Developmental Cell, Vol. 5, 139–148, July, 2003, Copyright 2003 by Cell Press

Unequal Segregation of Neuralized Biases Notch Activation during Asymmetric Cell Division Roland Le Borgne and Franc¸ois Schweisguth* Departement de Biologie Ecole Normale Supe´rieure CNRS UMR 8542 46, rue d’Ulm 75230 Paris Cedex France

Summary In Drosophila, Notch signaling regulates binary fate decisions at each asymmetric division in sensory organ lineages. Following division of the sensory organ precursor cell (pI), Notch is activated in one daughter cell (pIIa) and inhibited in the other (pIIb). We report that the E3 ubiquitin ligase Neuralized localizes asymmetrically in the dividing pI cell and unequally segregates into the pIIb cell, like the Notch inhibitor Numb. Furthermore, Neuralized upregulates endocytosis of the Notch ligand Delta in the pIIb cell and acts in the pIIb cell to promote activation of Notch in the pIIa cell. Thus, Neuralized is a conserved regulator of Notch signaling that acts as a cell fate determinant. Polarization of the pI cell directs the unequal segregation of both Neuralized and Numb. We propose that coordinated upregulation of ligand activity by Neuralized and inhibition of receptor activity by Numb results in a robust bias in Notch signaling. Introduction During metazoan development, different cell fates are generated by cell-cell interactions or by the unequal segregation of cell fate determinants during asymmetric cell divisions. Both mechanisms contribute to the pIIa/ pIIb decision during sensory organ development in Drosophila. In the pupal thorax, each sensory organ precursor cell (pI) divides along the anterior-posterior axis of the fly to generate a posterior pIIa cell and an anterior pIIb cell (Gho et al., 1999). The cell fate determinant Numb localizes asymmetrically at the anterior cortex of the dividing pI cell and segregates into the anterior daughter cell (Rhyu et al., 1994). Genetic analysis indicates that Numb antagonizes Notch (N) signaling and that Numb-mediated inhibition of N requires the ␣-adaptin (Berdnik et al., 2002; Guo et al., 1996). Numb binds both the intracellular domain of N (Guo et al., 1996) and the ␣-adaptin ear domain (Berdnik et al., 2002; Santolini et al., 2000). This suggests that Numb may directly regulate N endocytosis. Signaling by N receptors involves three successive cleavages (for review, see Fortini, 2001). N is first processed in the trans-Golgi network at the extracellular S1 site to produce a functional heterodimeric receptor. A second ligand-dependent cleavage of N at the extracellular S2 site generates *Correspondence: [email protected]

a membrane-bound activated form which is then processed at an intramembraneous S3 site, thereby releasing the active Notch intracellular domain (NICD). Numb genetically acts upstream of S3 cleavage (Guo et al., 1996). Thus, Numb may promote internalization and degradation of N prior to S3 cleavage. Alternatively, the importance of dynamin activity for N signal transduction (Seugnet et al., 1997) raises the possibility that S3 cleavage depends on dynamin-dependent endocytosis with Numb inhibiting endocytosis of membrane-bound activated N. Regardless of the mechanism of Numb action, trafficking of N is suggested to play an important role in generating asymmetry. Two ligands of N are known in Drosophila, Delta (Dl), and Serrate (Ser). These two ligands act redundantly to activate N during the pIIa/pIIb decision (Zeng et al., 1998). Recent studies have indicated that endocytosis of Dl is critical for N activation. First, dynamin-dependent endocytosis is not only required for signal transduction as mentioned above but is also required in signal-sending cells to promote N activation (Seugnet et al., 1997). Second, endocytosis-defective Dl proteins have reduced signaling capacity (Parks et al., 2000). Third, the E3-ubiquitin ligases Neuralized (Neur) in Drosophila and Mind bomb (Mib) in zebrafish promote endocytosis of Dl and appear to be required for efficient activation of N by Dl (Deblandre et al., 2001; Itoh et al., 2003; Lai et al., 2001; Pavlopoulos et al., 2001; Yeh et al., 2001). It has been proposed that Dl endocytosis facilitates the S2 cleavage of N at the surface of the signal-receiving cell (Parks et al., 2000). Here we show that Neur is unequally segregated during asymmetric division of the pI cell, upregulates endocytosis of Dl in the pIIb cell, and plays a critical role in generating cell fate diversity. We propose that Neur acts as a cell fate determinant during asymmetric cell divisions. Results A numb-Independent Asymmetry in Delta Endocytosis To examine whether asymmetry in N ligands distribution may play a role in generating cell fate diversity during asymmetric divisions, we have analyzed the subcellular distribution of Dl and Ser in the sensory organ lineage. In mitotic pI cells, Dl and Ser were uniformly distributed around the cell cortex and were equally partitioned into both daughter cells (data not shown). In both pI daughter cells, Dl and Ser accumulated at the apical cell cortex as well as in intracellular dots of 0.5 ⫾ 0.2 ␮m in diameter (Figures 1A–1A″ and E–E″; see Supplementary Figure S1 at http://www.developmentalcell.com/cgi/content/ full/5/1/139/DC1). These dots were coated by Hrs (hepatocyte growth factor-regulated tyrosine kinase substrate) (Figures 1C–1D″). Hrs binds ubiquitinated proteins via its ubiquitin-interacting motif and sorts endocytic cargos into the lumen of multivesicular bodies (MVBs) (Lloyd et al., 2002; Raiborg et al., 2002; Shih et al., 2002). Therefore, these Dl-positive vesicles appeared to

Developmental Cell 140

Figure 1. A numb-Independent Asymmetry of Dl Localization (A–B″) Dl ([A, A⬘, B, and B’]; red in [A and B]) colocalized with NECD ([A, A″, B, and B″]; blue dots in [A and B]) into large intracellular vesicles in pIIa, pIIb, and epidermal cells, both in wild-type (A–A″) and in numb15 (B–B″) cells. Sensory cells were identified by using Cut ([A and A″]; blue nuclear staining in [A]), Pon-GFP (green in [A]; Pon-GFP was expressed under the control of neurP72GAL4), or Asense (green in [B]) as markers. A higher number of Dl- and NECD-containing vesicles was detected in anterior pIIb cells in both wild-type (A–A″) and numb15 mutant pupae (B–B″). (C–D”) Dl ([C, D, and D″]; blue in [C and D]) localized within Hrs-coated vesicles ([C–D⬘]; red in [C and D]) in pIIb (C–D″), pIIa, and epidermal cells (data not shown). Sensory cells were identified by using Pon-GFP (green in [C and D]; Pon-GFP was expressed under the control of neurP72GAL4). (E–E″) Ser ([E and E″]; green in [E]) colocalized with Dl ([E and E⬘]; red in [E]) within Hrscoated vesicles (data not shown). Bar is 5 ␮m in (A)–(B″) and 1 ␮m in (D)–(E″). Unless specified, anterior is left in all figures.

be large endocytic vesicles that probably correspond to MVBs. These Dl-positive vesicles also contained Notch extracellular domain (NECD) and NICD epitopes (Figures 1A–1A″ and data not shown). Strikingly, a higher number of large Dl-positive vesicles was seen in the anterior signal-sending pIIb cell (5.0 ⫾ 2.2, n ⫽ 130) than in the posterior signal-receiving pIIa cell (2.0 ⫾ 1.5, n ⫽ 130; Figure 3C). This difference was significant (p ⬍ 0.001; Student’s t test). This asymmetry in Dl endocytosis is established independently of the unequal partitioning of Numb. Indeed, anterior pI daughter cells are shown to accumulate a higher number of Dl-positive vesicles than posterior pI daughter cells in numb2 and numb15 mutant clones (Figures 1B–1B″; numb2: 5.1 ⫾ 1.8 versus 2.8 ⫾ 1.6, n ⫽ 24, p ⬍ 0.001; numb15: 5.7 ⫾ 2.5 versus 3.1 ⫾ 1.9, n ⫽ 16, p ⬍ 0.001). Thus, asymmetry in Dl endocytosis did not depend on Numb. Neuralized Upregulates the Endocytosis of Delta in the pIIb Cell Recent studies have suggested that endocytosis of Dl is promoted by the ubiquitination of Dl by Neur, a RING finger-type E3-ubiquitin ligase required for N signaling (Deblandre et al., 2001; Lai et al., 2001; Pavlopoulos et al., 2001; Yeh et al., 2001). Neur is found in a complex with Dl and is required for Dl ubiquitination. Finally, Neur stimulates the accumulation of Dl into intracellular vesicles in imaginal disc cells. The latter conclusion was, however, based on the analysis of steady-state levels of Dl, making it difficult to unambiguously conclude whether Neur promotes Dl endocytosis or favors direct sorting from the Golgi to intracellular vesicles. To discriminate between these two possibilities and to test whether Neur regulates Dl trafficking in sensory cells, we developed an ex vivo assay for endocytosis. Internalization of Dl was followed in living epithelial cells using antibodies recognizing the extracellular part of Dl. Briefly, the single-layered epithelium corresponding to

the pupal notum was dissected and cultured in presence of anti-Dl antibodies (Figure 2A). Following medium changes and fixation, the uptake of anti-Dl antibodies was revealed using secondary antibodies. Anti-Dl antibodies were specifically internalized in the pIIa and pIIb cells (Figures 2B and B⬘). Internalized anti-Dl antibodies colocalized with Dl into large Dl-positive vesicles (Figures 2B–2B″). Internalization of anti-Dl required dynamin activity (Figure 2C) and was not observed at 4⬚C (data not shown). Together, these results indicate that antiDl interacts with Dl at the cell surface and that Dl-antiDl complexes are endocytosed in sensory cells. We then used this assay to examine the function of neur. Clones of neur1F65 mutant cells have been shown to exhibit a neurogenic phenotype with too many pI cells being specified. The progeny of these mutant pI cells produced no external sensory structures indicating that pIIa cells have been transformed into pIIb-like cells (Lai and Rubin, 2001a; Yeh et al., 2000). These cell fate transformations are associated with defects in Dl trafficking. High levels of anti-Dl remain at the surface of neur1F65 mutant cells and internalization of anti-Dl was drastically reduced (Figures 2D and 2D⬘). We conclude that neur is required for the endocytosis of Dl in sensory cells. This defect in Dl endocytosis was quantified on fixed tissues. neur mutant pI cells and pIIb-like progeny cells were found to accumulate high levels of Dl at the cell surface (Figures 3A–3B″). Accumulation of Dl at the cell surface is consistent with the proposed function of Neur in the internalization and degradation of Dl (Deblandre et al., 2001; Lai et al., 2001; Pavlopoulos et al., 2001). Quantification of Dl-positive vesicles in neur mutant clones revealed that mutant pIIb-like cells contained much fewer Dl-positive vesicles (1.5 ⫾ 0.9, n ⫽ 135; Figure 3C) than wild-type pIIb cells (5.0 ⫾ 2.2, n ⫽ 130). Thus, in the absence of neur function, both pI daughter cells have the same reduced number of Dl-positive vesicles. Furthermore, a similar distribution of Dl-containing

Unequal Segregation of Neuralized 141

Figure 2. neur-Dependent Endocytosis of Dl in Living Epithelial Cells (A) Schematic representation of the endocytosis assay. Pupal nota were dissected in culture medium (1) and incubated with a mouse antiDl recognizing the extracellular domain of Dl (2). Dl-anti-Dl complexes were allowed to be internalized for 15 min (3) prior to washing, fixation, and immunodetection of internalized antibodies. (B–B″) Mouse anti-Dl ([B and B⬘], red in [B]) were specifically internalized in the pIIa and pIIb cells (Pon-GFP expressed under the control of neurP72GAL4, green in [B]). Internalized anti-Dl localized into large Dl-positive vesicles (red arrowheads) detected using guinea pig anti-Dl ([B and B″]; blue in [B]). Dl is not internalized at a detectable level in epidermal cells (blue arrowheads in [B⬘]). (C) Uptake of anti-Dl requires Dynamin activity. Anti-Dl (red) remained at the cell surface of the pIIa/pIIb cells (Asense in blue) in epithelia mutant for the temperature sensitive dynamin mutant shits cultured at restrictive temperature (30⬚C). Bar is 5 ␮m in (B)–(C). (D and D⬘) Neur regulates endocytosis of Dl. Anti-Dl antibodies were efficiently internalized in wild-type pIIa and pIIb cells (white arrowheads in [D and D⬘]; Asense in blue). In contrast, anti-Dl uptake was inhibited in neur1F65 mutant sensory cells (marked by the loss of nlsGFP) and high levels of anti-Dl were detected at the surface of neur mutant sensory cells. Bar is 20 ␮m.

vesicles was seen in the wild-type pIIa cells (2.0 ⫾ 1.5, n ⫽ 130; Figures 1A–1A″) that do not inherit Neur and in the neur mutant pIIb-like cells (1.5 ⫾ 0.9, n ⫽ 135; Figures 3B and 3C). These comparisons indicate that neur is required to upregulate the endocytosis of Dl in the pIIb cell. Neuralized Is Unequally Segregated into the pIIb Cell Upregulation of Dl endocytosis in the pIIb cell may result from higher levels of Neur in this cell. To test this hypothesis, we examined the localization of Neur. The Neur protein was detectable in the pI cell and in its progeny cells, but not in epidermal cells. Neur was perinuclear in prophase and localized asymmetrically at the anterior cortex during prometaphase (Figures 4A–4B″). At telophase, Neur specifically segregated into the anterior daughter cell (Figures 4C–4C″) At cytokinesis, Neur uniformly redistributed at the cortex and in the cytoplasm in the pIIb cell (Figures 4D–4D″). Localization of Neur at mitosis is identical to the one described for Partner of Numb (Pon) (Bellaı¨che et al., 2001a; Lu et al., 1998). Consistently, Neur colocalized with Pon-GFP throughout mitosis (Figures 4A–4D″). Asymmetric localization

of Neur was also seen in the pIIb and pIIa dividing cells (see Supplementary Figure S2 at http://www. developmentalcell.com/cgi/content/full/5/1/139/DC1). Specificity of anti-Neur antibodies was demonstrated by absence of staining in neur mutant pI cells (Figures 4E–4F⬘). Unequal segregation of Neur did not depend on numb activity (Figures 4G–4I). Conversely, unequal segregation of Numb did not depend on neur activity (Figures 4J–4L⬘). Thus, the numb-independent unequal segregation of Neur into the pIIb cell provides a simple explanation for the upregulation of Dl endocytosis in the pIIb cell. To test the functional significance of Neur unequal segregation, Neur was overexpressed in pI cells. Overexpression of Neur using neurP72GAL4 (Bellaı¨che et al., 2001a) failed to affect the unequal partitioning of Neur at pI mitosis (see Supplementary Figure S3A–S3C at http://www.developmentalcell.com/cgi/content/full/5/ 1/139/DC1) and the pIIa/pIIb decision (data not shown) but resulted in a weak double-socket phenotype associated with a shaft-to-socket transformation (Supplementary Figures S3H–S3K at http://www.developmentalcell. com/cgi/content/full/5/1/139/DC1). This fate transformation is known to result from high levels of Delta-Notch

Developmental Cell 142

Figure 3. neur Upregulates Dl Endocytosis in the pIIb Cell (A–B″) Localization of Dl (red) and NECD (blue vesicular staining) was analyzed in neur1F65 mutant cells (marked by loss of nlsGFP in green) at a stage when pairs of pIIa/pIIb cells (nuclear Cut in blue) are seen in surrounding wild-type tissue. (B)–(B⬘) are enlargments of the region boxed in (A⬘). Loss of neur activity resulted in a dramatic increase in Dl accumulation at the cell surface (compare the wild-type and neur mutant sensory cells indicated by arrowheads in [A″]) and in a reduction in the number of Dl-positive (arrowheads in [B and B″]) and NECDpositive (B⬘) vesicles. In both wild-type and neur mutant cells, most Dl-positive vesicles were coated by Hrs (data not shown). A low level of neur-independent trafficking of Dl toward Hrs-positive vesicles was expected since epidermal cells, that did not express neur, also contained a few Hrs-positive Dl-containing vesicles (see Figure 1 and data not shown). Loss of neur function also led to the specification of a large excess of pI cells. Bar is 5 ␮m in (B)–(B″). (C) Plot showing the distribution of the number of Dl-positive vesicles in wild-type pIIa (blue bars) and pIIb (red bars) cells as well as in neur1F65 mutant pIIb-like (green bars) cells. Counting of Dl-positive vesicles in neur pIIb-like cells was performed on confocal z-sections only within rows containing both dividing pI cells and pIIa/pIIb pairs outside the clone. The mutant pIIb-like cells were distinguished from pI cells based on their small nuclei.

signaling (Bang and Posakony, 1992; Schweisguth and Posakony, 1994) and is opposite to the socket-to-shaft transformation seen in neur mutant clones (see below). Moreover, this shaft-to-socket transformation may result from the equal partitioning of Neur (but not Numb) in the two pIIa daughter cells which can also be observed at low frequency (Supplementary Figures S3G–S3G″ at http://www.developmentalcell.com/cgi/content/full/ 5/1/139/DC1). Thus, these observations support the notion that unequal segregation of Neur is functionally important.

A Similar Mode of Asymmetric Distribution for Neuralized and Numb We next investigated the mechanisms by which Neur localized at the anterior cortex of the dividing pI cell. The role of the cytoskeleton was studied by applying drugs to cultured nota. Colcemid, a microtubule-depolymerizing agent, was found to have no significant effect (Figures 5A and 5D). In contrast, both Latrunculin A, an agent that depolymerizes actin microfilaments, and the myosin motor inhibitor butanedione-2-monoxime (BDM) strongly impaired (Figures 5B and 5C) or completely

Unequal Segregation of Neuralized 143

Figure 4. Unequal Segregation of Neur (A–D″) Localization of Neur (red) and PonGFP (green) was studied in mitotic pI cells. Sensory cells were detected using Pon-GFP expressed under the control of neur P72GAL4. At prophase (A–A″), Neur colocalized with Pon-GFP in a perinuclear compartment and in a weak cortical crescent at the anterior pole. At prometaphase, Neur colocalized with Pon-GFP at the anterior cortex (B–B″) and unequally segregated into the anterior pIIb daughter cell (C–C″). At cytokinesis, Neur uniformly redistributed at the cortex and in the cytoplasm in the pIIb cell (D–D″). (E–F⬘) No anti-Neur immunoreactivity (red) was detected in mitotic neur1F65 mutant pI cells (Cut in blue; nlsGFP in green served as a clone marker). An enlarged view of the inset in (E) is shown in (F)–(F⬘). (G–I) Neur (red) localized asymmetrically in mitotic numb2 mutant pI cells (Cut in blue; nlsGFP in green). The localization of Neur in wild-type cells (I) was compared to that in numb mutant cells (H). Anterior is up in (G)–(I). (J–L⬘) Numb (blue) localized asymmetrically in mitotic neur1F65 mutant pI cells (nlsGFP in green). Note the strong and uniform cortical accumulation of Dl (red) in dividing neur1F65 mutant pI cells ([K]; inset in [J] corresponds to [K]). A wild-type control is shown in (L)–(L⬘). Bar is 25 ␮m in (E) and (J), 10 ␮m in (F), and 5 ␮m in all other panels.

inhibited (data not shown) the asymmetric localization of Neur. Thus, both myosin motor activity and an intact actin cytoskeleton are required for the formation and/ or maintenance of the Neur crescent at the anterior cortex of the dividing pI cell. These requirements for Neur localization are similar to the ones seen earlier for Numb (Berdnik and Knoblich, 2002; Knoblich et al., 1997) and Pon (Lu et al., 1999). Neur also behaves similarly to Numb and Pon in that localization of Neur at the anterior cortex of the pI cell depends on planar polarity genes (data not shown) and on the polarity genes discs-large and pins (Figures 5E–5G) (Bellaı¨che et al., 2001b). Moreover, mispartitioning of Neur in dlg and pins mutant cells correlated with a loss in asymmetric internalization of Dl (Figures 5H–5I⬘). These data indicate that Neur and Numb share part of the same molecular machinery to localize asymmetrically in the pI cell.

neuralized Activity Is Required in Signal-Sending Cells Unequal segregation of Neur in the anterior pIIb cell suggests that Neur acts in this cell to promote adoption of the pIIa fate by the posterior cell. To test whether neur activity is indeed required in the pIIb cell, we generated clones within the sensory organ lineage. Mitotic recombination in the pI cell produces one neur mutant cell and one wild-type cell (Figure 6A). Importantly, the anterior daughter cell will inherit Neur, regardless of its genotype. Thus, when the anterior cell is neur mutant, the posterior cell is predicted to adopt a pIIa fate whatever the requirement for neur activity. However, two different outcomes are predicted when the posterior cell is mutant. If neur activity is required in the signal-receiving cell, the posterior cell is predicted to adopt a pIIb-like fate activity (Figure 6A). This should result in a bristle loss phenotype.

Developmental Cell 144

Figure 5. Requirements for Neur Asymmetric Localization (A–D) Asymmetric localization of Neur (red) was examined in dividing pI cells (Cut in blue in the small insets) from dissected nota cultured in control medium (A) or in the presence of 2 ␮M Latrunculin A (B), 5 mM BDM (C), or 1 ␮M colcemid (D). Depolymerization of microtubules did not affect Neur localization (n ⫽ 39). In contrast, depolymerization of microfilaments by Latrunculin A and inhibition of myosin motor activity by BDM strongly impaired ([B], n ⫽ 4/17; [C], n ⫽ 5/19) or completely abolished Neur crescent formation (data not shown; Latrunculin A, n ⫽ 13/17 and BDM, n ⫽ 14/19). Pon-GFP behaved similarly to Neur in these assays (data not shown). (E–G) Neur (red) was mostly cytoplasmic in dlg1P20 ([F], n ⫽ 18) and pins⌬62 ([G], n ⫽ 20) mutant pI cells (Cut in blue). A wild-type control is shown in (E). (H–I⬘) Similar numbers of Dl- (red) and NECDcontaining (green; arrowheads in [H⬘ and I⬘]) vesicles were seen in both pI daughter cells of dlg1P20 ([H and H⬘]; 2.4 ⫾ 1.7 and 2.7 ⫾ 1.6 vesicles in the anterior and posterior pI daughter cells, respectively; n ⫽ 19) and pins⌬62 ([I and I⬘]; 3.2 ⫾ 1.3 and 3.0 ⫾ 1.5 vesicles in the anterior and posterior pI daughter cells, respectively; n ⫽ 32) mutant pupae. Sensory cells were detected using Cut (green nuclear staining). TRITC-phalloidin staining (shown in blue) was used to outline the cells. Bar is 5 ␮m.

In contrast, if neur acts in the signal-sending cell, the mutant posterior cell is predicted to become a pIIa cell (Figure 6A). This mutant pIIa cell should then produce two mutant cells unable to signal, hence leading to bristle duplication. Mitotic recombination induced at 0–6 hr before puparium formation (PF), when most macrochaete pI cells are specified but have not yet divided (Huang et al., 1991), produced flies with double-shaft bristles on the head, thorax and at the wing margin (Figures 6B and 6C). No macrochaete loss was detectable. This double-shaft phenotype appears to result from wild-type pIIb/mutant pIIa pairs because sensory organs composed of two mutant shaft cells and wildtype pIIb progeny cells were detected at 20 hr after PF (Figures 6D–6E⬘; n ⫽ 9). Reciprocally, a sheath-toneuron transformation was observed in mutant pIIb/ wild-type pIIa pairs (data not shown; n ⫽ 6). These data show that neur is required for the socket/shaft and neuron/sheath fate decisions and further indicate that neur acts in the pIIb cell to specify the pIIa cell. Previous studies on the cell autonomy of neur have led to contradictory results (Lai and Rubin, 2001a; Pavlopoulos et al., 2001; Yeh et al., 2000) (see Discussion). Thus, we further tested the autonomy of neur activity during the pI/epidermal fate decision in the developing notum. Analysis of cell fate decisions at the clone border has previously been shown to be extremely powerful to distinguish activities required for signal reception, like N, from activities required for signal production, like Dl (Heitzler and Simpson, 1991). Mutant cells unable to receive inhibitory signaling autonomously adopt the pI fate and inhibit their wild-type neighbors to become pI

cells. Thus, most of the pI cells at the clone border are mutant. Conversely, mutant cells unable to produce inhibitory signals fail to inhibit their wild-type neighbors but still receive inhibitory signaling produced by neighboring wild-type cells. Thus, most of the pI cells at the clone border are wild-type. We have used this assay to study the autonomy of neur. The genotype of the pI cells that contact the border of neur1F65 mutant clones running across microchaete rows 1–5 was examined (Figures 3A and A⬘). One position per microchaete row was scored. Wild-type pI cells were found in 79% of these positions (blue arrows, Figure 3A⬘, n ⫽ 73; 34 clones). Thus, neur mutant cells are not defective in receiving the inhibitory signal sent by wild-type cells but fail to efficiently inhibit wild-type cells from becoming pI cells. We conclude that neur activity is primarily required in signal-sending cells. Discussion Notch signaling regulates binary cell fate decisions in a wide variety of developmental contexts. Following asymmetric division, up- or downregulation of N signaling in one daughter cell may generate an asymmetry in fate. In this study, we identify Neur as an important factor that establishes an asymmetry in N signaling in the sensory bristle lineage. First, loss of neur activity in dividing pI and pIIa precursor cells results in pIIa-topIIb and socket-to-shaft transformations, respectively. Conversely, overexpression of Neur in the bristle lineage may result in the opposite socket-to-shaft transformation. Thus, Neur regulates the binary pIIa/pIIb and

Unequal Segregation of Neuralized 145

Figure 6. Neur Acts in the Signal-Sending pIIb Cell (A) Predicted bristle phenotypes. Mitotic recombination was induced in heterozygous pI cells (nuclear GFP in green, Neur in red). If no recombination occurs (1), the pIIa and pIIb cells are correctly specified and produce a wild-type bristle. Two different phenotypes are predicted when recombination produces a posterior cell mutant for neur (2). If neur activity is required in the signal-receiving cell (3), the posterior cell is predicted to adopt a pIIb-like fate. This mutant pIIb-like cells should then produce mutant neurons, hence resulting in a bristle loss phenotype. In contrast, if neur acts in the signal-sending cell (4), the mutant posterior cell should adopt the pIIa fate. This mutant pIIa cell should then produce two mutant cells unable to signal. These two cells are therefore predicted to produce two shafts and no socket. (B and C) Examples of double-shaft bristles with no socket (position of the missing socket indicated by a red arrowhead) on the thorax (B) and at the wing margin (C) of adult flies. (D–E⬘) Wild-type (D and D⬘) and double-shaft (E and E⬘) bristles in the notum of a 20 hr APF pupa. Two neur mutant shaft cells (E and E⬘) were associated to a wild-type neuron (data not shown), demonstrating that a mutant pIIa had been specified following recombination in the pI cell. Nuclear GFP (green) identifies wild-type cells, Cut (red) identifies sensory cells, and Su(H) (blue) identifies socket cells. (D⬘) and (E⬘) show the clone marker (nls-GFP in green).

socket/shaft decisions. Second, Neur localizes asymmetrically in the dividing pI and pIIa cells and unequally segregates into the pIIb and shaft cells, respectively. Third, endocytosis of Dl is upregulated in the pIIb cell, and this upregulation depends on neur activity. Fourth, clonal analysis indicates that Neur acts nonautonomously in the pIIb cell to promote activation of N in the pIIa cell. The mechanisms by which endocytosis of Dl in the pIIb cell may be linked to the activation of N in the pIIa cell are discussed below. Together, these results indicate that Neur acts as a cell fate determinant in the sensory bristle lineage. Thus, two distinct pathways establish an asymmetry in N signaling in this lineage. A first pathway involves inhibition of N by Numb (Berdnik et al., 2002; Guo et al., 1996; Rhyu et al., 1994), while a second pathway involves the Neur-dependent upregulation of Dl signaling. These two parallel pathways may have evolved to reinforce fidelity and robustness in generating fate asymmetry. Previous studies have shown that Neur is found in a complex with Dl and is required for Dl monoubiquitination and internalization, suggesting that Dl may be a

substrate for Neur (Deblandre et al., 2001; Lai et al., 2001; Pavlopoulos et al., 2001; Yeh et al., 2001). In yeast, monoubiquitination of surface membrane protein has been shown to serve as a signal for endocytosis (Galan et al., 1996; Polo et al., 2002). By analogy, Neur may ubiquitinate an endocytic motif of Dl, thereby regulating Dl endocytosis. We have used here an endocytosis assay to follow the internalization of Dl in living epithelial cells and have shown that neur is required for the endocytosis of Dl in sensory cells. The mechanism by which Neur regulates N signaling has remained controversial. One model proposes that Neur acts in signal-receiving cells to promote signal reception. In this model, Neur antagonizes the inhibition mediated by Dl in cis on N. In this model, Neur sorts Dl away from N (Deblandre et al., 2001; Lai et al., 2001). A second model proposes that Neur acts in signal-producing cells to upregulate Dl signaling (Pavlopoulos et al., 2001). Our analysis clearly indicates that Neur is required in signal-sending cells and, therefore, supports the latter model. First, clonal analysis in the bristle lineage has revealed that neur mutant pIIa cells produce shaft cells,

Developmental Cell 146

indicating that they have been correctly specified. Similarly, following mitotic recombination in the pI cell, double-shaft bristles produced by pairs of neur mutant cells have been independently observed by Lai and Rubin (2001a). This indicates that neur is not autonomously required for the specification of the pIIa cell. Second, our clonal analysis indicates that neur is required in the signal-sending pI cell during lateral inhibition and therefore confirms the results of Pavlopoulos et al. (2001). Our conclusion that Neur acts in signal-sending cells to upregulate Dl signaling is further strengthened by the observation that forced expression of Neur and Dl, but not Dl alone, in clones induce activation of N target genes in cells surrounding the clones (Pavlopoulos et al., 2001). Finally, this conclusion is consistent with the observed accumulation of Neur in the signalsending pI, pIIb, pIIIb, and shaft cells (Lai and Rubin, 2001b; Yeh et al., 2000) (this study). However, this conclusion contradicts two earlier studies (Lai and Rubin, 2001a; Yeh et al., 2000). In a first study, neur mutant bristles were found to preferentially develop at the clone border in adult flies. This led the authors to propose that neur acts cell autonomously (Yeh et al., 2000). While the approach is similar to the one used here, two methodological differences likely account for the different results. First, in order to detect mutant bristles in adult flies, a weak allele of neur was used. We therefore suggest that these mutant cells can still signal and that they overall produce a stronger inhibitory signal than the wild-type cells due to their higher density. Second, the ratio of wild-type versus mutant bristles was not corrected for the higher density of mutant bristles. By contrast, our analysis had no such bias since only one position per microchaete row was taken into account. In a second study (Lai and Rubin, 2001a), neur was proposed to act cell autonomously because mutant cells at the clone border did not differentiate bristles. This interpretation is based on the hypothesis that pIIa cells receive inhibitory signals from epidermal cells. There is no data to support this hypothesis (see Zeng et al., 1998). We suggest instead that neur mutant pIIb cells fail to signal to their sister cells that therefore adopt a pIIb-like fate. Finally, Lai and Rubin (2001a) also observed at the wing margin that mutant pI cells develop at the clone border, but this phenotype was not quantified. This is in contrast with the detailed study of Pavlopoulos et al. (2001) who have analyzed the same decision and have concluded that neur is primarily required in signal-sending cells. Although we do not exclude the possibility that Neur has additional cell-autonomous functions in signal-receiving cells, our analysis as well as previously published data point to the conclusion that neur is required in signal-sending cells to upregulate Dl signaling. Our proposal that neur-mediated Dl endocytosis in the signal-sending pIIb cell promotes N receptor activation in the signal-receiving pIIa cell is counterintuitive. However, several observations suggest the existence of an unknown mechanism required for S2 cleavage that is associated with ligand endocytosis. First, dynamindependent endocytosis, which is required for both N signaling and Dl endocytosis (Parks et al., 2000; Seugnet et al., 1997), does not appear to be required for the presenilin-dependent transmembrane S3 cleavage of N

in Drosophila (Struhl and Adachi, 2000). Indeed, requirement for dynamin activity can be bypassed by the deletion of the extracellular part of the receptor, which mimicks the extracellular S2 cleavage of N (Struhl and Adachi, 2000). This therefore suggests that endocytosis is required upstream of the S3 cleavage. Second, an unknown activity residing within the intracellular domain of mouse Delta1 is required for or upstream of the S3 cleavage (Shimizu et al., 2002). This observation has been interpreted to suggest that multimerization and/ or endocytosis mediated by signals in the intracellular domain of Delta1 promote N activation. Third, a recent functional analysis of the mind bomb (mib) gene indicates that endocytosis of Dl is required for nonautonomous activation of N in the zebrafish neural tube (Itoh et al., 2003). Like Neur, the E3 ubiquitin ligase Mib coimmunoprecipitates with Dl, promotes Dl ubiquitination and upregulates Dl endocytosis. Interestingly, Mib is structurally distinct from Neur and both Mib and Neur have been conserved during evolution. The function of Drosophila mib is not yet known. Results form elegant cell transplantation studies indicate that mib mutant cells are not defective in receiving inhibitory signals but are less effective at producing inhibitory signals in the neural tube (Itoh et al., 2003). Thus, both Neur- and Mibmediated internalization of Dl appear to nonautomously activate N. Finally, endocytosis-defective Dl proteins remain accumulated at the cell surface and have reduced signaling capacity (Parks et al., 2000). Together, these results have suggested that Dl endocytosis might either expose N on the receiving cell to extracellular S2 cleavage or dissociate the cleaved ectodomain from the rest of the receptor (Parks et al., 2000; Pavlopoulos et al., 2001; Struhl and Adachi, 2000). One prediction from this model is that NECD, but not NICD, should colocalize with Dl into endocytic vesicles. This prediction was not verified in the pIIb cell since both NECD and NICD were found to colocalize with Dl into large endocytic vesicles. It is possible, however, that NECD and NICD are targeted by independent mechanisms toward Hrs-positive vesicles. Whether Neur upregulates N activation by clustering N receptors in signal-receiving cells, by trans-endocytosis of Dl-NECD complexes in signal-producing cells or by yet another mechanism, such as the targeting of Dl into signaling vesicles budding in the lumen of MVBs (exosomes), remains to be studied (Le Borgne and Schweisguth, 2003). Homologs of Drosophila neur have been identified in frogs and mice. The regulatory function of Xneur in Dl endocytosis has been conserved in Xenopus (Deblandre et al., 2001). Xneur RNA is detectable in the developing nervous system and skin at neural plate stages (Deblandre et al., 2001). At this stage, the Xneur-expressing cells may correspond to primary neurons within the neural plate and to specialized ciliated cells in the skin. Expression of Xneur is blocked by constitutive activation of N and is upregulated by the inhibition of N (Deblandre et al., 2001). Thus, the neur gene is expressed in signalsending cells that are selected by lateral inhibition in both Xenopus and Drosophila. However, elucidation of the exact function of Xneur in N signaling awaits Xneur inactivation. Knock out of m-neu1, one of the two murine homologs of Drosophila neur, reveals that m-neu1 is not an essential gene. m-neu1 mutant mice are viable

Unequal Segregation of Neuralized 147

and show no obvious developmental defects (Ruan et al., 2001; Vollrath et al., 2001). It is possible that m-neu2 compensates for the loss of m-neu1 function. Considering the conservation of Neur function from flies to frogs, it will be of interest to examine the distribution of Neur during asymmetric cell divisions in vertebrate species. In summary, Neur and Numb are unequally segregated into the same daughter cell, independently of each other, during the asymmetric division of the pI cell. Neur promotes ligand endocytosis in the anterior cell, thereby activating N in its posterior sister. Concomitantly, Numb inhibits signal transduction in the anterior cell. We propose that Neur and Numb act in parallel to bias N activation.

critical reading. This work was supported in part by a grant from the ARC (#4512).

Experimental Procedures

Bellaı¨che, Y., Radovic, A., Woods, D.F., Hough, C.D., Parmentier, M.L., O’Kane, C.J., Bryant, P.J., and Schweisguth, F. (2001b). The Partner of Inscuteable/Discs-large complex is required to establish planar polarity during asymmetric cell division in Drosophila. Cell 106, 355–366.

Drosophila Stocks The neurP72GAL4 (Bellaı¨che et al., 2001a) driver was used to express Pon-GFP (Lu et al., 1999) and UAS-Neur (Lai and Rubin, 2001a). Mitotic clones for neur1F65, numb2, and numb15 were induced using the FLP-FRT technique by heat shocking first instar larvae (30 min at 37⬚C). The following genotypes were used: (1) y w hsFLP/ w ; FRT82B neur1F65/FRT82B nlsGFP; (2) y w hsFLP/ w ; FRT40A numb2 (or numb15)/FRT40A nlsGFP. neur1F65, numb2, and numb15 (Berdnik et al., 2002) are amorphic alleles, while dlg1P20 is an hypomorphic allele. The amorphic pins⌬62 allele (Yu et al., 2000) is associated with a deletion that extends from ⫺898 to 1210 (⫹1 is at the ATG) (Y. Bellaı¨che and F.S., unpublished data). Immunofluorescence Nota were dissected and processed as previously described (Bellaı¨che et al., 2001a). Primary antibodies used were rabbit anti-Neur antibody (a gift from E. Lai; 1:600), guinea pig anti-Dl (a gift from M. Muskavitch; 1:3000), mouse anti-Dl (C594-9B, Developmental Studies Hybridoma Bank [DSHB]), rabbit anti-Ser antibody (a gift from E. Knust; 1:200), mouse anti-NECD (C458-2H, DSHB; 1:3000), mouse anti-Cut (2B10, DSHB; 1:1000), rabbit anti-Asense (a gift from Y.-N. Jan; 1:2000), guinea pig anti-Hrs (a gift from H. Bellen; 1/600), rabbit anti-Numb (a gift from Y.-N. Jan; 1:2000), rat anti-Elav (7E8, DSHB), rabbit anti-GFP (Molecular Probes; 1:1000) and rat antiSu(H) (1:1000). All Alexa-coupled and Cy3/5-coupled antibodies (1:1000) were from Molecular Probes and Jackson Laboratories. Images were acquired on a Leica SP2 microscope and assembled using Adobe Photoshop. Endocytosis Assay Pupal nota were dissected in Schneider’s Drosophila medium (GIBCO-BRL, Life Technology, Carlsbad, CA) containing 1% fetal calf serum (GIBCO-BRL). After dissection, medium was replaced and supplemented with 1 ␮g/ml 20-OH ecdysone (Sigma, Saint Louis, MO). Pupal nota were cultured for 15 min in presence of the mouse monoclonal anti-Dl antibody C594-9B that recognizes the extracellular portion of Dl. Following medium changes, epithelial cells were fixed. Localization of anti-Dl antibodies was revealed using secondary antibodies. Dl was detected using guinea pig anti-Dl. Drug Treatment of Nota For drug treatment, pupae were dissected as described above. Dissected nota were incubated for 1 hr at 25⬚C with 1 ␮g/ml 20-OH ecdysone (Sigma) alone or with 2 ␮M Latrunculin A (Sigma), 5 mM of BDM (Sigma), or 1 ␮M colcemid (Sigma). Pupal nota were then fixed and processed for immunofluorescence as described above. Acknowledgments We thank H. Bellen, W. Chia, Y.N. Jan, J. Knoblich, E. Knust, E. Lai, M. Muskavitch, and the DSHB (University of Iowa) for flies and antibodies. We thank A. Bardin, Y. Bellaı¨che, A. Brand, C. Goridis, D. Henrique, S. Lee, V. Orgogozo, A. Prochiantz, and I. Stu¨ttem for

Received: January 21, 2003 Revised: May 2, 2003 Accepted: May 2, 2003 Published: July 7, 2003 References Bang, A.G., and Posakony, J.W. (1992). The Drosophila gene Hairless encodes a novel basic protein that controls alternative cell fates in adult sensory organ development. Genes Dev. 6, 1752–1769. Bellaı¨che, Y., Gho, M., Kaltschmidt, J.A., Brand, A.H., and Schweisguth, F. (2001a). Frizzled regulates localization of cell-fate determinants and mitotic spindle rotation during asymmetric cell division. Nat. Cell Biol. 3, 50–57.

Berdnik, D., and Knoblich, J.A. (2002). Drosophila Aurora-A is required for centrosome maturation and actin-dependent asymmetric protein localization during mitosis. Curr. Biol. 12, 640–647. Berdnik, D., Torok, T., Gonzalez-Gaitan, M., and Knoblich, J. (2002). The endocytic protein alpha-adaptin is required for numb-mediated asymmetric cell division in Drosophila. Dev. Cell 3, 221–231. Deblandre, G.A., Lai, E.C., and Kintner, C. (2001). Xenopus neuralized is a ubiquitin ligase that interacts with XDelta1 and regulates Notch signaling. Dev. Cell 1, 795–806. Fortini, M.E. (2001). Notch and presenilin: a proteolytic mechanism emerges. Curr. Opin. Cell Biol. 13, 627–634. Galan, J.M., Moreau, V., Andre, B., Volland, C., and HaguenauerTsapis, R. (1996). Ubiquitination mediated by the Npi1p/Rsp5p ubiquitin-protein ligase is required for endocytosis of the yeast uracil permease. J. Biol. Chem. 271, 10946–10952. Gho, M., Bellaı¨che, Y., and Schweisguth, F. (1999). Revisiting the Drosophila microchaete lineage: a novel intrinsically asymmetric cell division generates a glial cell. Development 126, 3573–3584. Guo, M., Jan, L.Y., and Jan, Y.N. (1996). Control of daughter cell fates during asymmetric division: interaction of Numb and Notch. Neuron 17, 27–41. Heitzler, P., and Simpson, P. (1991). The choice of cell fate in the epidermis of Drosophila. Cell 64, 1083–1092. Huang, F., Dambly-Chaudiere, C., and Ghysen, A. (1991). The emergence of sense organs in the wing disc of Drosophila. Development 111, 1087–1095. Itoh, M., Kim, C.H., Palardy, G., Oda, T., Jiang, Y.J., Maust, D., Yeo, S.Y., Lorick, K., Wright, G.J., Ariza-McNaughton, L., et al. (2003). Mind bomb is a ubiquitin ligase that is essential for efficient activaiton of Notch signaling by Delta. Dev. Cell 4, 67–82. Knoblich, J.A., Jan, L.Y., and Jan, Y.N. (1997). The N terminus of the Drosophila Numb protein directs membrane association and actin-dependent asymmetric localization. Proc. Natl. Acad. Sci. USA 94, 13005–13010. Lai, E.C., and Rubin, G.M. (2001a). neuralized functions cell-autonomously to regulate a subset of notch-dependent processes during adult Drosophila development. Dev. Biol. 231, 217–233. Lai, E.C., and Rubin, G.M. (2001b). Neuralized is essential for a subset of Notch pathway-dependent cell fate decisions during Drosophila eye development. Proc. Natl. Acad. Sci. USA 98, 5637–5642. Lai, E.C., Deblandre, G.A., Kintner, C., and Rubin, G.M. (2001). Drosophila neuralized is a ubiquitin ligase that promotes the internalization and degradation of delta. Dev. Cell 1, 783–794. Le Borgne, R., and Schweisguth, F. (2003). Notch signaling: endocytosis makes delta signal better. Curr. Biol. 13, R273–R275. Lloyd, T.E., Atkinson, R., Wu, M.N., Zhou, Y., Pennetta, G., and Bellen, H.J. (2002). Hrs regulates endosome membrane invagination

Developmental Cell 148

and tyrosine kinase receptor signaling in Drosophila. Cell 108, 261–269. Lu, B., Rothenberg, M., Jan, L.Y., and Jan, Y.N. (1998). Partner of Numb colocalizes with Numb during mitosis and directs Numb asymmetric localization in Drosophila neural and muscle progenitors. Cell 95, 225–235. Lu, B., Ackerman, L., Jan, L.Y., and Jan, Y.N. (1999). Modes of protein movement that lead to the asymmetric localization of partner of Numb during Drosophila neuroblast division. Mol. Cell 4, 883–891. Parks, A.L., Klueg, K.M., Stout, J.R., and Muskavitch, M.A. (2000). Ligand endocytosis drives receptor dissociation and activation in the Notch pathway. Development 127, 1373–1385. Pavlopoulos, E., Pitsouli, C., Klueg, K.M., Muskavitch, M.A., Moschonas, N.K., and Delidakis, C. (2001). neuralized encodes a peripheral membrane protein involved in delta signaling and endocytosis. Dev. Cell 1, 807–816. Polo, S., Sigismund, S., Faretta, M., Guidi, M., Capua, M.R., Bossi, G., Chen, H., De Camilli, P., and Di Fiore, P.P. (2002). A single motif responsible for ubiquitin recognition and monoubiquitination in endocytic proteins. Nature 416, 451–455. Raiborg, C., Bache, K.G., Gillooly, D.J., Madshus, I.H., Stang, E., and Stenmark, H. (2002). Hrs sorts ubiquitinated proteins into clathrincoated microdomains of early endosomes. Nat. Cell Biol. 4, 394–398. Rhyu, M.S., Jan, L.Y., and 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. Ruan, Y., Tecott, L., Jiang, M.M., Jan, L.Y., and Jan, Y.N. (2001). Ethanol hypersensitivity and olfactory discrimination defect in mice lacking a homolog of Drosophila neuralized. Proc. Natl. Acad. Sci. USA 98, 9907–9912. Santolini, E., Puri, C., Salcini, A.E., Gagliani, M.C., Pelicci, P.G., Tacchetti, C., and Di Fiore, P.P. (2000). Numb is an endocytic protein. J. Cell Biol. 151, 1345–1352. Schweisguth, F., and Posakony, J.W. (1994). Antagonistic activities of Suppressor of Hairless and Hairless control alternative cell fates in the Drosophila adult epidermis. Development 120, 1433–1441. Seugnet, L., Simpson, P., and Haenlin, M. (1997). Requirement for dynamin during Notch signaling in Drosophila neurogenesis. Dev. Biol. 192, 585–598. Shih, S.C., Katzmann, D.J., Schnell, J.D., Sutanto, M., Emr, S.D., and Hicke, L. (2002). Epsins and Vps27p/Hrs contain ubiquitin-binding domains that function in receptor endocytosis. Nat. Cell Biol. 4, 389–393. Shimizu, K., Chiba, S., Saito, T., Takahashi, T., Kumano, K., Hamada, Y., and Hirai, H. (2002). Integrity of intracellular domain of Notch ligand is indispensable for cleavage required for release of the Notch2 intracellular domain. EMBO J. 21, 294–302. Struhl, G., and Adachi, A. (2000). Requirements for presenilin-dependent cleavage of notch and other transmembrane proteins. Mol. Cell 6, 625–636. Vollrath, B., Pudney, J., Asa, S., Leder, P., and Fitzgerald, K. (2001). Isolation of a murine homologue of the Drosophila neuralized gene, a gene required for axonemal integrity in spermatozoa and terminal maturation of the mammary gland. Mol. Cell. Biol. 21, 7481–7494. Yeh, E., Zhou, L., Rudzik, N., and Boulianne, G.L. (2000). Neuralized functions cell autonomously to regulate Drosophila sense organ development. EMBO J. 19, 4827–4837. Yeh, E., Dermer, M., Commisso, C., Zhou, L., McGlade, C.J., and Boulianne, G.L. (2001). Neuralized functions as an E3 ubiquitin ligase during Drosophila development. Curr. Biol. 11, 1675–1679. Yu, F., Morin, X., Cai, Y., Yang, X., and Chia, W. (2000). Analysis of partner of inscuteable, a novel player of Drosophila asymmetric divisions, reveals two distinct steps in inscuteable apical localization. Cell 100, 399–409. Zeng, C., Younger-Shepherd, S., Jan, L.Y., and 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.