Endocytic trafficking during Drosophila development

Mechanisms of Development 120 (2003) 1265–1282 www.elsevier.com/locate/modo

Endocytic trafficking during Drosophila development Marcos Gonza´lez-Gaita´n Max-Planck Institute for Molecular Cell Biology and Genetics, Pfotenhauerstr. 108, Dresden D-01307, Germany Accepted 12 June 2003

Abstract During the last decade, many of the factors and mechanisms controlling membrane and protein trafficking in general and endocytic trafficking in particular have been uncovered. We have a detailed understanding of the different endocytic trafficking steps: plasma membrane budding, endocytic vesicle motility and fusion with the endosome, recycling, transcytosis and lysosomal degradation. The kinetics and trafficking pathway of many signaling receptors and the relevance of endocytic trafficking during signaling in many mammalian cultured cells are also well understood. However, only in recent years has the role of endocytic trafficking during cell-to-cell communication during development, i.e. during patterning, induction and lateral inhibition, begun to be explored. The contribution of Drosophila developmental genetics and cell biology has been fundamental in elucidating the essential role of endocytosis during these processes. Reviewed here are some of the recent developments on the role of endocytic trafficking during long- and short-range signaling and during lateral inhibition. q 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: Endocytic trafficking; Morphogenetic signaling; Morphogens; Decapentaplegic; Wingless; Hedgehog; Notch; Delta

1. Endocytic trafficking during long-range morphogenetic signaling: Dpp in the wing 1.1. Organizers and long-range morphogens During morphogenesis, cells within developing tissues acquire positional information in a process that involves a particular kind of signaling molecules: the morphogens (Turing, 1952; Wolpert, 1969). Morphogens are ligands which are secreted from a restricted groups of cells, a source. Upon secretion, morphogens move away from the source to form gradients of concentration. Positional information is encoded in the graded distribution of morphogens: cells read their distance to the source by computing the morphogen concentrations which surround them. Long-range morphogens act as organizers, being secreted from a small source and providing positional information to cells far away from the producing cells. Therefore morphogens move far from their source and establish a stable gradient that lasts during several days of development. What cellular mechanisms control the longrange movement of morphogens? What ensures their graded E-mail address: [email protected] (M. Gonza´lez-Gaita´n). 0925-4773/$ - see front matter q 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.mod.2003.06.002

distribution? How are their slopes and ranges stably maintained? To answer these questions it is necessary to know: (i) the key factors involved in the morphogenetic signaling event, i.e. the signaling cascade from the ligand, through the receptor and the effecting transcription factors; (ii) the developmental properties of the signaling event: where is the source? which are the receiving cells? which target genes are activated at which distance from the source? and (iii) how developing cells (as opposite to cultured cells) control membrane and protein trafficking at the molecular level and establish the kinetic parameters of protein movements within and between cells in the tissues where morphogenetic signaling is taking place. This rich pool of interdisciplinary information is starting to be available in the case of morphogenetic signaling in Drosophila. In particular, the role of endocytic trafficking in three morphogenetic signaling pathways, Dpp, Wingless and Hedgehog, are the focus of a number of Drosophila laboratories in recent years. The comparison of the role of endocytosis during these three signaling events uncovers how endocytic trafficking mediates long-range signaling (Dpp) as well as shorter signaling events (Wg, Hh).

1266

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

1.2. Dpp and endocytosis: spreading by planar transcytosis Dpp acts as a long-range morphogen that provides information along the anterior –posterior axis during wing development (Capdevila et al., 1994). Dpp directly activates the expression of target genes in a concentration dependent manner (Lecuit et al., 1996; Nellen et al., 1996). Thus, two target transcription factors, the zinc-finger proteins Spalt and Optomotorblind are expressed in nested domains of expression around the Dpp expressing cells during wing development. Dpp forms a gradient spanning 40 cells diameters that activates Spalt in wing cells at a distance of 15 cells and Omb as far as 30 cells away from the source. Dpp spreading to form such a long-range gradient requires endocytosis at the receiving cells (Entchev et al., 2000; Gonza´lez-Gaita´n and Ja¨ckle, 1999). Three kind of experiments uncovered the role of endocytosis during longrange movement of Dpp: the ‘shibire rescue’ assay, the ‘shadow’ assay and the receptor mosaic assay (Fig. 1A – D) (Entchev et al., 2000). In a set of experiments (shibire rescue; Fig. 1B), endocytosis was blocked in the receiving cells using the thermosensitive Dynamin mutant shibire, while the producing cells at the source were rescued by expressing there a Dynamin transgene. In these developing wings, fluid phase endocytosis was controlled by observing dextran internalization into endosomes: endocytosis was restricted to the source of Dpp and internalization was tightly blocked in the receiving cells. In this scenario, Dpp internalization into endosomes was blocked in the receiving cells and its distribution was reduced to the first three cell rows adjacent to the source, where it appeared as an extracellular weak staining. These observations suggested that long-range spreading requires Dpp endocytosis by the receiving cells. It is, however, still possible that extracellular Dpp is diluted in a larger volume than the endosomes: low levels of Dpp would only be visible when concentrated in the endosomes. The shadow assay, however, shows that indeed endocytosis is essential during long-range Dpp trafficking (Fig. 1C). In the shadow assay, the progression of Dpp emanating from the source is challenged by a patch of shibire mutant cells which cannot perform endocytosis. The assay takes advantage of the thermosensitivity of Dpp expression driven by the gal4 system and the thermosensitivity of shibire. The experiment starts at low temperature, when (i) the shibire cells can perform endocytosis at the permissive temperature and (ii) low levels of Dpp are expressed and therefore it cannot be seen in the receiving cells. Then, raising the temperature has two consequences: (i) endocytosis is blocked in the patch of shibire cells; (ii) a wave of Dpp secreted from the source is travelling through the tissue. When confronted with the shibire clone of cells the wave of Dpp cannot cross the endocytosis defective tissue and therefore a shadow with low levels of Dpp is formed distal to the mutant mosaic. The shadow is,

however, transient because Dpp moves rapidly and in all directions, entering shortly thereafter (within few hours) the shadow area. The Dpp receptor, composed of two proteins, Punt and Thickveins (Tkv), is required for Dpp internalization and for its long-range movement. A patch of Tkv mutant cells is incapable of internalizing Dpp. As a consequence Dpp accumulates around the mutant cells. But extracellular Dpp accumulation is restricted to the cells of the clone which face the Dpp source (Fig. 1D). An explanation of this behaviour of the Tkv clone is that Dpp accumulated around the cells facing the source is unable to progress further into the mutant territory, suggesting again a scenario in which Dpp movement through the target tissue requires its internalization by endocytosis, specifically by Dynamindependent receptor-mediated endocytosis. Based on these three observations, a planar transcytosis model has been proposed in which Dpp movement through the target tissue involves short-range extracellular diffusion (across less than 3 cell diameters) and active transport of the morphogen through the receiving cells by vesicular trafficking involving the endocytosis and re-release of the morphogen molecules (Fig. 1E) (Entchev et al., 2000). In this scenario, endocytic block allows the short-range extracellular diffusion of the morphogen, but impedes its active transport through the cells and thereby its long-range distribution in the tissue. 1.3. An alternative model: Dpp spreading by diffusion The above three assays show that the control of receptormediated endocytosis is required in order to achieve the long-range spreading of Dpp. The planar transcytosis model assumes that Dpp internalized in the receiving cells is resecreted to be internalized (and signal) in the neighbouring cells. The event of Dpp re-secretion from the receiving cells has, however, not yet been directly monitored. Therefore, an alternative formal possibility is that internalization is not required as the first step of Dpp planar transcytosis, but that endocytosis down-regulates another critical molecule by removing it from the plasma membrane. Thus it has been suggested that endocytosis block causes the accumulation of the Dpp receptor at the cell surface and this in turn traps Dpp while moving extracellularly by free-diffusion across the extracellular matrix (Fig. 2A) (Lander et al., 2002). This possibility is supported by the fact that Dpp signaling range is decreased by overexpression of the receptors at the receiving cells (Lecuit and Cohen, 1998). In this scenario, a proper internalization of the receptor ensures that enough Dpp free molecules could circulate to build-up the longrange gradient. To address this possibility, the differential equations governing Dpp trafficking by free-diffusion through the extracellular space were studied (Fig. 2B) (Lander et al., 2002). The model of Lander et al. assumes: (i) extracellular diffusion of Dpp according to the Fick’s law; (ii) binding/release of Dpp to the receptors at the plasma

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

1267

Fig. 1. Experimental assays to study the role of endocytosis during Dpp morphogen movement. (A) Dpp long-range gradient. Dpp forms a gradient of concentration from the source, (red cells, left) and appears graded both extracellularly and when internalized in the endosomes. (B) ‘Shi rescue assay’. It consists of mosaic discs where the producing cells at the source (red cells, left) are wildtype (shiþ) while the receiving cells (right) are mutants for a thermosensitive mutation in Dynamin (shits) and thereby endocytosis defective cells. Endocytosis is blocked for 6 h. Dpp, degraded, disappears from the endosome of receiving cells, while extracellular Dpp is restricted to the cells adjacent to the source. (C) ‘Shadow assay’. It starts by producing a tissue void of Dpp at the receiving cells. A patch of shits cells is generated at the permissive temperature. Then Dpp is pulsed from the source and endocytosis is blocked at the restrictive temperature of shits. The front of Dpp advancing through the tissue meets the mutant clone. It cannot move across the endocytosis defective territory and forms a shadow behind the clone. (D) ‘Receptor mosaic assay’. Consist in a mutant clone that lacks the receptor. The clone is exposed to the gradient for many cell generations. As a consequence, Dpp, which cannot be internalized by receptor-mediated endocytosis, accumulates in between the cells of the clone which are nearer the source. (E) A model of Dpp spreading by planar transcytosis. Dpp is secreted to an apical location below the adherens junctions from the producing cells. At the extracellular space, it does not move a long distance by simple diffusion. Instead it is internalized by endocytosis and accumulates in an apical endosome from where it is degraded or recycled again. This way it moves on through the next cells and thereby spreads through the target tissue forming a gradient. Degradation in each of the receiving cells ensures the formation of a stable gradient of concentration.

membrane; (iii) endocytosis, recycling and degradation of the free receptors and endocytosis þ degradation of the ligand bound to the receptors; (iv) boundary conditions in the modeling area: Dpp is degraded in a sink region of

the developing tissue and (v) before GFP-Dpp emanates from its source, there is no Dpp in the receiving tissue. In the model, the developing wing is simplified in its geometry to a one-dimensional diffusion problem in which

1268

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

Fig. 2. Modelling the free-diffusion of morphogens. (A) Endocytosis block may cause surface receptor accumulation. (i) Internalization of receptors (II) by endocytosis regulates the amount of surface receptors at the plasma membrane. Receptors internalized by endocytosis are targeted to degradation. Receptors at the plasma membrane bind to the morphogen (green) and hamper its further spreading through the target tissue. (ii) In endocytosis defective cells (grey cells) surface receptors are not down-regulated by endocytosis, thereby causing accumulation of receptors at the plasma membrane. Accumulated at the plasma membrane, high levels of receptors block the progression of the morphogen through the receiving cells. (B) Parameters affecting the free-diffusion of morphogens when surface receptors hamper their spreading. (C) System of differential equations which describes the free-diffusion and titration of morphogen spreading by surface receptors. (D –F) Behaviour of the shadow assay under a model of free-diffusion of the morphogen and spreading is impaired by surface receptors. (D) Internalized Dpp in the shadow assay (gray, shibire-clone) before the steady-state (5 h). Note that a shadow is seen behind the clone (i), but also before the clone (ii). Also note that, in the clone where internalization is blocked internalized Dpp is increased (iii). This is due to the fact that the levels of extracellular receptors are 10 times those of the wildtype situation and endocytosis is not

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

the morphogen is introduced at a rate n at one location (the source), and absorbed at another, a sink at the edge of the gradient where the morphogen is totally degraded. While it is clear that there is a source of Dpp in the developing wing, no sink exists in the developing wing primordium, i.e. no region in the disc is dedicated to the total degradation of the remaining morphogen. The existence of such sink by itself would generate gradients resembling the one observed for Dpp regardless of the mechanism of transport proposed. Within the modeling area, diffusion of Dpp would behave according to Fick’s second law

›½L=›t ¼ D›2 ½L=›x2 where ½L is the concentration of the morphogen, t is time, x is distance and D the diffusion coefficient. In addition to this, Dpp spreading is governed by Dpp/receptor binding and dissociation (with constants kon and koff, respectively) as well as the rates of biosynthesis, endocytosis and degradation of the receptor and the receptor/ligand complexes (Fig. 2B,C). Then, the receptor/ligand association rate constant (kon ¼ 3 £ 105 or 1.2 £ 105 M21 s21) is introduced from the literature (Iwasaki et al., 1995; Natsume et al., 1997). The values of the other parameters were scanned to fit the gradient shape and kinetics of gradient formation observed for Dpp. The values of the different trafficking rates selected this way were smaller than those previously estimated experimentally for EGF and its receptor (Lauffenburger and Linderman, 1993). This analysis shows that there is a universe of parameters for which a gradient resembling that observed for Dpp can be formed by extracellular diffusion and receptor binding. Lander et al. argued that another interesting consequence of the model is that in a scenario of diffusion and receptor binding as proposed by their model, shibire clones would also generate distal shadows. This is due to the fact that endocytosis block would cause the accumulation of surface receptors at the shibire mutant cells thereby trapping Dpp on its travel to form the gradient (Fig. 2D – F). However, in the model, both the shadow and the clone behave different from the experimentally observed situation: (i) surprisingly shadows can be seen distally (Fig. 2Di), but also proximally to the shibire clone (Fig. 2Dii); (ii) the result of blocking internalization in the clone is that the internalized Dpp is increased with respect to wildtype cell (Fig. 2Diii), which was not observed in the shibire clones: internalized Dpp disappeared in the endocytosis-defective cells, it was not increased! and (iii) the amount of extracellular Dpp at the clone increased by a factor of 40 (Fig. 2Eiv), which was also not observed.

1269

In summary, the shibire clone in a scenario of diffusion þ receptor binding of Dpp behaves different to the observed result (cf. Figs. 2G vs. 1B). In addition, in the simulation of the shibire clone, in the endocytosis defective cells, the receptor levels increase by a factor of 10 instantaneously, instead of increasing progressively according to the differential equations and the biosynthetic rates proposed when endocytosis is blocked in the shibire clone. As a consequence, when the Dpp propagation front reaches the clone, it finds a concentration of the receptor 10 times bigger than normal causing the formation of the shadow behind. After 6 h of endocytic block, the levels of the Dpp receptor determined experimentally using antibodies do not increase by a factor of 10, but only suffer a marginal increase (Pantazis and G-G, unpublished results). Another issue is the plausibility of the spreading of Dpp by intracellular trafficking. It has been argued that if moving by intracellular trafficking, Dpp has to be transported across a single cell in 150 s on average in order to account for the speed of expansion of the Dpp gradient in less than 8 h. For various mesenchymal cells in culture, the rates of recycling of ligands such as Transferrin or EGF imply mean transit times of minutes to hours (Sheff et al., 1999; Shitara et al., 1998). No data is, however, available in the case of developing epithelial cells. Interestingly, secretion to the apical-junctional complex in MDCK cells mediated by the exocyst (Grindstaff et al., 1998) and accumulation of receptors to the junctional region (Banerjee et al., 1987; Fehon et al., 1991; Tomlinson et al., 1987) may make the transport from cell to cell more effective than the threedimensional random walk predicted by Fick’s law which is on the basis of the model. It is worth noting that Dpp accumulates in apical endosomes (Entchev et al., 2000), also suggesting that the trafficking routes of Dpp in the developing epithelial wing cells might be far from random throughout the full volume of the cell. It will be interesting to estimate experimentally the rates of dissociation of Dpp from its receptor and, more generally, its trafficking routes and kinetics at the receiving cells. In the future, a fluent crosstalk between experimental data and modeling of non-linear systems will be necessary in order to integrate trafficking and developmental approaches. 1.4. Endosomal dynamics during Dpp signaling Trafficking through the endocytic pathway involves a number of intermediate compartments, including the early endosome, where the decision is made to target the endocytic

R totally blocked in the simulation, but decreased by a factor of 10. Therefore, although endocytosis is decreased, the huge amount of surface receptors account for increased receptor-mediated endocytosis inside the clone. This was not seen in the experimental data. (E) Surface Dpp in the shadow assay (gray, shibire-clone) before the steady-state (5 h). Surface Dpp is increased by a factor of 40 (iv). (F) Internal Dpp after 24 h. No shadow can be seen. This, however, does not represent the steady-state situation. (G) Schematic representations of the behaviour of the shadow assay in the diffusion/ hampering receptors model. The result differs substantially from the observed results (cf. Fig. 1C).

1270

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

cargo to recycling or to degradation at the late endosome/lysosome (reviewed in Mukherjee et al., 1997). Each particular trafficking step through the endocytic pathway is controlled by a small GTPase of the Rab family. Thus, Rab5 controls the step from the plasma membrane to the early endosome, Rab7 controls the targeting to the degradation pathway and Rab4/Rab11, the recycling pathway (reviewed in Novick and Zerial, 1997). Consistent with their role in endosomal dynamics, these Rab proteins accumulate in distinct endosomal compartments: Rab4/Rab5 in the early endosome, Rab11 in a recycling endosome and Rab7 in the late endosome. In Drosophila, these factors are highly conserved in aminoacid sequence (between 75 and 85% identity, 83– 91% similarity), localization and function (Entchev et al., 2000 and unpublished results). The mutant phenotype of these Rab proteins and their localization with respect to Dpp uncovered an essential role of endosomal dynamics during Dpp signaling (Entchev et al., 2000). Thus, blocking Rab5 in the receiving cells causes a reduction of Dpp signaling range, as monitored by the distance to the source at which target genes are activated. Conversely, Rab5 overexpression causes an expansion of the Dpp signaling range (Fig. 3C). Similarly, enhanced degradation caused by expression of a gain of function Rab7 mutant protein in the receiving cells reduces the signaling range. These results are consistent with a scenario in which internalized Dpp is directed to degradation and recycling in the receiving cells in a mechanism controlled by Rab proteins which determines the efficiency of Dpp spreading and thereby the range and slope of the gradient. In addition to controlling Dpp spreading, endosomal dynamics may also determine the efficiency of Dpp signal transduction. Indeed, endosomal trafficking turned out to critically control the signal transduction event by other TGF-beta-like ligands in mammalian cells. 1.5. TGF-beta signal transduction and endosomal dynamics in mammalian cells: SARA, Smurfs, Clathrin and Caveolin TGF-beta signals through heteromeric complexes of Type II and Type I transmembrane Ser-Thr kinase receptors (reviewed in Massague´, 1998). Ligand induces the assembly of a heteromeric receptor complex within which Type II receptor transphosphorylates and activates the Type I receptor (Wrana et al., 1994). The Type I receptor in turn phosphorylates the receptor regulated Smads, R-Smads (Smad1,2,3,5,8) (Miyazono, 2000). Once phosphorylated, the R-Smads bind to another Smad, the Co-Smad Smad4, forming a complex which is imported in the nucleus and elicits transcription of target genes. Finally the inhibitory Smads, I-Smad (Smad6,7), negatively regulate TGF-beta signaling by binding to the Type I receptor and recruiting the Smurf E3 ubiquitin ligases, which direct the ubiquitindependent degradation of the TGF-beta receptor– Smad7 complexes (Ebisawa et al., 2001; Kavsak et al., 2000).

Like other receptors, the TGF-beta receptors are down-regulated by internalization both constitutively and upon ligand binding (Massague and Kelly, 1986; Sathre et al., 1991). Endocytosis here may affect negatively the signal transduction event by removing the receptors from the plasma membrane where the ligand binds the receptors and the receptors phosphorylate the transcription factor. The discovery of an endosomal protein, SARA, as an essential adaptor between the activated Type I receptor and the transcription factors, Smad2/3, uncovered the endosome as a platform during signal transduction (Tsukazaki et al., 1998). In mammalian cells, SARA acts specifically during Smad signaling involving Smad2/3 which mediate TGF-beta/Activin signaling, but not Smad1/5/8 which mediate BMP signaling (Tsukazaki et al., 1998). In Drosophila, however, it has been suggested that Drosophila SARA is also involved in Dpp signaling, a BMP pathway mediated by an Smad1 homolog, Mad (Bennett and Alphey, 2002). SARA is targeted to the endosome through its FYVE finger domain (Hayes et al., 2002; Itoh et al., 2002; Panopoulou et al., 2002), which specifically binds to PI(3)P when in a bilayer, a situation specific of the early endosomal membrane and the internal vesicles of the multivesicular bodies (Gillooly et al., 2000). In addition to its role as an adaptor, SARA itself is involved in endosomal dynamics in a process that also requires its FYVE domain (Hayes et al., 2002; Itoh et al., 2002; Panopoulou et al., 2002), reminiscent of other FYVE domain proteins such as EEA1 and Rabenosyn which are involved in endocytic vesicle tethering and fusion with the endosome (Gillooly et al., 2001; Wurmser et al., 1999). What role does SARA play during Smad phosphorylation? Three possibilities have been proposed: (i) targeting of Smad to the receptor when localized at the endosome, (ii) Smad phosphorylation at the endosome and (iii) diversion of activated receptors away from the degradation pathway. Based on the fact that SARA binds directly to the Type I receptor and Smad2 it was proposed that SARA acts by targeting Smad to the receptor. This event would take place at the endosome, mediated by the targeting of SARA itself to this compartment by means of its FYVE domain (Fig. 3D) (Tsukazaki et al., 1998). However, if endocytosis is blocked, SARA is still found in a trimeric complex with the receptor and Smad2, suggesting that this interaction does not need the endosomal milieu (Penheiter et al., 2002). While the trimeric complex forms at the plasma membrane, phosphorylation of Smad2 by the Type I receptor is blocked in the absence of endocytosis, suggesting that SARA endosomal localization is essential to allow Smad2 phosphorylation. An interesting model for the role of the endocytic pathway during TGF-beta signal transduction has been suggested recently by the Wrana Laboratory, which originally discovered SARA (Fig. 3E) (Di Guglielmo et al.,

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

1271

Fig. 3. Role of endocytic trafficking during TGF-beta-like signaling. (A–C) Essential and rate-limiting role of Rab5 during Dpp signaling. (A) Wild-type. Dpp emanating from the source (green cells) activates target genes such as Spalt at a distance from the source (red cells). (B) Dominant negative Rab5 mutant in the posterior receiving cells (dash) reduced the range of Dpp signaling: only cells adjacent to the source activate the target gene. (C) Rab5 overexpression in the posterior receiving cells expands the range of Dpp signaling. To drive the mutant Rab5 proteins the Engrailed-gal4 driver (en < ) was used. A/P, anterior/posterior boundary. Arrows represent the range of Dpp signaling. (D) Model of SARA-mediated TGF-beta signal transduction from the endosome. Binding of TGF-beta to the receptor at the plasma membrane causes phosphorylation of the Type I receptor and activation of its kinase. At the endosome, SARA brings the Smad2 transcription factor to the receptor. Then Smad2 is phosphorylated and becomes active and imported into the nucleus. (E) A model of endosome/caveolae competition for TGF-beta signaling. CCP, Clathrin-coated pit. For details, see text.

2003; Tsukazaki et al., 1998). They showed that TGF-beta receptors internalize both into Caveolin- and EEA1-positive vesicles. These vesicular structures are the target compartments of the endocytic vesicles generated by two different molecular mechanisms: Clathrin-mediated endocytosis generates EEA1-positive vesicles and caveolae, deep invaginations of the membrane, generate vesicular structures containing the raft-protein Caveolin. Rafts are lipid microdomains enriched in cholesterol which has been

proposed to play an essential role as signaling platforms (Simons and Ikonen, 1997). However, in the case of TGFbeta signaling, it seems that the Clathrin dependent internalization of the receptors into the EEA1-positive endosome, where the Smad2 anchor SARA is enriched, promotes TGF-beta signaling, while the lipid raft-caveolar internalization pathway is required for rapid receptor degradation. Thus, in the Caveolin-positive compartment the receptor colocalizes with Smad7 –Smurf2, two factors

1272

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

Fig. 4. Endocytosis, lysosomal degradation and asymmetric Wg signaling. (A –E) Embryonic segmentation of the cuticle involves four different signaling pathways. (A) Stages 9 and 10. Mutual activation of Wg (red) and Hh (blue) in adjacent cells. Wg is distributed symmetrically at both sides of its source. (B) Stage 11. Ser (grey) is expressed between the Hh and Wg stripes due to repression from both signaling pathways. Wg is already distributed asymmetrically. (C) Stage 12. Rho (green) is expressed in the cells between Hh and Ser. This is because either ligand can activate Rho expression. Anterior to Hh and posterior to Ser, Rho is not expressed due to repression by Wg. (D) Spi (green) is secreted from the Rho expressing cells and activate the expression of Svb (orange), the master regulator of denticle differentiation. Spi activates Svb expression in a longer range towards anterior than towards posterior. This is due to long-range

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

which are involved in receptor degradation and signaling down-regulation. Thus, segregation of TGF-beta receptors into distinct endocytic compartment regulates Smad activation and receptor turnover mediated by Smurf. Smurf proteins are E3 ubiquitin ligases that inhibit TGFbeta signaling and ubiquitinate both the TGF-beta receptors and Smad proteins (Kavsak et al., 2000; Lin et al., 2000; Zhu et al., 1999). Consistently, a Drosophila Smurf mutation is unable to inhibit Dpp signaling and expands indeed the signaling range (Podos et al., 2001). Smurf proteins seem to act at three levels during TGF-beta signaling inhibition: (i) ubiquitination of the R-Smads to target it to the proteosome for degradation (Lin et al., 2000; Zhu et al., 1999); (ii) binding to the I-Smad Smad7 to target it to the receptor where it competes with the R-Smad for receptor binding (Ebisawa et al., 2001; Kavsak et al., 2000) and (iii) target the receptor –Smad7 complex for degradation a process that at least partially takes place in the lysosome, thereby down-regulating the receptor (Kavsak et al., 2000).

1273

While Dpp signals over a long-range during wing development across 40 cell diameter, other signaling molecules, like Wingless, are also distributed as a gradient, but with a short-range. During embryogenesis, the Wg signal extends across 4 cell diameters in the developing epidermis. While it is not clear whether Wingless acts itself as a morphogen (Martinez Arias, 2003), its distribution as a gradient and/or its signal transduction mechanism seems also to be mediated by endocytic trafficking. The role of endocytosis during Wingless signaling was best studied during segmentation (Fig. 4A –G).

6 –11 show a smooth surface also called ‘naked cuticle’ (Fig. 4E). During embryonic segmentation, four different signaling pathways, (Hedgehog (Hh), Wingless (Wg), Notch (N) and EGF-receptor) interact with one another to give rise to the spatial information which lies on the basis of this stereotyped spatial distribution of cuticular elements. Early during embryogenesis, Hh and Wg are expressed in adjacent stripes of cells (Fig. 4A). At this time, Hh and Wg are involved in a paracrine signaling loop: Hh activates expression of Wg in the adjacent cells and vice versa (DiNardo et al., 1988; Ingham et al., 1988). In addition, both Hh and Wg repress expression of a third signaling molecule, the Notch ligand Serrate (Ser) which is thereby expressed in the middle of the segment (rows 4– 6) where neither Hh (rows 11,12) nor Wg (row 10) reach (Fig. 4B) (Alexandre et al., 1999; Wiellette and McGinnis, 1999). Finally a fourth ligand, the EGF-receptor ligand Spitz, is secreted from a stripe of cells (rows 1– 3) between Hh and Ser. Restricted secretion of Spitz is mediated by Rhomboid (Rho), which is expressed in these cells (Golembo et al., 1996). Rho expression is activated by both Hh and Ser and repressed by Wg, hence the localization of the Rho stripe and restricted Spitz secretion (Fig. 4C) (Alexandre et al., 1999; Gritzan et al., 1999; Sanson et al., 1999). A single master transcription factor, Shavenbaby (Svb), controls the differentiation of the denticles in the cell rows 12/1 – 5 (Payre et al., 1999). Svb expression is activated by Spitz emanating from rows 1– 3 and repressed by Wg secreted from row 10 (Fig. 4D). Wg spreading from row 10 is asymmetric, with a range of four rows towards anterior and only one cell row towards posterior. This explains the skewed localization of Svb at both sides of the Spitz secreting cells. Therefore, positional information is determined in this case by an array of signaling molecules arranged in a spatial sequence and by the asymmetric range of spreading of one of the ligands, Wg.

2.1. Segmentation: an array of signaling molecules and the asymmetric spreading of Wg

2.2. Asymmetric wingless signaling in the embryo: asymmetric degradation

Drosophila larval epidermis shows a repetitive pattern of cuticular structures that manifest the segmental organization of the insect body (Fig. 4E). Each epidermal segment derives from a stripe of 12 rows of embryonic cells. On the ventral surface of the larva, cell rows 1 – 5 and 12 differentiate denticles which are small hair-like structures at the apical side of the epidermal cells, while cell rows

Wg is distributed symmetrically at the beginning of embryogenesis (Fig. 4A) and its posterior range of spreading is restricted later (Fig. 4B –D) (Gonzalez et al., 1991). Secreted Wg seems to be stabilized by its receptors at the anterior side, since posterior restriction correlates with a posterior reduction of the RNA levels of Wg receptors, Frizzled and Frizzled2 (Dubois et al., 2001). Another level

2. Endocytic trafficking during short-range morphogenetic signaling: Wg in the embryo

R repression of Svb by wingless towards anterior and short Wg repression towards posterior. As a consequence Rho expression domain is not centered with respect to Svb. (E) Differentiation of denticles depends on Svb which is due to the interplay of four signaling pathways (Wg, Hh, Rho, Ser) distributed in an array of cells along the anterior–posterior axis. Positioning of Svb depends on Rho/Spi and asymmetric signaling by Wg. 1 –12, cell rows within the segmental unit (delimited by vertical lines). (F) Secreted Wg is internalized into in an early endosomal structure. Wg is degraded in a lysosome where it could not be detected. Its distribution is asymmetric. (G) HRP-Wg reveals the asymmetric degradation of Wg. HRP-Wg is symmetrically distributed. Due to the HRP moiety, it can be detected both in the early endosome and the lysosome. It therefore uncovers enhanced lysosomal degradation in the posterior cells. As a consequence it shows a symmetric distribution: HRP-Wg in early endosomes of anterior cells; HRP in the lysosomes of posterior cells. In posterior cells, enhanced degradation is controlled by Spi signaling (green dots) emanating from Rho expressing cells (green nuclei).

1274

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

of regulation for the posterior spreading of Wg is its specific degradation at the lysosome: Wg is degraded at higher rates in the posterior cells than in the anterior ones (Dubois et al., 2001). This was shown by monitoring a functional horseradish peroxidase (HRP)-Wg fusion (Dubois et al., 2001). HRP in the fusion persists in the lumen of the degradative compartment even after Wg itself has been digested. At a stage when Wg already spreads asymmetrically from row 10 and is only seen in the posterior cells adjacent to the source (Fig. 4F), HRPwingless shows a symmetric distribution and can be seen in posterior cells distant from the source (Fig. 4G). Therefore, the posterior restriction of Wg requires its differential degradation in posterior cells (Fig. 4F,G). At the EM level, HRP-Wg is indeed seen in degradative compartments: multivesicular (MVB) and multilamellar bodies (MLB), confirming previous immuno-EM results (Gonzalez et al., 1991; van den Heuvel et al., 1989). Although MVB/MLB containing HRP-Wg are seen in both anterior and posterior cells, Wg seems to be more actively degraded in the posterior cells, since HRP-Wg-labelled MVB/MLB in the posterior cells are four times more frequent than in anterior cells. Consistent with the idea that the asymmetric distribution of Wg is mediated by its differential degradation, treatment with chloroquine, a lysosomal inhibitor, or reduction of internalization by endocytosis and lysosomal degradation with Clathrin heavy chain and deep orange mutations abolishes the asymmetric distribution of Wg (Dubois et al., 2001). How is this differential degradation controlled? Restricted Wg spreading towards posterior is under the control of Hh: in Hh pathway mutants, Wg distribution remains symmetric (Sanson et al., 1999). As described above, Hh activates Rho expression in posterior cells determining the restricted secretion of the EGF-receptor ligand Spitz, which in turn elicits the transcription of an unknown factor mediating enhanced Wg degradation in the cells that received the Spitz signal (Fig. 4G). In a number of cases, EGF-receptor signaling has been shown to control endocytic trafficking by interfering with the Rab5 endocytic machinery (Lanzetti et al., 2000; Tall et al., 2001), suggesting this small GTPase and its regulators/ effectors as possible targets of Spitz control. 2.3. Endocytosis and Wg signaling What signaling step is affected by the differential degradation of Wg? Spreading? Transduction? The fact that HRP-Wg is seen far from the source in the posterior cells, suggests that Wg spreading is not affected by differential degradation. An alternative possibility is that Wg signal transduction is affected by the degradation of Wg. Future work will elucidate which of this two possibilities are more plausible, and whether wg signal transduction occurs from an intracellular locale. Wg is internalized by Dynamin-mediated endocytosis into vesicular structures in the target tissue (Bejsovec and Wieschaus, 1995). In shibirets mutants, Wg is only able to

elicit signaling in cells adjacent to row 10. Therefore, long-range Wg activity requires endocytosis. Again, is endocytosis required for Wg release from the producing cells, its spreading or its transduction? The answer is unclear. On one hand, the shits embryonic phenotype is reminiscent of a secretion-defective Wg mutant, suggesting that endocytic block affects Wg release (Bejsovec and Wieschaus, 1995; Gonzalez et al., 1991). This possibility has also been proposed for Wg during wing development (Strigini and Cohen, 1999). On the other hand, endocytic block restricted to the receiving cells also affects Wg signaling, consistent with a endocytic role during either spreading or transduction of the signal (Moline et al., 1999). Like in the case of Dpp, a possible role of endocytosis during the spreading of Wg is that it initiates the planar transcytosis of the morphogen: Wg is internalized by endocytosis, traffics through the endocytic pathway and is re-secreted again to move further into the target tissue. Wg recycling in this way was observed using a functional Wg-GFP fusion (Pfeiffer et al., 2002). Fluorescence timelapse microscopy of Wg-GFP shows that the Wg internalized in endocytic vesicles can return to the cell surface. Although recycling could only be observed in the apical surface, which is optically accessible, it is still possible that re-secretion also occurs laterally within the epithelium, thereby directing the ligand to the neighbouring cells. Ligand recycling is a key feature of the planar transcytosis model of transport: the ligand spreads by repeated cycles of internalization, recycling, and presentation to further cells. The observation of Wingless recycling shows that planar transcytosis is plausible. However, one should note that, in the above experimental situation, very few endocytic vesicles containing GFPWingless were present in non-expressing cells. Therefore, either a small number of vesicles are sufficient for transport or transcytosis is not an important contributor to Wg transport. In the wing, it has been argued that endocytosis is indeed not required for the dispersal of the signal. This was based on the absence of Wg shadows (Strigini and Cohen, 1999) when performing a shadow assay as described above for Dpp (Entchev et al., 2000). However, in the case of the Wg experiment, the shadow assay was performed in a steadystate situation. In this experimental condition, before Wg disappears from the distal side of the endocytosis-defective mutant clone, molecules of Wg emanating from the sides will fill this distal region rapidly. As a consequence, like in the case of Dpp, no distal Wg shadows will be formed if the experiment is started in a steady-state situation if Wg moves rapidly and in all directions. Wg moves rapidly: the gradient can expand at a speed of 15 cells in 30 min. It also moves in all directions. To observe shadows the experiment should be done in conditions in which the propagation front of Wg gradient expansion meets a clone of Shibire mutant cells as performed in the Dpp shadow assay.

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

Whatever the mechanism of Wg dispersal (diffusion vs. planar transcytosis), is Wg transported in vesicular structures while spreading through the receiving tissue? A recent report shows that, like Hh (see below), Wingless is palmitoylated (Willert et al., 2003), suggesting that it would not appear as a soluble protein in the extracellular space, but associated to a lipidic millieu. Interestingly, Wg has been seen in the same endocytic compartment as the argosomes, lipidic fragments that disperse over long distances in the wing epithelium (Greco et al., 2001). The molecular and cellular nature of the argosomes and the question whether they serve as lipidic carriers for signaling molecules such as Wg or Hh remains yet to be fully understood, but such mechanism of morphogen dispersal is certainly tantalizing.

3. Trafficking during short-range morphogenetic signaling: Hh in the wing and the embryo Like Dpp and Wg, Hh forms concentration gradients during development which instructs cells about their positional information. Hh shows post-tranlational modifications by covalent binding to two lipid moieties: cholesterol (Porter et al., 1996) and palmitoyl acid (Chamoun et al., 2001; Pepinsky et al., 1998). This implies that Hh is a membrane-tethered molecule. Indeed Hh range of spreading is shorter than for other morphogens, such as Dpp. In the case of the developing wing Hh elicits signaling not farther beyond 10 cells and in the embryo, not beyond four cells. But being membrane-tethered, how is Hh released from the producing cells? How does it spread through the target tissues? What are the mechanisms underlying the transduction of the signal? These are indeed unsolved issues. In the producing cells, Hh family proteins undergo a autoproteolytic cleavage under the control of the C-terminal domains. Concomitantly, a cholesterol moiety is added to Hh N-terminal domain at its C-terminus to give raise to the active Hh fragment, which is thereby linked to the membrane (Beachy et al., 1997) and is found in lipid rafts (Rietveld et al., 1999). Strikingly, the cholesterol-modified membrane-tethered N-terminal Hh fragment carries the morphogenetic signaling functions both at long- and shortrange, while the C-terminal domain, secreted and soluble, is inactive for signaling (Beachy et al., 1997). Another lipid modification of Hh, a palmitoylation of its N-terminal cysteine, has been shown to occur in vertebrates and Drosophila (Chamoun et al., 2001; Pepinsky et al., 1998). The factor that catalyses Hh palmitoylation, Skinny hedgehog, shares sequence homology with O-linked acyl transferases. These are cytosolic enzymes that therefore add the palmitoyl moiety to the cytoplasmic side of transmembrane proteins. It is therefore intriguing how, in the case of Hh, they catalyse the lipid modification on the luminal side of the protein.

1275

3.1. Hh release and spreading: cholesterol modification What are these lipid modifications good for? In the case of the cholesterol modification, it is controversial whether cholesterol modification restricts or widens the range of secreted Hh. This question was addressed using a truncated N-terminal Hedgehog transgene which does not undergo cholesterol modification (Burke et al., 1999; Lewis et al., 2001). In a mesenchymal situation, the developing mouse limb, this non-modified Hh is only able to elicit short-range signaling, suggesting that the cholesterol modification is essential for long-range spreading of the ligand (Lewis et al., 2001). In contrast, in an epithelial situation, the Drosophila developing wing, the corresponding non-modified Hh shows a longer spreading range, leading to the opposite proposal that the cholesterol moiety restricts Hh spreading. In the case of the wing, absence of cholesterol modification leads to a mistargeting of Hh release (Burke and Basler, 1996): in the wildtype, Hh is seen in the basolateral side of the epithelium, while non-cholesterol-modified Hh is found in the apical side of the developing wing. This raises the possibility that cholesterol controls Hh secretion into a side of the epithelium where the extracellular milieu might determine the range of spreading. Thus, while the apical, luminal side of the developing wing will allow freediffusion of Hh to a long distance from the source, in the case of the mesenchymal cells, the limiting factor might not be release/spreading, but the transduction of the signal (see below). In another developing context, the embryonic segmentation of the epidermis, the cholesterol modification of Hh has also been recently shown to be responsible for its targeting to large punctate structures, possibly an endocytic compartment, at the apical pole of the receiving cells (Gallet et al., 2003). The proper targeting of cholesterol-modified Hh is mediated by a sterol-sensing domain protein, Dispatched (Burke et al., 1999; Ma et al., 2002). It is therefore possible that cholesterol modification endows Hh with targeting information during trafficking in polarized epithelial cells, a role for cholesterol which is reminiscent of cholesterol-enriched lipid rafts in apical targeting in the biosynthetic pathway of epithelial cells (Simons and Ikonen, 1997). Once properly targeted to the right side of the cells, Hh is released to a extracellular space of the epithelium where proteoglycans synthesized by the GAG transferase enzyme Tout-velou mediate the movement of modified Hh throughout the tissue (Bellaiche et al., 1998; Gallet et al., 2003; The et al., 1999). In addition to the role of cholesterol modification and proteoglycans, the Hh receptor Patched (Ptc) also plays a key role to restrict Hh spreading. High levels of Ptc restricts Hh movement throughout the tissue (Chen and Struhl, 1996): Ptc overexpressing clones cast shadows behind, like in the shadow assay. Hh is involved this way in an autoregulatory loop that limits its spreading: Hh signaling upregulates Ptc, which in turn binds to Hh and titrates it, impeding its long-range movement. Subsequently, the levels

1276

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

of Ptc at the plasma membrane are regulated by Dynamindependent internalization initiated by Ptc binding to Hh (Denef et al., 2000), which lead to targeting to endocytic vesicles and MVBs (Capdevila et al., 1994). It is, however, unclear whether endocytosis itself plays any role during Hh spreading and to date no data suggests that Hh movement involves a planar transcytosis mechanism. 3.2. Hh signal transduction and the endocytic pathway Ptc acts as a Hh receptor, but instead of eliciting signaling, its function is to repress the signaling pathway. In the absence of the ligand, Ptc inhibits a latent, tonic signaling activity of another transmembrane protein, Smoothened (Smo) (reviewed in Ingham and McMahon, 2001). Ptc binding destabilizes Smo (Denef et al., 2000). Binding of Hh to Ptc, releases its inhibition of Smo. This release is mediated by Hh-induced internalization of Ptc, followed by phosphorylation of Smo, which in turn is stabilized and accumulates at the plasma membrane (Denef et al., 2000). These observations suggested that Hh elicits signaling by diverting Ptc into a distinct trafficking route that would release Smo from Ptc repression. Consistent with this, a mutant in the sterolsensing domain of Ptc which is not affected in binding/ sequestration of Hh, is unable to repress the intrinsic signaling activity of Smo (Chen and Struhl, 1996; Martin et al., 2001; Strutt et al., 2001). The sterol-sensing domain has been implicated in the targeting of proteins and lipids through specific membrane compartments (Kuwabara and Labouesse, 2002). Indeed, the sterol-sensing domain mutants in Ptc are mistargeteds and do not anymore colocalize with Smo in the endocytic compartment (Martin et al., 2001). In summary, the role of Hh as a signaling molecule would then be to get rid of a molecule, Ptc, that acts as a repressor of a Smo-dependent tonic signaling. In mammalian cells in culture, the intricacies of this fascinating possibility have been studied in some detail by the Roelink group (Incardona et al., 2002). Ptc and Smo colocalize extensively in the absence of Sonic Hedgehog (Shh), a mammalian Hh homolog. Upon ligand binding, Shh, Ptc and Smo are internalized together, but become subsequently segregated into different compartments: the Ptc-Hh complex is targeted for degradation at the lysosome while Smo is not. The sorting of Smo requires an acid luminal endosomal environment in which the membrane is enriched in lysobiphosphatidic acid (LBPA), suggesting a specific trafficking pathway during Hh signaling. 3.3. Cholesterol-dependent Hh trafficking and Hh functional diversity Cholesterol modification of Hh and the sterol-sensing domain in Ptc may play key roles in trafficking through this specific pathway. An interesting possibility is that, during Hh signaling, activation of different target genes is mediated by Hh presentation from different membrane compartments

determined by the lipid modification of Hh (Gallet et al., 2003). This has been suggested for Hh signaling during epidermal embryonic segmentation (Fig. 5). As described above, Hh is expressed in two rows of cells abutting the parasegmental boundary in the embryonic epidermis and from there (i) activates Wg in the cells adjacent in the anterior direction and (ii) activates Rho in three cell rows towards the posterior direction (Fig. 5A,B). Targeting of Hh to an apical compartment in the receiving cells (Fig. 5A,C) depends on cholesterol modification and the function of the sterol-sensing domain protein Dispatched (disp), which has being implicated previously in Hh release from the producing cells (Burke et al., 1999; Ma et al., 2002). In non-cholesterol-modified hh or in disp mutants, apical targeting of Hh in the receiving cells is absent. These mutant conditions abolish Hh signaling in the anterior adjacent cells, but do not affect signaling towards the posterior cells, which thereby activate Rho normally (Gallet et al., 2003). Similarly, Tout-velu mutants affect anterior Wg expression, but not posterior signaling. Tout-velu promotes the spreading of cholesterol-modified Hh through the extracellular space by catalyzing the synthesis of a proteoglycan that mediates Hh movement. In summary, anterior signaling requires Hh targeting to the apical endosome to be properly released from the producing cells in a process which requires Cholesterol modification of Hh, its release in Disp-dependent manner and the proteoglycan-dependent spreading of cholesterolmodified Hh mediated by Tout-velu (Fig. 5C). In contrast, posterior signaling does not require Hh modification with cholesterol, its release or proteoglycan-mediated movement (Fig. 5C). This uncovers a scenario in which different cellular environments are able to elicit the expression of different target genes depending on the lipid modification and thereby the subcellular compartment (i.e. the apical endosomal compartment where Hh is normally found) in which the Hh ligand is present. It is, however, unclear how a Hh molecule which has not been cholesterol-modified, that cannot be released properly (in disp mutants) or cannot spread through a long-range (in Tout-velu mutants) is nevertheless able to elicit signaling across 3 – 4 cell diameters in the posterior direction. Conversely, it is surprising that all these cellular requirements, which are essential for long-range spreading of the ligand, are essential to signal to the neighbour anterior cell. These paradoxes reflect that probably we are missing some essential cellular/molecular process at the basis of asymmetric Hh signaling during the process of segmentation.

4. Endocytic trafficking during lateral inhibition: Notch signaling and asymmetric cell division Dpp, Hh and Wg form gradients of concentration that elicits signaling in cells at a distance from the source.

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

1277

Fig. 5. Cholesterol and asymmetric Hh signaling? (A,B) Hh signaling (blue) activates Wg (red) in the anterior adjacent cells and Rho (green) within a range of 3 cell diameters towards the posterior. In receiving cells, Hh is internalized into ‘large punctate structures’ (LPS) probably corresponding to an apical endosomal structure. (A) Z section through the epithelium. (B) XY section. (C) Hh is secreted from the source (blue nucleus) by a process that requires Disp. It is internalized into an LPS in a process that requires Ttv. Anterior signaling to activate Wg (red) requires Disp, Ttv and the cholesterol moiety (Hh-Np). Posterior signaling to activate Rho (green) is independent of Hh release, Disp, Ttv and the cholesterol moiety.

During asymmetric cell division (ACD), the Notch signaling pathway mediates lateral inhibition between two adjacent cells contacting each other. Asymmetric Notch signaling probably capitalizes on asymmetric endocytic trafficking. 4.1. Asymmetric localization of Numb controls Notch signaling during ACD The external sense organs (ES) of the PNS are composed of two outer cells (hair and socket), two inner cells (neuron and sheath) and a glial cell (Fig. 6A) (Gho

et al., 1999). These four cells arise from a single sensory organ precursor (SOP) cell in a stereotyped lineage: the SOP divides into a posterior pIIa and an anterior pIIb cell; pIIb generates the pIIIb and a glial cell, pIIa the hair and the socket and pIIIb the neuron and the sheath (Fig. 6B) (Gho et al., 1999). Numb, a phosphotyrosine binding protein, determines the fate of the SOP progeny (Rhyu et al., 1994; Uemura et al., 1989): in loss of function numb mutants SOPs divide symmetrically into two pIIa cells, which subsequently generate two sockets (Fig. 6C). Conversely, Numb overexpression in the SOP lead to the generation of four neurons.

1278

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

Fig. 6. Endocytosis and asymmetric cell division. (A) Schematic representation of an ES organ. (B) Schematic representation of the asymmetric cell divisions that give rise to the five cells in the ES organ. (C,D) numb2/a-adaptinear (C) and Nts (D) mutant phenotypes. (E) Molecular interactions between Notch, Numb, a-adaptin and Clathrin. (F) Notch binding of Numb recruits a-adaptin/Clathrin thereby initiating its internalization by endocytosis. (G) A model for the role of endocytosis during asymmetric cell division. Numb distributed in a crescent in the SOP causes the asymmetric distribution of a-adaptin. Higher levels of Numb/a-adaptin in one cell causes the enhanced down-regulation of Notch by endocytosis in that cell. This gives advantage to one of the two cells during mutual competition for Delta/Notch signaling. This initial bias is increased until it is converted into a directional Notch signaling from one of the two cells, that inhibits the PIIa fate in the adjacent cell.

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

Interestingly, in the SOPs, the Numb protein concentrates in the area of the cortex that overlies one of the two centrosomes (Knoblich et al., 1995) and is segregated into one of the two daughter cells upon cytokinesis (Fig. 6G) (Rhyu et al., 1994). Therefore, Numb behaves as a segregating determinant in the SOP lineage. Numb controls ACD by repressing Notch signaling. Notch is the receptor of two ligands, Delta and Serrate, which upon binding lead to Notch cleavage. The intracellular Notch fragment is then imported into the nucleus and converts Suppressor of Hairless, a repressor, into a transcriptional activator. Inactivation of Notch function during ACD lead to cell fate transformations which are the opposite to those observed in numb mutants: SOP divides into two pIIb which generate four neurons (Fig. 6D) (Hartenstein and Posakony, 1989). Numb acts upstream of Notch (Guo et al., 1996) and both molecules bind to each other (Guo et al., 1996). A model has been suggested (Zeng et al., 1998) in which both Delta and Serrate bind to Notch in the SOP lineage: Notch signaling induces pIIa fate in one daughter cell, while Numb prevents Notch signal transduction in the other, which thereby adopts the pIIb fate. How does Numb control Notch signal transduction? A mammalian Numb homolog localizes to endocytic vesicles and binds to a-Adaptin, a component of the AP2 complex (Santolini et al., 2000). AP2 initiates Clathrin coating in endocytic vesicles by recruiting Clathrin to the endocytic bud and triggering Clathrin polymerization into a lattice. That prompted the possibility that Numb controls Notch signaling by orchestrating the endocytosis of the receptor. It was already known that Dynamin plays a key role during Notch signaling (Seugnet et al., 1997). 4.2. Endocytosis, Notch signaling and ACD Another endocytic factor, a-Adaptin, plays a specific role during ACD. a-Adaptin mutants were recovered in a screen for factors involved in ACD (Berdnik et al., 2002). a-Adaptin is composed of a head domain, an ear domain and a hinge domain. While loss of function mutations in a-Adaptin blocks endocytosis and cause embryonic lethality (Gonza´lez-Gaita´n and Ja¨ckle, 1997), mutations in the ear domain of a-Adaptin cause a milder phenotype and resemble Numb loss of function mutations during ES development (Fig. 6C). The ear domain of a-Adaptin has a regulatory role during endocytosis by binding to accessory endocytic proteins including AP-180, Epsin, EPS-15, Dynamin and Amphiphysin. It was also shown to bind to mammalian Numb (Santolini et al., 2000). Furthermore, it was shown that the ear domain of Drosophila a-Adaptin is able to bind Numb both in vitro and in vivo (Berdnik et al., 2002). The Numb binding domain contains a DPF domain that was shown to be essential for Notch repression (Frise et al., 1996). In the asymmetrically dividing SOPs, a-Adaptin, like Numb, is concentrated to one side of the cell cortex

1279

during metaphase (Fig. 6G) (Berdnik et al., 2002). During cell fate specification in the Drosophila ES lineage, aAdaptin acts downstream of Numb and upstream of Notch: in numb mutants a-Adaptin fails to localize asymmetrically. Since Numb can directly interact with Notch intracellular domain, Numb may antagonize Notch activity by recruiting a-Adaptin (Fig. 6E). This would in turn initiate Notch internalization by Clathrin-mediated endocytosis (Fig. 6F) and ultimately cause down-regulation of Notch signaling in the cells that get the higher levels of Numb/a-Adaptin (Fig. 6G). Testing this hypothesis will require Notch internalization assays in vivo during the development of ES.

5. Concluding remarks Different signaling events capitalize on endocytic trafficking in different ways: endocytic trafficking powers the long-range spreading of Dpp, but is also used to restrict the range of Wg signaling in the embryo; it is used to implement Hh signaling by degrading the repressing receptor, Ptc and trafficking it away from the tonic activator Smo and, during Notch signaling in ACD, to have a biased internalization of the receptor, thereby skewing a reciprocal signaling event. These proposed models of action for endocytosis are qualitative, while their basis is the quantitative change in the rates of internalization, degradation and recycling. To fully understand the mechanisms, we will need to examine these processes in quantitative detail. Therefore, future directions will include the quantitative assessment of the rates of trafficking and signaling. In Drosophila, tools like the thermosensitive shibire mutant have provided the possibility of studying phenotypes in real time, a key tool when monitoring trafficking. The analysis of other kinds of mutants only allows monitoring of the effects of the mutations after the developing system reaches its steady-state. This means that many of these phenotypes could be indirect, the consequence of the reaction of a system which probably has backups built in. In the future, a useful tool to help us understand the trafficking mechanisms during signaling would be conditional mutant situations that allow us to study the direct cell biological effects of depleting a gene function in real time.

Acknowledgements I would like to thank Karen Echeverri, Christian Bo¨kel and Alfonso Martı´nez-Arias for critically reading this manuscript. My work is supported by the Max-Planck Society and DFG.

1280

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

References Alexandre, C., Lecourtois, M., Vincent, J., 1999. Wingless and Hedgehog pattern Drosophila denticle belts by regulating the production of shortrange signals. Development 126, 5689–5698. Banerjee, U., Renfranz, P.J., Hinton, D.R., Rabin, B.A., Benzer, S., 1987. The sevenless þ protein is expressed apically in cell membranes of developing Drosophila retina; it is not restricted to cell R7. Cell 51, 151– 158. Beachy, P.A., Cooper, M.K., Young, K.E., von Kessler, D.P., Park, W.J., Hall, T.M., et al., 1997. Multiple roles of cholesterol in hedgehog protein biogenesis and signaling. Cold Spring Harb Symp Quant Biol 62, 191 –204. Bejsovec, A., Wieschaus, E., 1995. Signaling activities of the Drosophila wingless gene are separately mutable and appear to be transduced at the cell surface. Genetics 139, 309–320. Bellaiche, Y., The, I., Perrimon, N., 1998. Tout-velu is a Drosophila homologue of the putative tumour suppressor EXT-1 and is needed for Hh diffusion. Nature 394, 85–88. Bennett, D., Alphey, L., 2002. PP1 binds Sara and negatively regulates Dpp signaling in Drosophila melanogaster. Nat. Genet. 31, 419 –423. Berdnik, D., Torok, T., Gonzalez-Gaitan, M., Knoblich, J.A., 2002. The endocytic protein alpha-Adaptin is required for numb-mediated asymmetric cell division in Drosophila. Dev. Cell 3, 221 –231. Burke, R., Basler, K., 1996. Dpp receptors are autonomously required for cell proliferation in the entire developing wing. Development 122, 2261–2269. Burke, R., Nellen, D., Bellotto, M., Hafen, E., Senti, K.A., Dickson, B.J., Basler, K., 1999. Dispatched, a novel sterol-sensing domain protein dedicated to the release of cholesterol-modified hedgehog from signaling cells. Cell 99, 803– 815. Capdevila, J., Pariente, F., Sampedro, J., Alonso, J.L., Guerrero, I., 1994. Subcellular localization of the segment polarity protein patched suggests an interaction with the wingless receptor complex in Drosophila embryos. Development 120, 987 –998. Chamoun, Z., Mann, R.K., Nellen, D., von Kessler, D.P., Bellotto, M., Beachy, P.A., Basler, K., 2001. Skinny hedgehog, an acyltransferase required for palmitoylation and activity of the hedgehog signal. Science 293, 2080–2084. Chen, Y., Struhl, G., 1996. Dual roles for Patched in sequestering and transducing Hedgehog. Cell 87, 553–563. Denef, N., Neubueser, D., Perez, L., Cohen, S.M., 2000. Hedgehog induces opposite changes in turnover and subcellular localization of patched and smoothened. Cell 102, 521– 531. Di Guglielmo, G.M., Le Roy, C., Goodfellow, A.F., Wrana, J.L., 2003. Distinct endocytic pathways regulate TGF-beta receptor signalling and turnover. Nat. Cell Biol. 5, 410–421. DiNardo, S., Sher, E., Heemskerk-Jongens, J., Kassis, J.A., O’Farrell, P.H., 1988. Two-tiered regulation of spatially patterned engrailed gene expression during Drosophila embryogenesis. Nature 332, 604 –609. Dubois, L., Lecourtois, M., Alexandre, C., Hirst, E., Vincent, J.P., 2001. Regulated endocytic routing modulates wingless signaling in Drosophila embryos. Cell 105, 613–624. Ebisawa, T., Fukuchi, M., Murakami, G., Chiba, T., Tanaka, K., Imamura, T., Miyazono, K., 2001. Smurf1 interacts with transforming growth factor-beta type I receptor through Smad7 and induces receptor degradation. J. Biol. Chem. 276, 12477–12480. Entchev, E.V., Schwabedissen, A., Gonzalez-Gaitan, M., 2000. Gradient formation of the TGF-beta homolog Dpp. Cell 103, 981 –991. Fehon, R.G., Johansen, K., Rebay, I., Artavanis-Tsakonas, S., 1991. Complex cellular and subcellular regulation of notch expression during embryonic and imaginal development of Drosophila: implications for notch function. J. Cell Biol. 113, 657 –669. Frise, E., Knoblich, J.A., Younger-Shepherd, S., Jan, L.Y., Jan, Y.N., 1996. The Drosophila Numb protein inhibits signaling of the Notch receptor

during cell –cell interaction in sensory organ lineage. Proc. Natl Acad. Sci. USA 93, 11925–11932. Gallet, A., Rodriguez, R., Ruel, L., Therond, P.P., 2003. Cholesterol modification of hedgehog is required for trafficking and movement, revealing an asymmetric cellular response to hedgehog. Dev. Cell 4, 191 –204. Gho, M., Bellaiche, Y., Schweisguth, F., 1999. Revisiting the Drosophila microchaete lineage: a novel intrinsically asymmetric cell division generates a glial cell. Development 126, 3573–3584. Gillooly, D.J., Morrow, I.C., Lindsay, M., Gould, R., Bryant, N.J., Gaullier, J.M., Parton, R.G., Stenmark, H., 2000. Localization of phosphatidylinositol 3-phosphate in yeast and mammalian cells. Eur. Mol. Biol. Org. J. 19, 4577– 4588. Gillooly, D.J., Simonsen, A., Stenmark, H., 2001. Cellular functions of phosphatidylinositol 3-phosphate and FYVE domain proteins. Biochem. J. 355, 249–258. Golembo, M., Raz, E., Shilo, B.Z., 1996. The Drosophila embryonic midline is the site of Spitz processing, and induces activation of the EGF receptor in the ventral ectoderm. Development 122, 3363– 3370. Gonzalez, F., Swales, L.S., Bejsovec, A., Skaer, H., Martinez Arias, A., 1991. Secretion and movement of wingless protein in the epidermis of the Drosophila embryo. Mech. Dev. 35, 43–54. Gonza´lez-Gaita´n, M.A., Ja¨ckle, H., 1997. Role of Drosophila a-adaptin during synaptic vesicle recycling. Cell 88, 767–776. Gonza´lez-Gaita´n, M.A., Ja¨ckle, H., 1999. The range of spalt-activating Dpp signalling is reduced in endocytosis-defective Drosophila wing discs. Mech. Dev. 87, 143–151. Greco, V., Hannus, M., Eaton, S., 2001. Argosomes: a potential vehicle for the spread of morphogens through epithelia. Cell 106, 633 –645. Grindstaff, K.K., Yeaman, C., Anandasabapathy, N., Hsu, S.C., RodriguezBoulan, E., Scheller, R.H., Nelson, W.J., 1998. Sec6/8 complex is recruited to cell– cell contacts and specifies transport vesicle delivery to the basal–lateral membrane in epithelial cells. Cell 93, 731– 740. Gritzan, U., Hatini, V., DiNardo, S., 1999. Mutual antagonism between signals secreted by adjacent wingless and engrailed cells leads to specification of complementary regions of the Drosophila parasegment. Development 126, 4107–4115. Guo, M., Jan, L.Y., Jan, Y.N., 1996. Control of daughter cell fates during asymmetric division: interaction of Numb and Notch. Neuron 17, 27 –41. Hartenstein, V., Posakony, J.W., 1989. Development of adult sensilla on the wing and notum of Drosophila melanogaster. Development 107, 389 –405. Hayes, S., Chawla, A., Corvera, S., 2002. TGF beta receptor internalization into EEA1-enriched early endosomes: role in signaling to Smad2. J. Cell Biol. 158, 1239–1249. van den Heuvel, M., Nusse, R., Johnston, P., Lawrence, P.A., 1989. Distribution of the wingless gene product in Drosophila embryos: a protein involved in cell –cell communication. Cell 59, 739 –749. Incardona, J.P., Gruenberg, J., Roelink, H., 2002. Sonic hedgehog induces the segregation of patched and smoothened in endosomes. Curr. Biol. 12, 983–995. Ingham, P.W., McMahon, A.P., 2001. Hedgehog signaling in animal development: paradigms and principles. Genes Dev. 15, 3059– 3087. Ingham, P.W., Baker, N.E., Martinez-Arias, A., 1988. Regulation of segment polarity genes in the Drosophila blastoderm by fushi tarazu and even skipped. Nature 331, 73–75. Itoh, F., Divecha, N., Brocks, L., Oomen, L., Janssen, H., Calafat, J., et al., 2002. The FYVE domain in Smad anchor for receptor activation (SARA) is sufficient for localization of SARA in early endosomes and regulates TGF-beta/Smad signalling. Genes Cells 7, 321–331. Iwasaki, S., Tsuruoka, N., Hattori, A., Sato, M., Tsujimoto, M., Kohno, M., 1995. Distribution and characterization of specific cellular binding proteins for bone morphogenetic protein-2. J. Biol. Chem. 270, 5476–5482. Kavsak, P., Rasmussen, R.K., Causing, C.G., Bonni, S., Zhu, H., Thomsen, G.H., Wrana, J.L., 2000. Smad7 binds to Smurf2 to form an E3

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282 ubiquitin ligase that targets the TGF beta receptor for degradation. Mol. Cell 6, 1365–1375. Knoblich, J.A., Jan, L.Y., Jan, Y.N., 1995. Asymmetric segregation of Numb and Prospero during cell division. Nature 377, 624– 627. Kuwabara, P.E., Labouesse, M., 2002. The sterol-sensing domain: multiple families, a unique role? Trends Genet. 18, 193 –201. Lander, A.D., Nie, Q., Wan, F.Y., 2002. Do morphogen gradients arise by diffusion? Dev. Cell 2, 785–796. Lanzetti, L., Rybin, V., Malabarba, M.G., Christoforidis, S., Scita, G., Zerial, M., Di Fiore, P.P., 2000. The Eps8 protein coordinates EGF receptor signalling through Rac and trafficking through Rab5. Nature 408, 374 –377. Lauffenburger, D.A., Linderman, J.J., 1993. Receptors. Models for Binding, Trafficking and Signaling, Oxford University Press, New York. Lecuit, T., Cohen, S.M., 1998. Dpp receptor levels contribute to shaping the Dpp morphogen gradient in the Drosophila wing imaginal disc. Development 125, 4901–4907. Lecuit, T., Brook, W.J., Ng, M., Calleja, M., Sun, H., Cohen, S.M., 1996. Two distinct mechanisms for long-range patterning by decapentaplegic in the Drosophila wing. Nature 381, 387– 393. Lewis, P.M., Dunn, M.P., McMahon, J.A., Logan, M., Martin, J.F., StJacques, B., McMahon, A.P., 2001. Cholesterol modification of sonic hedgehog is required for long-range signaling activity and effective modulation of signaling by Ptc1. Cell 105, 599–612. Lin, X., Liang, M., Feng, X.H., 2000. Smurf2 is a ubiquitin E3 ligase mediating proteasome-dependent degradation of Smad2 in transforming growth factor-beta signaling. J. Biol. Chem. 275, 36818–36822. Ma, Y., Erkner, A., Gong, R., Yao, S., Taipale, J., Basler, K., Beachy, P.A., 2002. Hedgehog-mediated patterning of the mammalian embryo requires transporter-like function of dispatched. Cell 111, 63–75. Martin, V., Carrillo, G., Torroja, C., Guerrero, I., 2001. The sterol-sensing domain of Patched protein seems to control Smoothened activity through Patched vesicular trafficking. Curr. Biol. 11, 601 –607. Martinez Arias, A., 2003. Wnts as morphogens? The view from the wing of Drosophila. Nat. Rev. Mol. Cell Biol. 4, 321–325. Massague´, J., 1998. TGF-b signal transduction. Annu. Rev. Biochem. 67, 753–791. Massague, J., Kelly, B., 1986. Internalization of transforming growth factor-beta and its receptor in BALB/c 3T3 fibroblasts. J. Cell Physiol. 128, 216 –222. Miyazono, K., 2000. TGF-beta signaling by Smad proteins. Cytokine Growth Factor Rev. 11, 15– 22. Moline, M.M., Southern, C., Bejsovec, A., 1999. Directionality of Wingless protein transport influences epidermal patterning in the Drosophila embryo. Development 126, 4375–4384. Mukherjee, S., Ghosh, R.N., Maxfield, F.R., 1997. Endocytosis. Physiol. Rev. 77, 759–803. Natsume, T., Tomita, S., Iemura, S., Kinto, N., Yamaguchi, A., Ueno, N., 1997. Interaction between soluble type I receptor for bone morphogenetic protein and bone morphogenetic protein-4. J. Biol. Chem. 272, 11535– 11540. Nellen, N., Burke, R., Struhl, G., Basler, K., 1996. Direct and long-range action of a DPP morphogen gradient. Cell 85, 357 –368. Novick, P., Zerial, M., 1997. The diversity of Rab proteins in vesicle transport. Curr. Opin. Cell Biol. 9, 496 –504. Panopoulou, E., Gillooly, D.J., Wrana, J.L., Zerial, M., Stenmark, H., Murphy, C., Fotsis, T., 2002. Early endosomal regulation of Smaddependent signaling in endothelial cells. J. Biol. Chem. 277, 18046– 18052. Payre, F., Vincent, A., Carreno, S., 1999. ovo/svb integrates Wingless and DER pathways to control epidermis differentiation. Nature 400, 271–275. Penheiter, S.G., Mitchell, H., Garamszegi, N., Edens, M., Dore, J.J. Jr., Leof, E.B., 2002. Internalization-dependent and -independent requirements for transforming growth factor beta receptor signaling via the Smad pathway. Mol. Cell Biol. 22, 4750–4759.

1281

Pepinsky, R.B., Zeng, C., Wen, D., Rayhorn, P., Baker, D.P., Williams, K.P., et al., 1998. Identification of a palmitic acid-modified form of human Sonic hedgehog. J. Biol. Chem. 273, 14037–14045. Pfeiffer, S., Ricardo, S., Manneville, J.B., Alexandre, C., Vincent, J.P., 2002. Producing cells retain and recycle Wingless in Drosophila embryos. Curr. Biol. 12, 957–962. Podos, S.D., Hanson, K.K., Wang, Y.C., Ferguson, E.L., 2001. The DSmurf ubiquitin-protein ligase restricts BMP signaling spatially and temporally during Drosophila embryogenesis. Dev. Cell 1, 567– 578. Porter, J.A., Young, K.E., Beachy, P.A., 1996. Cholesterol modification of Hedgehog signaling proteins in animal development. Science 274, 255– 259. Rhyu, M.S., Jan, L.Y., Jan, Y.N., 1994. Asymmetric distribution of numb protein during division of the sensory organ precursor cell confers distinct fates to daughter cells. Cell 76, 477 –491. Rietveld, A., Neutz, S., Simons, K., Eaton, S., 1999. Association of steroland glycosylphosphatidylinositol-linked proteins with Drosophila raft Lipid microdomains. J. Biol. Chem. 274, 12049–12054. Sanson, B., Alexandre, C., Fascetti, N., Vincent, J.P., 1999. Engrailed and hedgehog make the range of Wingless asymmetric in Drosophila embryos. Cell 98, 207 –216. Santolini, E., Puri, C., Salcini, A.E., Gagliani, M.C., Pelicci, P.G., Tacchetti, C., Di Fiore, P.P., 2000. Numb is an endocytic protein. J. Cell Biol. 151, 1345–1352. Sathre, K.A., Tsang, M.L., Weatherbee, J.A., Steer, C.J., 1991. Binding and internalization of transforming growth factor-beta 1 by human hepatoma cells: evidence for receptor recycling. Hepatology 14, 287– 295. Seugnet, L., Simpson, P., Haenlin, M., 1997. Requirement for dynamin during Notch signaling in Drosophila neurogenesis. Dev. Biol. 192, 585– 598. Sheff, D.R., Daro, E.A., Hull, M., Mellman, I., 1999. The receptor recycling pathway contains two distinct populations of early endosomes with different sorting functions. J. Cell Biol. 145, 123 –139. Shitara, Y., Kato, Y., Sugiyama, Y., 1998. Effect of brefeldin A and lysosomotropic reagents on intracellular trafficking of epidermal growth factor and transferrin in Madin–Darby canine kidney epithelial cells. J. Control Release 55, 35– 43. Simons, K., Ikonen, E., 1997. Functional rafts in cell membranes. Nature 387, 569–572. Strigini, M., Cohen, S.M., 1999. Formation of morphogen gradients in the Drosophila wing. Semin. Cell Dev. Biol. 10, 335–344. Strutt, H., Thomas, C., Nakano, Y., Stark, D., Neave, B., Taylor, A.M., Ingham, P.W., 2001. Mutations in the sterol-sensing domain of Patched suggest a role for vesicular trafficking in Smoothened regulation. Curr. Biol. 11, 608– 613. Tall, G.G., Barbieri, M.A., Stahl, P.D., Horazdovsky, B.F., 2001. Rasactivated endocytosis is mediated by the Rab5 guanine nucleotide exchange activity of RIN1. Dev. Cell 1, 73– 82. The, I., Bellaiche, Y., Perrimon, N., 1999. Hedgehog movement is regulated through Tout velu-dependent synthesis of a heparan sulfate proteoglycan. Mol. Cell 4, 633 –639. Tomlinson, A., Bowtell, D.D., Hafen, E., Rubin, G.M., 1987. Localization of the sevenless protein, a putative receptor for positional information, in the eye imaginal disc of Drosophila. Cell 51, 143 –150. Tsukazaki, T., Chiang, T.A., Davidson, A.F., Attisano, L., Wrana, J.L., 1998. SARA, a FYVE domain protein that recruits Smad2 to the TGFbeta receptor. Cell 95, 779–791. Turing, A.M., 1952. The chemical basis of morphogenesis. Philos. Trans. R. Soc. (London) 237, 37–72. Uemura, T., Shepherd, S., Ackerman, L., Jan, L.Y., Jan, Y.N., 1989. Numb, a gene required in determination of cell fate during sensory organ formation in Drosophila embryos. Cell 58, 349 –360. Wiellette, E.L., McGinnis, W., 1999. Hox genes differentially regulate Serrate to generate segment-specific structures. Development 126, 1985–1995.

1282

M. Gonza´lez-Gaita´n / Mechanisms of Development 120 (2003) 1265–1282

Willert, K., Brown, J.D., Danenberg, E., Duncan, A.W., Weissman, I.L., Reya, T., et al., 2003. Wnt proteins are lipid-modified and can act as stem cell growth factors. Nature 423, 448–452. Wolpert, L., 1969. Positional information and the spatial pattern of cellular differentiation. J. Theor. Biol. 25, 1 –47. Wrana, J.L., Attisano, L., Wieser, R., Ventura, F., Massague, J., 1994. Mechanism of activation of the TGF-beta receptor. Nature 370, 341–347. Wurmser, A.E., Gary, J.D., Emr, S.D., 1999. Phosphoinositide 3-kinases and their FYVE domain-containing effectors as regulators of vacuolar/

lysosomal membrane trafficking pathways. J. Biol. Chem. 274, 9129–9132. Zeng, C., Younger-Shepherd, S., Jan, L.Y., Jan, Y.N., 1998. Delta and Serrate are redundant Notch ligands required for asymmetric cell divisions within the Drosophila sensory organ lineage. Genes Dev. 12, 1086–1091. Zhu, H., Kavsak, P., Abdollah, S., Wrana, J.L., Thomsen, G.H., 1999. A SMAD ubiquitin ligase targets the BMP pathway and affects embryonic pattern formation. Nature 400, 687 –693.