signaling from neighboring wild-type cells to activate cut (Wang, 2004).

By contrast, Dl^LDLm accumulates predominantly on the apical cell surface, but not in intracellular puncta, and lacks signaling activity. Indeed, clones of cells overexpressing Dl^LDLm that abut the DV boundary block normal Notch signaling across the boundary, as would be expected if Dl^LDLm can inhibit Notch transduction within the same cell (like wild-type Dl), but is devoid of the capacity to activate Notch in adjacent cells (Wang, 2004).

(2) It was found serendipitously that replacement of the missing cytosolic domain of Dl^DeltaC with a random peptide, R+, of 50 amino acids (Dl^R+) also restored normal behavior. Dl^R+ accumulates in intracellular puncta as well as on the apical cell surface; in addition it activates Notch in neighboring cells. The R+ peptide contains two Lysines that might potentially serve as acceptors for ubiquitination. Replacement of both Lysines with Arginine blocked the rescuing activity of the R+ peptide. The mutant protein, Dl^Rm, accumulated predominantly only on the cell surface and lacked signaling activity; moreover, clones of Dl^Rm that abutted the DV boundary interrupted signaling across the boundary (Wang, 2004).

To assess the possibility that mono-ubiquitination of native Dl, as well as the Dl^R+ chimera, might suffice to provide an internalization signal, the missing cytosolic domain of Dl^DeltaC was replaced with Ubiquitin itself, and with a corresponding mutant form of Ubiquitin in which the Isoleucine at position 44 was mutated to Alanine, which functionally inactivates the internalization signal. All seven Lysine residues in each Ubiquitin domain were also replaced by Arginine to avoid additional ubiquitination. Both the resulting proteins, Dl^Ubi+ and Dl^Ubim, accumulated on the cell surface, as well as in intracellular puncta. However, far fewer puncta were found in Dl^Ubim-expressing cells than in Dl^Ubi+-expressing cells, and only Dl^Ubi+ was able to signal to neighboring cells. These data indicate that mono-ubiquitination is sufficient for Dl endocytosis and signalling, and suggest that at least one of the Lysines in the R+ peptide serves as Ubiquitin acceptor, allowing the protein to be internalized and to signal. It is noted that the Dl^Ubi+ protein appears to have only weak signaling activity relative to Dl or Dl^R+, since induction could only be detected of vg boundary-specific expression, but not cut or wg expression (Wang, 2004).

It is concluded that: (1) the cytosolic domain of Dl is essential for its endocytosis; (2) mono-ubiquitination is sufficient for Dl internalization; and (3) Dl endocytosis is essential for signaling activity (Wang, 2004).

Given that Epsin has been implicated in endocytosis, lqf^- cells may fail to send DSL signals because they are generally impaired for endocytosis. However, Dpp, Hh and Wg signaling, Wg internalization, and cell growth and proliferation are not adversely affected by the absence of Lqf, suggesting that endocytosis is not significantly impaired overall (Wang, 2004).

Alternatively, Lqf might be required specifically for the endocytosis of DSL ligands. To assess this possibility, the effects of lqf^- clones were first examined in the developing retina, where they have been reported to cause abnormally high levels of Dl on the cell surface, consistent with impaired Dl endocytosis (Overstreet, 2003). Such clones do indeed cause elevated surface expression of Dl, but it was also observed that endogenous Dl transcription (as assayed using a Dl-lacZ reporter gene), is strongly upregulated in the mutant cells, apparently as a consequence of the lack of Notch signaling. Furthermore, Dl staining can be detected in intracellular puncta in such lqf^- eye disc clones. Thus, the elevated surface accumulation of Dl observed in lqf^- eye clones can be ascribed to elevated Dl expression in the mutant cells, and may not reflect impaired Dl endocytosis (Wang, 2004).

lqf^- clones were generated in wing discs expressing uniformly high levels of exogenous Dl under Gal4 control, and Dl staining was compared in lqf^- cells and their wild-type neighbors. Under this condition, the level of Dl expression does not vary between wild-type and lqf^- cells, simplifying analysis. No difference was detected in the subcellular distribution of Dl between lqf^- and adjacent wild-type cells. In both cases, Dl was localized predominantly at the cell surface, as well as in similar numbers of intracellular puncta, many of which co-localize with the endosomal protein Hrs. The same result was obtained in separate experiments in which only the subcellular distribution of endogenous Dl was assayed (i.e., in the absence of overexpressed Dl) (Wang, 2004).

It was reasoned that if the Dl-positive puncta in lqf^- clones are indeed endocytic, the appearance of such puncta should change in the absence of hrs activity, which interferes with the maturation of early into late endosomes, and causes the formation of abnormal endosomal structures. To test this, both hrs^- and hrs^- lqf^- clones were generated. Endogenous Dl was found to accumulate in abnormally large puncta in both types of clones, and similar results were obtained when these clones expressed exogenous Dl under Gal4 control. The block in endosomal maturation caused by the removal of Hrs does not interfere with signaling by Dl; nor does it alter the requirement for Lqf. Clones of hrs^- cells that express exogenous Dl induce Cut expression in surrounding cells, whereas corresponding hrs^- lqf^- clones do not (Wang, 2004).

To determine unequivocally whether the abnormal puncta that accumulate Dl in hrs^- and hrs^- lqf^- cells are indeed endosomal, use was made of the finding that Wg secreted from prospective wing margin cells accumulates in similar, abnormally large puncta in hrs^- cells positioned at a distance from the secreting cells. The same result was obtained in double mutant hrs^- wg^- cells, establishing that the accumulation of Wg in these puncta serves as an in vivo marker for endocytosis. Then Wg and Dl staining was examined in triple mutant hrs^- wg^- lqf^- clones that express an HRP-tagged form of Dl under Gal4 control. In this case, as in corresponding hrs^- wg^- double mutant clones, co-localization of Wg and Dl was observed in large intracellular puncta. Thus, bulk endocytosis of both endogenous and overexpressed Dl appear normal in lqf^- cells (Wang, 2004).

Although bulk Dl endocytosis appears unaffected by the absence of Lqf, blockage of a relatively small, but specific, subset of Dl endocytic events might escape detection, and this subset might be crucial for signaling activity. To examine this possibility, Dl was co-expressed together with the E3 Ubiquitin Ligase Neuralized (Neur), under Gal4 control, to drive efficient ubiquitination and internalization of the exogenous Dl. It was reasoned that under these conditions, even modest reductions in the rate of Dl endocytosis might cause an abnormal persistence of Dl at the apical cell surface (Wang, 2004).

Wing discs that express uniformly high levels of Dl under Gal4 control accumulate high levels of Dl on the apical cell surface. However, in discs that co-express high levels of both Dl and Neur, this surface accumulation is strongly reduced and Dl accumulates instead in an abnormally large number of intracellular puncta. Clones of lqf^- cells generated in such co-expressing discs do not appear to alter the number or general appearance of these Dl-positive puncta, many of which co-localize with Hrs. However, they do affect the level of Dl staining associated with the apical cell surface (as visualized in discs processed either with, or without, detergent). Such lqf^- clones show residual surface staining of Dl, in contrast to neighboring wild-type cells where surface-associated staining is depleted. It is inferred that lqf^- cells cannot endocytose Dl as efficiently as their wild-type neighbors, accounting for why a difference was detected under sensitized conditions in which the rate of surface clearance appears to be limiting (Wang, 2004).

Significantly, the residual staining of Dl on the surface of lqf^- cells that overexpress Neur and Dl correlates with the failure of these cells to signal. Clones of lqf^- cells that overexpress Neur and Dl fail to activate cut in neighboring cells, even though clones of otherwise wild-type cells that overexpress Neur and Dl show enhanced Dl signaling. Hence, it appears that the impairment in Dl endoctyosis detected in lqf^- clones in this sensitized background correlates with an absolute block in signaling activity (Wang, 2004).

The cytosolic domain of DSL ligands contains multiple Lysines at least some of which serve as acceptors for Ubiquitin. Lqf contains two Ubiquitin Interacting Motifs (UIMs) (Hofmann, 2001). Hence, mono-ubiquitination of DSL ligands might allow Lqf to target DSL ligands for a special subset of endocytic events that are required for signaling activity. By contrast, bulk endocytosis of DSL ligands mediated by interactions with other Ubiquitin-binding adaptor proteins might not suffice to confer signaling activity. To test this hypothesis, whether the signaling activity of the Dl^R+ protein depends on Lqf activity, was investigated (Wang, 2004).

Endocytosis and signaling activity of Dl^R+ depends on the presence of at least one of the two Lysines in the R+ peptide comprising the cytosolic domain. Clones of lqf^- cells that express Dl^R+ fail to induce cut expression in adjacent wing disc cells. However, Dl^R+ protein in these lqf^- clones accumulates both on the apical surface and in intracellular puncta. Moreover, no difference was detected in the punctate, cytosolic accumulation of Dl^R+ between lqf^- and wild-type cells in wing discs that generally overexpress Dl^R+. Both results indicate that bulk endocytosis of Dl^R+ is not significantly altered in the absence of Lqf. Because substitution of both Lysines by Arginine blocks internalization and signaling activity of Dl^Rm, it is inferred that Dl^R+ is targeted for internalization solely by ubiquitination at one or both of these Lysines. Hence, it is suggested that other Ubiquitin-interacting proteins aside from Lqf can target mono-ubiquitinated cargo proteins, such as Dl^R+ or endogenous Dl, for internalization. However, only Lqf appears able to direct endocytosis of these proteins in a way that allows DSL ligands to signal (Wang, 2004).

Both endocytosis and signaling activity of Dl^LDL+ depends on the FDNPVY internalization signal. However, unlike either native Dl or Dl^R+, it was found that clones of lqf^- cells expressing Dl^LDL+ can induce cut expression in adjacent wild-type cells, indicating that the presence of the LDL internalization signal in the chimeric Dl^LDL+ protein bypasses the requirement for Lqf. As observed for clones of wild-type cells overexpressing Dl^LDL+, the 'rescued' lqf^- clones induced cut only when located close to the DV boundary. Nevertheless, their ability to signal, albeit weakly, contrasts with that of lqf^- clones that overexpress native Dl, native Dl plus Neur, or Dl^R+, all of which are devoid of signaling activity. Hence, it is concluded that the FDNPVY signal directs internalization of Dl^LDL+ in a manner that permits the protein to acquire signaling activity even in the absence of Lqf activity (Wang, 2004).

Lqf-dependent endocytosis of DSL ligands might be accompanied by modifications of these ligands, either as a pre-requisite for, or a consequence of, signaling activity. To examine this possibility, it was asked whether the size of Dl protein changes as a consequence of Lqf-dependent endocytosis. Initially, clones of wild-type and lqf^- cells were generated that express Dl tagged by the insertion of six copies of the Myc epitope in the extracellular juxtamembrane domain, and the profile of Dl peptides that retain the Myc epitope was examined by Western blotting. Under these conditions, similar, complex profiles were observed of Myc-tagged Dl peptides from both wild-type and lqf^- cells, corresponding to full-length Myc-Dl protein, as well as several lower molecular weight peptides (Wang, 2004).

This experiment was then repeated using wild-type and lqf^- cells that overexpress Neur and Myc-tagged Dl, the sensitized condition under which residual surface expression can be detected of Myc-tagged Dl in lqf^-, but not in wild-type, cells. In this case, the profile of Myc-tagged Dl is remarkably simple. Wild-type cells show two bands, one corresponding by size to full-length Myc-tagged Dl (~105 kDa) and the other to a Myc-tagged cleavage product of ~50 kDa. By contrast, lqf^- cells show only a single band, corresponding to full-length Myc-tagged Dl. Thus, the failure to clear Dl from the cell surface of lqf^- cells is associated with an apparent failure in Dl processing. These results provide evidence for a Lqf-dependent cleavage of Dl that correlates with Lqf-dependent endocytosis and signaling activity (Wang, 2004).

It is noted that the expected size of the Myc-tagged extracellular domain of Dl is ~75 kDa, whereas that of the complementary, Myc-tagged portion of the ligand containing the transmembrane and cytosolic domains is ~40 kDa. Hence, the 50 kDa Myc-tagged cleavage product must be composed of a C-terminal portion of the extracellular domain, and possibly some or all of the transmembrane and cytosolic domains as well. The relationship of this truncated peptide to the active ligand is presently unknown. It could comprise part, or all, of the active ligand, or alternatively, a non-signaling C-terminal fragment cleaved off in the process of generating an N-terminal signaling fragment. Alternatively, it might be a degradation product generated as a consequence of the activation of Notch by Dl (Wang, 2004).

EVOLUTIONARY HOMOLOGS

Genetic interaction studies show that Liquid facets is a candidate for the critical substrate of Fat facets in the eye

Fat facets is a deubiquitinating enzyme required in a cell communication pathway that limits to eight the number of photoreceptor cells in each facet of the Drososphila compound eye. Genetic data support a model whereby Faf removes ubiquitin, a polypeptide tag for protein degradation, from a specific ubiquitinated protein, thus preventing its degradation. Mutations in the liquid facets gene have been identified as dominant enhancers of the fat facets mutant eye phenotype. The liquid facets locus encodes epsin, a vertebrate protein associated with the clathrin endocytosis complex. Genetic experiments reveal that fat facets and liquid facets facilitate endocytosis and function in common cells to generate an inhibitory signal that prevents ectopic photoreceptor determination. The fat facets mutant phenotype is extraordinarily sensitive to the level of liquid facets expression. It is proposed that Liquid facets is a candidate for the critical substrate of Fat facets in the eye (Cadavid, 2000).

There are three key components of the endocytosis complex: (1) clathrin, which forms a cage structure engulfing the cell membrane; (2) AP-2, the core adaptor complex, which binds to clathrin and brings it to the cell surface, and (3) dynamin, a GTPase required for vesicle formation. Additional proteins associated with AP-2 have been identified, many of which contain protein-protein interaction domains called EH-domains and EH-domain-binding motifs. Epsin is an EH-domain-binding protein identified as a partner for Eps15, an EH-domain protein that also binds AP-2. The large number of AP-2-binding proteins identified suggests that many of them may have temporal and/or tissue-specific functions (Marsh and McMahon, 1999). The precise roles of Eps15 and epsin in endocytosis are unknown (Cadavid, 2000 and references therein).

The lqf gene itself is essential in Drosophila; in an otherwise wild-type background, lqf null mutants die as embryos. Clones of cells in the eye in which there is little or no lqf gene function have severely disrupted eye morphology, indicating that lqf is required also after embryogenesis for eye development. The mutant phenotypes associated with two weak lqf mutant alleles reveal specific roles for lqf in eye, wing and leg development (Cadavid, 2000).

The eye defects in homozygous adults for a weak allele of lqf resemble those in faf null mutants. As in faf mutants, the additional photoreceptors in lqf mutants arise from specific precursor cells (M-cells) present early during eye development. In contrast to lqf null mutants, faf null mutants are viable, have normal wings and legs and have less severe eye defects. Thus, lqf functions more broadly than faf, but both the lqf and faf genes are required during eye development in order to prevent the M-cells from becoming photoreceptors (Cadavid, 2000).

The faf mutant eye phenotype is unusually sensitive to a decrease in the dose of the lqf gene, suggesting strongly that the two genes function in a common pathway. Genetic interactions with endocytosis and Ub pathway mutants show that faf and lqf facilitate endocytosis and antagonize ubiquitination. In addition, although lqf is more broadly required than faf in the eye and elsewhere in the fly, weak lqf mutations reveal that like faf, lqf is required to prevent the misdetermination of M-cells as photoreceptors. Moreover, when expressed only in the rough positive cells surrounding the facet preclusters, both faf and lqf genes rescue completely to wild-type their respective M-cell misdetermination mutant phenotypes in the eye. Finally, given the relationship between Faf and its substrate protein, it would be expected that increasing the dose of the substrate should suppress the faf mutant phenotype. Slight overexpression of lqf completely obviates the need for faf in eye development. The simplest model consistent with all of this genetic data is that Lqf is the substrate of Faf. Other more complicated explanations are, of course, possible (Cadavid, 2000).

There is biochemical evidence that AF-6, a scaffolding protein thought to modulate cell-cell junctions in response to Ras activation may be an in vivo substrate of Fam, the mouse homolog of Faf; AF-6 and Fam bind each other in vitro and ubiquitinated AF-6 can be detected and deubiquitinated by Fam in cultured cells. Like lqf, the Drosophila Af6 homolog, canoe, is required pleiotropically for Drosophila eye development. In contrast to lqf mutations, however, canoe mutations do not act as strong dominant enhancers of the faf mutant eye phenotype. Given the striking genetic interactions between faf and lqf, it seems that canoe is unlikely to play a significant role in the essential faf pathway in the eye (Cadavid, 2000 and references therein).

While only one Faf/substrate interaction may be essential to normal eye development in Drosophila, Faf and Fam may have several substrates in vivo. Normally non-essential roles for faf later in eye development have been revealed in particular mutant backgrounds and Faf could have different substrates for its critical role in M-cell fate determination than in its redundant roles. Moreover, in addition to its essential role in eye development, faf is required maternally for cellularization of embryos and the critical maternal substrate of Faf is unknown. Because faf has mouse and human homologs, the modes of regulation by Faf are likely to be conserved. However, it is possible that the critical substrate(s) of Faf in Drosophila may differ from those in vertebrates (Cadavid, 2000 and references therein).

If Lqf is the substrate of Faf, then epsin levels, determined by the balance between its ubiquitination and deubiquitination, could regulate endocytosis. Mono-ubiquitination, however, has been shown previously to regulate endocytosis in two different ways. (1) Mono-ubiquitination of cell surface receptors can act as a signal for receptor endocytosis, which leads to lysosomal degradation. Here, the Ub moiety is somehow recognized by the endocytosis machinery; this process has nothing to do with the proteasome. (2) Eps15, an endocytosis complex component in mammalian cells, is mono-ubiquitinated in response to EGF receptor activation and Eps15 may require this modification to stimulate receptor endocytosis. In addition, Pan1p, a yeast protein similar to Eps15, is required for endocytosis in yeast. Although it is unknown whether Pan1p is mono-ubiquitinated in yeast, there is evidence that ubiquitination of an endocytic complex component is required for endocytosis in yeast; Rsp5p, a component of the ubiquitination machinery called a ubiquitin-ligase, may bind to Pan1p and is required generally for endocytosis in yeast, even for endocytosis of proteins with non-Ub endocytosis signals (Cadavid, 2000 and references therein).

Since Eps15 binds to epsin, could a mono-ubiquitinated Drosophila Eps15 homolog be the substrate of Faf? Two experimental results are inconsistent with this model. (1) It has been shown previously that the activity of Faf antagonizes proteolysis, not just ubiquitination; mutations in a gene encoding a proteasome subunit act as strong suppressors of the faf mutant eye phenotype. This result strongly suggests that Faf activity antagonizes proteolysis and thus that Faf deubiquitinates a protein containing a Ub chain targeting it for degradation, rather than a mono-ubiquitinated protein. (2) If mono-ubiquitination of Eps15 activates it, as the available data suggests, then deubiquitination of Eps15 by Faf would render Eps15 inactive and thus the function of Faf would antagonize endocytosis. The data presented here clearly indicate the opposite; mutations in endocytosis complex genes (particularly lqf and Clathrin heavy chain) act as strong dominant enhancers of faf, suggesting that the normal function of Faf is to facilitate endocytosis (Cadavid, 2000 and references therein).

Elevated levels of Lqf obviate the need for Faf, presumably by stimulating epsin-dependent endocytosis generally or stimulating endocytosis of a specific cell surface protein. How can the observation that Lqf and Faf function outside the M-cells to determine M-cell fate be reconciled with a role for Lqf in endocytosis? Endocytosis is known to modulate ligand/receptor interactions by a variety of mechanisms. One possibility is that M-cell fate is affected by a diffusible ligand that, like Wingless, travels via endocytosis through several cell distances. Alternatively, regulation of a membrane-bound receptor by endocytosis in cells adjacent to the M-cells could affect M-cell fate indirectly. For example, EGF receptor activity is downregulated by endocytosis following ligand binding. By contrast, activity of the Notch receptor may be up-regulated by endocytosis of activated receptors whose intracellular domains have been cleaved off prior to their translocation into the nucleus. Membrane-bound Notch receptors lacking their intracellular domains display dominant negative activity and endocytosis of cleaved Notch receptors may be required normally for precise modulation of Notch activity. Patterning of the photoreceptor preclusters in the developing eye may require that both Notch and the EGF receptor are activated in the rough-expressing cells surrounding the facet preclusters. Thus, Faf could regulate the activity of one or both of these receptors (Cadavid, 2000 and references therein).

Liquid facet functions in endocytosis of Delta in the developing eye

Epsin is part of a protein complex that performs endocytosis in eukaryotes. Drosophila epsin, Liquid facets (Lqf), was identified because it is essential for patterning the eye and other imaginal disc derivatives. Previous work has provided only indirect evidence that Lqf is required for endocytosis in Drosophila. Epsins are modular and have an N-terminal ENTH (epsin N-terminal homology) domain that binds PIP2 at the cell membrane and four different classes of protein-protein interaction motifs. The current model for epsin function in higher eukaryotes is that epsin bridges the cell membrane, a transmembrane protein to be internalized, and the core endocytic complex. This study shows directly that Drosophila epsin (Lqf) is required for endocytosis. Specifically, Lqf is essential for internalization of the Delta (Dl) transmembrane ligand in the developing eye. Using this endocytic defect in lqf mutants, a transgene rescue assay has been developed and a structure/function analysis of Lqf has been performed. When Lqf is divided into two pieces, an ENTH domain and an ENTH-less protein, each part retains significant ability to function in Dl internalization and eye patterning. These results challenge the model for epsin function that requires an intact protein (Overstreet, 2003).

To test for endocytosis defects in lqf^- mutants, the localization of the transmembrane receptor Dl was monitored in developing eyes. Dl normally undergoes endocytosis in the eye, and as the internalized protein is not degraded rapidly, internalized Dl can be detected in vesicles (Overstreet, 2003).

The Drosophila eye, composed of 800 identical 22-cell ommatidia, or facets, develops in the larval and pupal stages in a monolayer epithelium called the eye imaginal disc. Eye development occurs as a wave, where the morphogenetic furrow forms at the posterior of the disc, and moves anteriorly into the monolayer of undifferentiated cells. Rows of ommatidia assemble stepwise posterior to the furrow one or two cells at a time, starting with the eight photoreceptors (R1-R8) (Overstreet, 2003).

In wild-type, Dl is detected exclusively as intracellular dots within developing ommatidial clusters throughout the eye disc. In larval eye discs homozygous for lqf^FDD9, a weak, viable mutant allele, Dl is detected mainly at the membranes of cells just posterior to the furrow. In clones of cells homozygous for lqf^ARI, a strong, lethal mutant allele, similar membrane localization of Dl is observed. It is concluded that lqf⁺ is required for Dl internalization (Overstreet, 2003).

All epsins have an amino-terminal ENTH domain that binds PIP₂ at the cell membrane and three or four types of protein-protein interaction motifs, whose copy numbers vary among different epsins. The ubiquitin interaction motifs (UIMs) bind ubiquitin (Ub) noncovalently. There are also clathrin binding motifs (CBMs), DPW motifs that bind the core endocytic adaptor complex, AP-2, and NPF motifs that bind Eps15, another accessory factor (Overstreet, 2003).

A step toward understanding the role of Lqf in endocytosis is the identification of the modules of Lqf protein that are required. In yeast, there are straightforward assays for the function of the two epsins (Ent1 and Ent2). Structure/function analyses have demonstrated that the ENTH domain of Ent1 is necessary and sufficient to rescue the lethality of ent1Δent2Δ double mutants. Moreover, the ENTH domain and to a lesser extent the UIMs have been shown to be required for endocytosis. Because there are mechanistic differences between endocytosis in yeast and higher eukaryotes, the yeast epsins might function somewhat differently from vertebrate epsins and Drosophila Lqf. The major difference between these systems is that the AP-2 core adaptor complex in yeast has no known function in endocytosis, and, accordingly, the yeast epsins lack DPW motifs. As in yeast, structure/function analyses of epsins in vertebrate cell culture have pointed to the importance of the ENTH domain. These assays, however, rely on dominant-negative effects of mutant epsin proteins on endocytosis, and their interpretation is difficult (Overstreet, 2003).

Either the ENTH domain alone, or an ENTH-less Lqf protein, rescues the patterning and Dl endocytosis defects in lqf^FDD9 homozygous eyes. Since experimental results in yeast and in vertebrates have emphasized the importance of the ENTH domain, the most remarkable result is that an ENTH-less Lqf protein can function. The simplest interpretation of the rescue results is that Lqf^ΔENTH can function independently of the ENTH domain (Overstreet, 2003).

Transgenes that express Rat epsin1 or human epsin 2b in Drosophila with pRO, each as full-length proteins or without the ENTH domain, rescue the eye defects in lqf^FDD9 homozygotes. Thus, there is unlikely to be a significant functional difference between the Drosophila and vertebrate epsins in the region C-terminal to the ENTH domain. In addition, the ENTH domains of Lqf and yeast epsin are functionally similar. It was shown previously that expression of the ENTH domain of Ent1, but not the complementary portion of the protein, restores viability to ent1Δent2Δ yeast. Similarly, expression of full-length Lqf or Lqf^ENTH rescues ent1Δent2Δ lethality but Lqf^ΔENTH expression does not (Overstreet, 2003).

Thus Drosophila epsin, Lqf, is essential for endocytosis of Dl in the developing eye. Moreover, the ENTH domain alone and an ENTH-less Lqf protein each retain significant function. The prevailing model in vertebrates is that epsin functions like a bridge, where the ENTH domain links the membrane to clathrin, a cell surface protein to be internalized, and to AP-2. Since this model requires an intact epsin protein, the results presented here suggest that the prevailing model cannot be the whole story. Moreover, the observation that either the ENTH domain or the remainder of the protein, which are functionally distinct, can be deleted without destroying Lqf function completely suggests that each fragment of Lqf may be partially redundant with another Drosophila endocytic protein. Candidates for the other endycotgic protein include the other ENTH domain protein in Drosophila, Epsin-2 and the Drosophila homolog of AP180, which, like the ENTH-less Lqf protein, binds clathrin and AP-2 (Overstreet, 2003).

Fat facets and Liquid facets promote Delta endocytosis and Delta signaling in the signaling cells

Endocytosis modulates the Notch signaling pathway in both the signaling and receiving cells. One recent hypothesis is that endocytosis of the ligand Delta by the signaling cells is essential for Notch activation in the receiving cells. Evidence is presented in strong support of this model. In the developing Drosophila eye Fat facets (Faf), a deubiquitinating enzyme, and its substrate Liquid facets (Lqf), an endocytic epsin, promote Delta internalization and Delta signaling in the signaling cells. While Lqf is necessary for three different Notch/Delta signaling events at the morphogenetic furrow, Faf is essential only for one: Delta signaling by photoreceptor precluster cells, which prevents recruitment of ectopic neurons. In addition, the ubiquitin-ligase Neuralized (Neur), which ubiquitinates Delta, is shown to function in the signaling cells with Faf and Lqf. The results presented bolster one model for Neur function in which Neur enhances Delta signaling by stimulating Delta internalization in the signaling cells. It is proposed that Faf plays a role similar to that of Neur in the Delta signaling cells. By deubiquitinating Lqf, which enhances the efficiency of Delta internalization, Faf stimulates Delta signaling (Overstreet, 2004).

Cells with decreased lqf⁺ activity accumulate Delta on apical membranes and fail to signal to neighboring cells. Three Notch/Delta signaling events were examined in the eye: proneural enhancement, lateral inhibition and R-cell restriction. Loss of lqf⁺-dependent endocytosis during all three events has identical consequences to loss of Delta function in the signaling cells. It is concluded that lqf⁺-dependent endocytosis of Delta is required for signaling, supporting the notion that endocytosis in the signaling cells activates Notch in the receiving cells. However, Lqf is not required absolutely for all Delta internalization in the eye. Even in lqf-null cells, which are incapable of Delta signaling, some vesicular Delta is present. Perhaps not all of the vesicular Delta present in wild-type discs results from signaling (Overstreet, 2004).

Genetic studies in Drosophila indicate clearly that deubiquitination of Lqf by Faf activates Lqf activity. Moreover, genetic and biochemical evidence in Drosophila suggests that Faf prevents proteasomal degradation of Lqf. In vertebrates, however, it is thought that epsin is mono-ubiquitinated to modulate its activity rather than poly-ubiquitinated to target it for degradation. If Lqf regulation by ubiquitin also occurs this way in the Drosophila eye, the removal of mono-ubiquitin from Lqf by Faf would activate Lqf activity (Overstreet, 2004).

Whatever the precise mechanism, given that both Faf and Lqf are expressed ubiquitously in the eye, two related questions arise. First, why is Lqf ubiquitinated at all if Faf simply deubiquitinates it everywhere? One possibility is that Faf is one of many deubiquitinating enzymes that regulate Lqf, and expression of the others is restricted spatially. This could also explain why Faf is required only for R-cell restriction. Another possibility is that Faf activity is itself regulated in a spatial-specific manner in the eye disc. Alternatively, Lqf ubiquitination may be so efficient that Faf is needed to provide a pool of non-ubiquitinated, active Lqf. Similarly, Faf could be part of a subtle mechanism for timing Lqf activation. Second, why is Faf essential only for R-cell restriction? One possibility is that there is a graded requirement for Lqf in the eye disc, such that proneural enhancement requires the least Lqf, lateral inhibition somewhat more and neural inhibitory signaling by R2/3/4/5 the most. The mutant phenotype of homozygotes for the weak allele lqf^FDD9 supports this idea, as R-cell restriction is most severely affected. Alternatively, Lqf may be expressed or ubiquitinated with dissimilar efficiencies in different regions of the eye disc. More experiments are needed to understand the precise mechanism by which the Faf/Lqf interaction enhances Delta signaling (Overstreet, 2004).

In neur mutants, Delta accumulates on the membranes of signaling cells and Notch activation in neighboring cells is reduced. These results support a role for Neur in endocytosis of Delta in the signaling cells to achieve Notch activation in the neighboring receiving cells, rather than in downregulation of Delta in the receiving cells. Because neur shows strong genetic interactions with lqf and both function in R-cells, Neur and Lqf might work together to stimulate Delta endocytosis. Lqf has ubiquitin interaction motifs (UIMs) that bind ubiquitin. One explanation for how Neur and Faf/Lqf could function together is that Lqf facilitates Delta endocytosis by binding to Delta after its ubiquitination by Neur. This is anattractive model that will stimulate further experiments (Overstreet, 2004).

One exciting observation is that the endocytic adapter Lqf may be essential specifically for Delta internalization. Although, hedgehog, decapentaplegic and wingless signaling pathways have not been examined directly, they appear to be functioning in the absence of Lqf. These three signaling pathways regulate movement of the morphogenetic furrow and are thought to require endocytosis. The furrow moves through lqf-null clones and at the same pace as the surrounding wild-type cells. Moreover, all mutant phenotypes of lqf-null clones can be accounted for by loss of Delta function. Further experiments will clarify whether this apparent specificity means that Lqf functions only in internalization of Delta, or if the process of Delta endocytosis is particularly sensitive to the levels of Lqf (Overstreet, 2004).

Lqf expands the small repertoire of endocytic proteins that are known targets for regulation of cell signaling. In addition to Lqf, the endocytic proteins Numb and Eps15 (EGFR phosphorylated substrate 15) are objects of regulation. In vertebrates, asymmetrical distribution into daughter cells of the alpha-adaptin binding protein Numb may be achieved through ubiquitination of Numb by the ubiquitin-ligase LNX (Ligand of Numb-protein X) and subsequent Numb degradation. In addition, in vertebrate cells, Eps15 is phosphorylated and recruited to the membrane in response to EGFR activation and is required for ligand-induced EGFR internalization. Given that endocytosis is so widely used in cell signaling, endocytic proteins are likely to provide an abundance of targets for its regulation (Overstreet, 2004).

Drosophila Epsin mediates a select endocytic pathway that DSL ligands must enter to activate Notch

In screens for mutations affecting wing pattern, six alleles of a single complementation group were obtained that cause phenotypes similar to those caused by the loss of Notch signaling, namely severe wing notching, wing vein thickening and bristle tufts. All six alleles fail to complement existing alleles of liquid facets (lqf), and are associated with nonsense or missense mutations in the lqf-coding sequence (Overstreet, 2003). One new allele, lqf¹²²⁷, truncates the coding sequence after amino acid 119 in the middle of the ENTH domain, the most N-terminal conserved domain, and abolishes Lqf protein expression in vivo. This allele is referred to as lqf^-, and it was used for all experiments described in this study. A transgene containing the intact lqf gene (Cadavid, 2000) rescues the lethality of lqf^- homozygotes, as well as all of the mutant phenotypes associated with lqf^- clones. Lqf encodes the sole ortholog of vertebrate Epsin1 (Cadavid, 2000); a second Drosophila protein, sometimes referred to as Dm Epsin2 (Overstreet, 2003), lacks several conserved domains found in Lqf and vertebrate Epsin1, and appears instead to be the Drosophila ortholog of vertebrate EpsinR (see Mills, 2003), a functionally distinct Epsin-related protein (Wang, 2004).

In imaginal wing discs, signaling by the DSL ligands Delta (Dl) and Serrate (Ser) specifies the wing margin at the dorsoventral (DV) compartment boundary, and can be assayed by boundary-specific expression of wing margin genes (or their protein products), such as cut, wingless (wg) and vestigial (vg). lqf^- clones resemble Dl^- Ser^- clones or N^- clones in that they cause the loss of cut, wg and vg boundary-specific expression when they abut or cross the DV compartment boundary, corroborating the Notch-related phenotypes of lqf^- clones observed in the adult wing (Wang, 2004).

The loss of margin gene expression in lqf^- clones is not cell autonomous. Instead, wild-type cells can rescue the expression of margin specific genes in adjacent lqf^- cells (e.g., cut). Similarly, non-autonomous rescue of lqf^- clones was observed in the adult, where the presence of wild-type cells can rescue the ability of neighboring lqf^- cells to form single bristles. In both respects, as well as in others, lqf^- clones resemble Dl^- Ser^- clones, but differ from N^- clones, which show a strictly cell-autonomous loss of Notch target gene expression (Wang, 2004).

Collectively, these data establish an obligate role for Lqf in Notch signaling, and implicate Lqf in sending, rather than receiving, DSL signals (Wang, 2004).

To determine whether Lqf is required in signal-sending cells, the MARCM technique was used to generate lqf^- clones that express either Dl or Ser under Gal4 control. Notch is normally expressed in both the D and V compartments of the wing primordium, but is modified in D cells by the action of the glycosyltransferase Fringe (Fng) so that it responds preferentially to Dl signaling from V cells. Ser is expressed predominantly in D compartment cells, and signals in the opposite direction, activating unmodified Notch in V cells. Clones of cells that express Dl under Gal4 control activate Notch strongly in adjacent wild-type cells only when located in the D compartment, as monitored by the expression of margin-specific genes like cut. Conversely, Ser-expressing clones activate Notch strongly only when located in the V compartment. In both cases, the levels of exogenous Dl and Ser expression are several fold higher than the peak levels of endogenous Dl and Ser generated along the DV boundary, and this overexpression autonomously inhibits the activation of Notch in cells within the clones (Wang, 2004).

Clones of lqf^- cells that overexpress either Dl or Ser fail to induce margin gene expression, irrespective of where they are located within the wing primordium. Indeed they behave like simple lqf^- clones in blocking normal margin gene expression when they abut, or cross, the DV compartment boundary. Thus, Lqf is required in DSL signal-sending cells to activate Notch in adjacent, signal-receiving cells (Wang, 2004).

Intact Dl and Ser normally accumulate in intracellular puncta, some of which co-localize with the endosomal marker Hepatocyte growth factor-regulated tyrosine kinase substrate (Hrs), as well as at the apical cell surface. By contrast, C-terminally truncated forms of Dl that lack the intracellular domain (Dl^DeltaC) accumulate predominantly at the cell surface and, like C-terminally truncated forms of Ser (Ser^DeltaC), cannot activate Notch on the surface of neighboring cells. If such truncated DSL ligands fail to signal because they cannot be endocytosed, replacement of the missing Dl cytosolic domain with heterologous domains that contain other internalization signals should rescue both endocytosis and signaling activity. Moreover, mutations in the internalization signals of these domains should eliminate their rescuing activity. These predictions have been tested and confirmed with two heterologous domains, each containing a different internalization signal (Wang, 2004).

(1) The missing intracellular domain of Dl^DeltaC was replaced with a 21 amino acid peptide from the Low Density Lipoprotein (LDL) receptor that contains either the wild-type internalization signal FDNPVY, or a mutant signal, ADAAVA. The LDL peptide contains two Lysines; these were replaced by Arginine to avoid their serving as possible acceptors for ubiquitination. Both the wild-type (Dl^LDL+) and mutant (Dl^LDLm) chimeric proteins were labeled by the insertion of six copies of the Myc epitope tag in the juxtamembrane portion of the extracellular domain. When expressed in the wing disc, Dl^LDL+ shows a similar subcellular distribution to wild-type Dl, accumulating on both the apical cell surface and in intracellular puncta. Dl^LDL+-expressing clones, like wild-type Dl-expressing clones, can induce cut activity in surrounding cells, indicating that the chimeric protein has signaling activity. However, they differ from wild-type Dl-expressing clones in that they only induce cut when located close to the DV boundary. Hence, it is inferred that Dl^LDL+-expressing clones have reduced signaling activity relative to wild-type Dl-expressing clones, and require the additional boost provided by endogenous signaling from neighboring wild-type cells to activate cut (Wang, 2004).

By contrast, Dl^LDLm accumulates predominantly on the apical cell surface, but not in intracellular puncta, and lacks signaling activity. Indeed, clones of cells overexpressing Dl^LDLm that abut the DV boundary block normal Notch signaling across the boundary, as would be expected if Dl^LDLm can inhibit Notch transduction within the same cell (like wild-type Dl), but is devoid of the capacity to activate Notch in adjacent cells (Wang, 2004).

(2) It was found serendipitously that replacement of the missing cytosolic domain of Dl^DeltaC with a random peptide, R+, of 50 amino acids (Dl^R+) also restored normal behavior. Dl^R+ accumulates in intracellular puncta as well as on the apical cell surface; in addition it activates Notch in neighboring cells. The R+ peptide contains two Lysines that might potentially serve as acceptors for ubiquitination. Replacement of both Lysines with Arginine blocked the rescuing activity of the R+ peptide. The mutant protein, Dl^Rm, accumulated predominantly only on the cell surface and lacked signaling activity; moreover, clones of Dl^Rm that abutted the DV boundary interrupted signaling across the boundary (Wang, 2004).

To assess the possibility that mono-ubiquitination of native Dl, as well as the Dl^R+ chimera, might suffice to provide an internalization signal, the missing cytosolic domain of Dl^DeltaC was replaced with Ubiquitin itself, and with a corresponding mutant form of Ubiquitin in which the Isoleucine at position 44 was mutated to Alanine, which functionally inactivates the internalization signal. All seven Lysine residues in each Ubiquitin domain were also replaced by Arginine to avoid additional ubiquitination. Both the resulting proteins, Dl^Ubi+ and Dl^Ubim, accumulated on the cell surface, as well as in intracellular puncta. However, far fewer puncta were found in Dl^Ubim-expressing cells than in Dl^Ubi+-expressing cells, and only Dl^Ubi+ was able to signal to neighboring cells. These data indicate that mono-ubiquitination is sufficient for Dl endocytosis and signalling, and suggest that at least one of the Lysines in the R+ peptide serves as Ubiquitin acceptor, allowing the protein to be internalized and to signal. It is noted that the Dl^Ubi+ protein appears to have only weak signaling activity relative to Dl or Dl^R+, since induction could only be detected of vg boundary-specific expression, but not cut or wg expression (Wang, 2004).

It is concluded that: (1) the cytosolic domain of Dl is essential for its endocytosis; (2) mono-ubiquitination is sufficient for Dl internalization; and (3) Dl endocytosis is essential for signaling activity (Wang, 2004).

Given that Epsin has been implicated in endocytosis, lqf^- cells may fail to send DSL signals because they are generally impaired for endocytosis. However, Dpp, Hh and Wg signaling, Wg internalization, and cell growth and proliferation are not adversely affected by the absence of Lqf, suggesting that endocytosis is not significantly impaired overall (Wang, 2004).

Alternatively, Lqf might be required specifically for the endocytosis of DSL ligands. To assess this possibility, the effects of lqf^- clones were first examined in the developing retina, where they have been reported to cause abnormally high levels of Dl on the cell surface, consistent with impaired Dl endocytosis (Overstreet, 2003). Such clones do indeed cause elevated surface expression of Dl, but it was also observed that endogenous Dl transcription (as assayed using a Dl-lacZ reporter gene), is strongly upregulated in the mutant cells, apparently as a consequence of the lack of Notch signaling. Furthermore, Dl staining can be detected in intracellular puncta in such lqf^- eye disc clones. Thus, the elevated surface accumulation of Dl observed in lqf^- eye clones can be ascribed to elevated Dl expression in the mutant cells, and may not reflect impaired Dl endocytosis (Wang, 2004).

lqf^- clones were generated in wing discs expressing uniformly high levels of exogenous Dl under Gal4 control, and Dl staining was compared in lqf^- cells and their wild-type neighbors. Under this condition, the level of Dl expression does not vary between wild-type and lqf^- cells, simplifying analysis. No difference was detected in the subcellular distribution of Dl between lqf^- and adjacent wild-type cells. In both cases, Dl was localized predominantly at the cell surface, as well as in similar numbers of intracellular puncta, many of which co-localize with the endosomal protein Hrs. The same result was obtained in separate experiments in which only the subcellular distribution of endogenous Dl was assayed (i.e., in the absence of overexpressed Dl) (Wang, 2004).

It was reasoned that if the Dl-positive puncta in lqf^- clones are indeed endocytic, the appearance of such puncta should change in the absence of hrs activity, which interferes with the maturation of early into late endosomes, and causes the formation of abnormal endosomal structures. To test this, both hrs^- and hrs^- lqf^- clones were generated. Endogenous Dl was found to accumulate in abnormally large puncta in both types of clones, and similar results were obtained when these clones expressed exogenous Dl under Gal4 control. The block in endosomal maturation caused by the removal of Hrs does not interfere with signaling by Dl; nor does it alter the requirement for Lqf. Clones of hrs^- cells that express exogenous Dl induce Cut expression in surrounding cells, whereas corresponding hrs^- lqf^- clones do not (Wang, 2004).

To determine unequivocally whether the abnormal puncta that accumulate Dl in hrs^- and hrs^- lqf^- cells are indeed endosomal, use was made of the finding that Wg secreted from prospective wing margin cells accumulates in similar, abnormally large puncta in hrs^- cells positioned at a distance from the secreting cells. The same result was obtained in double mutant hrs^- wg^- cells, establishing that the accumulation of Wg in these puncta serves as an in vivo marker for endocytosis. Then Wg and Dl staining was examined in triple mutant hrs^- wg^- lqf^- clones that express an HRP-tagged form of Dl under Gal4 control. In this case, as in corresponding hrs^- wg^- double mutant clones, co-localization of Wg and Dl was observed in large intracellular puncta. Thus, bulk endocytosis of both endogenous and overexpressed Dl appear normal in lqf^- cells (Wang, 2004).

Although bulk Dl endocytosis appears unaffected by the absence of Lqf, blockage of a relatively small, but specific, subset of Dl endocytic events might escape detection, and this subset might be crucial for signaling activity. To examine this possibility, Dl was co-expressed together with the E3 Ubiquitin Ligase Neuralized (Neur), under Gal4 control, to drive efficient ubiquitination and internalization of the exogenous Dl. It was reasoned that under these conditions, even modest reductions in the rate of Dl endocytosis might cause an abnormal persistence of Dl at the apical cell surface (Wang, 2004).

Wing discs that express uniformly high levels of Dl under Gal4 control accumulate high levels of Dl on the apical cell surface. However, in discs that co-express high levels of both Dl and Neur, this surface accumulation is strongly reduced and Dl accumulates instead in an abnormally large number of intracellular puncta. Clones of lqf^- cells generated in such co-expressing discs do not appear to alter the number or general appearance of these Dl-positive puncta, many of which co-localize with Hrs. However, they do affect the level of Dl staining associated with the apical cell surface (as visualized in discs processed either with, or without, detergent). Such lqf^- clones show residual surface staining of Dl, in contrast to neighboring wild-type cells where surface-associated staining is depleted. It is inferred that lqf^- cells cannot endocytose Dl as efficiently as their wild-type neighbors, accounting for why a difference was detected under sensitized conditions in which the rate of surface clearance appears to be limiting (Wang, 2004).

Significantly, the residual staining of Dl on the surface of lqf^- cells that overexpress Neur and Dl correlates with the failure of these cells to signal. Clones of lqf^- cells that overexpress Neur and Dl fail to activate cut in neighboring cells, even though clones of otherwise wild-type cells that overexpress Neur and Dl show enhanced Dl signaling. Hence, it appears that the impairment in Dl endoctyosis detected in lqf^- clones in this sensitized background correlates with an absolute block in signaling activity (Wang, 2004).

The cytosolic domain of DSL ligands contains multiple Lysines at least some of which serve as acceptors for Ubiquitin. Lqf contains two Ubiquitin Interacting Motifs (UIMs) (Hofmann, 2001). Hence, mono-ubiquitination of DSL ligands might allow Lqf to target DSL ligands for a special subset of endocytic events that are required for signaling activity. By contrast, bulk endocytosis of DSL ligands mediated by interactions with other Ubiquitin-binding adaptor proteins might not suffice to confer signaling activity. To test this hypothesis, whether the signaling activity of the Dl^R+ protein depends on Lqf activity, was investigated (Wang, 2004).

Endocytosis and signaling activity of Dl^R+ depends on the presence of at least one of the two Lysines in the R+ peptide comprising the cytosolic domain. Clones of lqf^- cells that express Dl^R+ fail to induce cut expression in adjacent wing disc cells. However, Dl^R+ protein in these lqf^- clones accumulates both on the apical surface and in intracellular puncta. Moreover, no difference was detected in the punctate, cytosolic accumulation of Dl^R+ between lqf^- and wild-type cells in wing discs that generally overexpress Dl^R+. Both results indicate that bulk endocytosis of Dl^R+ is not significantly altered in the absence of Lqf. Because substitution of both Lysines by Arginine blocks internalization and signaling activity of Dl^Rm, it is inferred that Dl^R+ is targeted for internalization solely by ubiquitination at one or both of these Lysines. Hence, it is suggested that other Ubiquitin-interacting proteins aside from Lqf can target mono-ubiquitinated cargo proteins, such as Dl^R+ or endogenous Dl, for internalization. However, only Lqf appears able to direct endocytosis of these proteins in a way that allows DSL ligands to signal (Wang, 2004).

Both endocytosis and signaling activity of Dl^LDL+ depends on the FDNPVY internalization signal. However, unlike either native Dl or Dl^R+, it was found that clones of lqf^- cells expressing Dl^LDL+ can induce cut expression in adjacent wild-type cells, indicating that the presence of the LDL internalization signal in the chimeric Dl^LDL+ protein bypasses the requirement for Lqf. As observed for clones of wild-type cells overexpressing Dl^LDL+, the 'rescued' lqf^- clones induced cut only when located close to the DV boundary. Nevertheless, their ability to signal, albeit weakly, contrasts with that of lqf^- clones that overexpress native Dl, native Dl plus Neur, or Dl^R+, all of which are devoid of signaling activity. Hence, it is concluded that the FDNPVY signal directs internalization of Dl^LDL+ in a manner that permits the protein to acquire signaling activity even in the absence of Lqf activity (Wang, 2004).

Lqf-dependent endocytosis of DSL ligands might be accompanied by modifications of these ligands, either as a pre-requisite for, or a consequence of, signaling activity. To examine this possibility, it was asked whether the size of Dl protein changes as a consequence of Lqf-dependent endocytosis. Initially, clones of wild-type and lqf^- cells were generated that express Dl tagged by the insertion of six copies of the Myc epitope in the extracellular juxtamembrane domain, and the profile of Dl peptides that retain the Myc epitope was examined by Western blotting. Under these conditions, similar, complex profiles were observed of Myc-tagged Dl peptides from both wild-type and lqf^- cells, corresponding to full-length Myc-Dl protein, as well as several lower molecular weight peptides (Wang, 2004).

This experiment was then repeated using wild-type and lqf^- cells that overexpress Neur and Myc-tagged Dl, the sensitized condition under which residual surface expression can be detected of Myc-tagged Dl in lqf^-, but not in wild-type, cells. In this case, the profile of Myc-tagged Dl is remarkably simple. Wild-type cells show two bands, one corresponding by size to full-length Myc-tagged Dl (~105 kDa) and the other to a Myc-tagged cleavage product of ~50 kDa. By contrast, lqf^- cells show only a single band, corresponding to full-length Myc-tagged Dl. Thus, the failure to clear Dl from the cell surface of lqf^- cells is associated with an apparent failure in Dl processing. These results provide evidence for a Lqf-dependent cleavage of Dl that correlates with Lqf-dependent endocytosis and signaling activity (Wang, 2004).

It is noted that the expected size of the Myc-tagged extracellular domain of Dl is ~75 kDa, whereas that of the complementary, Myc-tagged portion of the ligand containing the transmembrane and cytosolic domains is ~40 kDa. Hence, the 50 kDa Myc-tagged cleavage product must be composed of a C-terminal portion of the extracellular domain, and possibly some or all of the transmembrane and cytosolic domains as well. The relationship of this truncated peptide to the active ligand is presently unknown. It could comprise part, or all, of the active ligand, or alternatively, a non-signaling C-terminal fragment cleaved off in the process of generating an N-terminal signaling fragment. Alternatively, it might be a degradation product generated as a consequence of the activation of Notch by Dl (Wang, 2004).

REFERENCES

Search PubMed for articles about Drosophila Liquid facets

Aguilar, R. C., Watson, H. A. and Wendland, B. (2003). The yeast Epsin Ent1 is recruited to membranes through multiple independent interactions. J. Biol. Chem. 278(12): 10737-43. 12529323

Cadavid, A. L. M., Ginzel, A. and Fischer, J. A. (2000). The function of the Drosophila Fat facets deubiquitinating enzyme in limiting photoreceptor cell number is intimately associated with endocytosis. Development 127: 1727-1736. 10725248

Chen, H., Fre, S., Slepnev, V. I., Capua, M. R., Takei, K., Butler, M. H., di Fiore, P. P. and de Camilli, P. (1998). Epsin is an EH-domain-binding protein implicated in clathrin-mediated endocytosis. Nature 394: 793-797. 9723620

Chen, H., Slepnev, V. I., Di Fiore, P. P. and De Camilli, P. (1999). The interaction of epsin and Eps15 with the clathrin adaptor AP-2 is inhibited by mitotic phosphorylation and enhanced by stimulation-dependent dephosphorylation in nerve terminals. J. Biol. Chem. 274(6): 3257-60. 9920862

Chen, H. and De Camilli, P. (2005). The association of epsin with ubiquitinated cargo along the endocytic pathway is negatively regulated by its interaction with clathrin. Proc. Natl. Acad. Sci. 102: 2766-2771. 15701696

Chen, X., Zhang, B. and Fischer, J. A. (2002). A specific protein substrate for a deubiquitinating enzyme: Liquid facets is the substrate of Fat facets. Genes Dev. 16: 289-294. 11825870

Chidambaram, S., et al. (2004). Specific interaction between SNAREs and epsin N-terminal homology (ENTH) domains of epsin-related proteins in trans-Golgi network to endosome transport. J. Biol. Chem. 279(6): 4175-9. 14630930

Drake, M. T., Downs, M. A. and Traub, L. M. (2000). Epsin binds to clathrin by associating directly with the clathrin-terminal domain. Evidence for cooperative binding through two discrete sites. J. Biol. Chem. 275: 6479-6489. 10692452

Duncan, M. C., Costaguta, G. and Payne, G. S. (2003). Yeast epsin-related proteins required for Golgi-endosome traffic define a gamma-adaptin ear-binding motif. Nat. Cell Biol. 5(1): 77-81. 12483220

Eugster, A., et al. (2004). Ent5p is required with Ent3p and Vps27p for ubiquitin-dependent protein sorting into the multivesicular body. Mol, Biol. Cell 15(7): 3031-41. 15107463

Eun, S. H., Banks, S. M. and Fischer, J. A. (2008). Auxilin is essential for Delta signaling. Development 135: 1089-95. PubMed Citation: 18256200

Ford, M. G., Mills, I. G., Peter, B. J., Vallis, Y., Praefcke, G. J., Evans, P. R. and McMahon, H. T. (2002). Curvature of clathrin-coated pits driven by epsin. Nature 419: 361-366. 12353027

Friant, S., et al. (2003). Ent3p Is a PtdIns(3,5)P2 effector required for protein sorting to the multivesicular body. Dev. Cell 5(3): 499-511. 12967568

Hofmann, K. and Falquet, L. (2001). A ubiquitin-interacting motif conserved in components of the proteasomal and lysosomal protein degradation systems. Trends Biochem. Sci. 26: 347-350. 11406394

Hussain, N. K., et al. (2003). A role for epsin N-terminal homology/AP180 N-terminal homology (ENTH/ANTH) domains in tubulin binding. J. Biol. Chem. 278(31): 28823-30. 12750376

Itoh, T., et al. (2001). Role of the ENTH domain in phosphatidylinositol-4,5-bisphosphate binding and endocytosis. Science 291(5506): 1047-51. 11161217

Kalthoff, C., et al. (2002). Unusual structural organization of the endocytic proteins AP180 and epsin 1. J. Biol. Chem. 277(10): 8209-16. 11756460

Koshiba, S., Kigawa, T., Kikuchi, A. and Yokoyama, S. (2002). Solution structure of the epsin N-terminal homology (ENTH) domain of human epsin. J. Struct. Funct. Genomics. 2(1): 1-8. 12836669

Mills, I. G., Praefcke, G. J., Vallis, Y., Peter, B. J., Olesen, L. E., Gallop, J. L., Butler, P. J., Evans, P. R. and McMahon, H. T. (2003). EpsinR: an AP1/clathrin interacting protein involved in vesicle trafficking. J. Cell Biol. 160: 213-222. 12538641

Le Borgne, R., Remaud, S., Hamel, S. and Schweisguth, F. (2005). Two distinct E3 ubiquitin ligases have complementary functions in the regulation of delta and serrate signaling in Drosophila. PLoS Biol. 3(4): e96. 15760269

Oldham, C. E., et al. (2002). The ubiquitin-interacting motifs target the endocytic adaptor protein epsin for ubiquitination. Curr. Biol. 12(13): 1112-6. 12121618

Overstreet, E., et al. (2003). Either part of a Drosophila Epsin protein, divided after the enth domain, functions in endocytosis of Delta in the developing eye. Curr. Biol. 13: 854-860. 12747835

Overstreet, E., Fitch, E. and Fischer, J. A. (2004). Fat facets and Liquid facets promote Delta endocytosis and Delta signaling in the signaling cells. Development. 131(21): 5355-66. 15469967

Owen, D. J., Vallis, Y., Noble, M. E., Hunter, J. B., Dafforn, T. R., Evans, P. R. and McMahon, H. T. (1999). A structural explanation for the binding of multiple ligands by the alpha-adaptin appendage domain. Cell 97: 805-815. 10380931

Pitsouli, C. and Delidakis, C. (2005). The interplay between DSL proteins and ubiquitin ligases in Notch signaling. Development 132(18): 4041-50. 16093323

Polo, S., et al. (2002). A single motif responsible for ubiquitin recognition and monoubiquitination in endocytic proteins. Nature 416(6879): 451-5. 11919637

Rosenthal, J. A., Chen, H., Slepnev, V. I., Pellegrini, L., Salcini, A. E., di Fiore, P. P. and de Camilli, P. (1999). The epsins define a family of proteins that interact with components of the clathrin coat and contain a new protein module. J. Biol. Chem. 274: 33959-33965. 10567358

Rosse, C., et al. (2003). RLIP, an effector of the Ral GTPases, is a platform for Cdk1 to phosphorylate epsin during the switch off of endocytosis in mitosis. J. Biol. Chem. 278(33): 30597-604. 12775724

Shih, S. C., et al. (2002). Epsins and Vps27p/Hrs contain ubiquitin-binding domains that function in receptor endocytosis. Nat. Cell Biol. 4(5): 389-93. 11988742

Sigismund, S., Woelk, T., Puri, C., Maspero, E., Tacchetti, C., Transidico, P., Di Fiore, P. P. and Polo, S. (2005). Clathrin-independent endocytosis of ubiquitinated cargos. Proc. Natl. Acad. Sci. USA 102: 2760-2765. 15701692

Stahelin, R. V., et al. (2003). Contrasting membrane interaction mechanisms of AP180 N-terminal homology (ANTH) and epsin N-terminal homology (ENTH) domains. J. Biol. Chem. 278(31): 28993-9. 12740367

Tian, X., Hansen, D., Schedl, T. and Skeath, J. B. (2004). Epsin potentiates Notch pathway activity in Drosophila and C. elegans. Development 131(23): 5807-15. 15539484

Wang, W. and Struhl, G. (2004). Drosophila Epsin mediates a select endocytic pathway that DSL ligands must enter to activate Notch. Development 131: 5367-5380. 15469974

Wang, W. and Struhl, G. (2005). Distinct roles for Mind bomb, Neuralized and Epsin in mediating DSL endocytos and signaling in Drosophila. Development 132(12): 2883-94. 15930117

Watson, H. A., et al. (2001). In vivo role for actin-regulating kinases in endocytosis and yeast epsin phosphorylation. Mol. Biol. Cell. 12(11): 3668-79. 11694597

Wendland, B. (2002). Epsins: adaptors in endocytosis? Nat. Rev. Mol. Cell Biol. 3, 971-977. 12461563

Wendland, B., Steece, K. E. and Emr, S. D. (1999). Yeast epsins contain an essential N-terminal ENTH domain, bind clathrin and are required for endocytosis. EMBO J. 18: 4383-4393. 10449404

Xie, X., Cho, B. and Fischer, J. A. (2012). Drosophila Epsin's role in Notch ligand cells requires three Epsin protein functions: the lipid binding function of the ENTH domain, a single Ubiquitin interaction motif, and a subset of the C-terminal protein binding modules. Dev. Biol. 363(2): 399-412. PubMed Citation: 22265678

date revised: 22 August 2022

Home page: The Interactive Fly © 2017 Thomas Brody, Ph.D.