Numb Regulates Post-endocytic Trafficking and Degradation of Notch1*

Notch is a transmembrane receptor that controls cell fate decisions during development and tissue homeostasis. Both activation and attenuation of the Notch signal are tightly regulated by endocytosis. The adaptor protein Numb acts as an inhibitor of Notch and is known to function within the intracellular trafficking pathways. However, a role for Numb in regulating Notch trafficking has not been defined. Here we show that mammalian Notch1 is constitutively internalized and trafficked to both recycling and late endosomal compartments, and we demonstrate that changes in Numb expression alter the dynamics of Notch1 trafficking. Overexpression of Numb promotes sorting of Notch1 through late endosomes for degradation, whereas depletion of Numb facilitates Notch1 recycling. Numb mutants that do not interact with the ubiquitin-protein isopeptide ligase, Itch, or that lack motifs important for interaction with endocytic proteins fail to promote Notch1 degradation. Our data suggest that Numb inhibits Notch1 activity by regulating post-endocytic sorting events that lead to Notch1 degradation.

Notch is a transmembrane receptor that controls cell fate decisions during development and tissue homeostasis. Both activation and attenuation of the Notch signal are tightly regulated by endocytosis. The adaptor protein Numb acts as an inhibitor of Notch and is known to function within the intracellular trafficking pathways. However, a role for Numb in regulating Notch trafficking has not been defined. Here we show that mammalian Notch1 is constitutively internalized and trafficked to both recycling and late endosomal compartments, and we demonstrate that changes in Numb expression alter the dynamics of Notch1 trafficking. Overexpression of Numb promotes sorting of Notch1 through late endosomes for degradation, whereas depletion of Numb facilitates Notch1 recycling. Numb mutants that do not interact with the ubiquitin-protein isopeptide ligase, Itch, or that lack motifs important for interaction with endocytic proteins fail to promote Notch1 degradation. Our data suggest that Numb inhibits Notch1 activity by regulating post-endocytic sorting events that lead to Notch1 degradation.
The Notch receptor is a heterodimeric, single pass, transmembrane protein that is an integral component of a conserved signaling pathway required for embryonic and postnatal development. Notch signaling controls cell fate decisions, maintenance of stem cells, cellular differentiation, proliferation, and apoptosis through direct cell to cell contact (1). Transmission of the Notch signal is initiated through binding of the Notch receptor to transmembrane ligands of the Delta/Serrate/Lag2 family expressed on neighboring cells (2,3). Ligand binding induces a sequential series of proteolytic cleavage events leading to a final intramembrane cleavage by the ␥-secretase complex that releases the active intracellular domain of Notch (4,5). This form of Notch functions within the nucleus as a cotransactivator with the CSL (CBF1/RBPj, Suppressor of Hairless Su(H), Lag1) family of transcription factors to modulate transcription of target genes such as Hes1 (the mammalian homologue of the hairy and enhancer of split gene complex in Drosophila) and Herp (Hes-related) genes (6,7).
The coordinated processing, endocytosis, and trafficking of the Notch receptor and its ligand are important in controlling Notch activation. Ubiquitin-dependent internalization and recycling of Delta/Serrate/Lag2 ligands have an established role in the activation of Notch signaling (8 -19). In contrast, much less is known regarding the role of endocytosis and trafficking of the Notch receptor itself. Studies to date suggest that endocytosis of Notch is important for ligand-mediated activation of Notch. Monoubiquitination and endocytosis of a constitutively active mutant form of the Notch1 receptor were shown to enhance ␥-secretase cleavage and activate downstream signaling events in a mammalian system (20). Similarly, mutational analysis in Drosophila showed that entry into the early endosome is necessary for Notch activation (21).
Ubiquitination and trafficking through the endocytic pathway also play a role in the down-regulation of Notch. A conserved di-leucine sorting motif identified in the cytoplasmic tail of the Caenorhabditis elegans Notch receptor, LIN-12, mediates its constitutive internalization and degradation during vulval development (22). WWP-1, the C. elegans orthologue of the Drosophila ubiquitin ligases Suppressor of Deltex (Su(Dx)) and dNedd4, is required for LIN-12 down-regulation (23). Furthermore, in Drosophila both Su(Dx) and dNedd4 act to limit Notch signaling by regulating post-endocytic sorting of Notch (24,25). Together these studies suggest that both entry and trafficking of Notch within the endocytic pathway are important in the regulation of its activity. However, the factors involved in sorting Notch into activation or degradative pathways are poorly characterized.
Genetic evidence in Drosophila has shown that the adaptor protein Numb acts as an inhibitor of Notch signaling during development of both the peripheral and central nervous systems and during muscle cell differentiation (26 -28). Mammalian Numb orthologues appear to function in a conserved fashion within the mammalian Notch1 pathway (26, 29 -31). For example, overexpression of mammalian Numb antagonizes Notch1-dependent transactivation of the Hes1 promoter (32) and inhibits Notch1 activity in neurite growth (33,34). Both Drosophila and mammalian Numb are asymmetrically localized in dividing precursor cells and are preferentially segregated to one of the two daughter cells upon cell division, thus ensuring cells adopt distinct cell fates, through the suppression of Notch signaling (26,28,(35)(36)(37)(38)(39)(40).
Numb interacts with components of the endocytic machinery and is localized to plasma membrane-associated and intracellular vesicles (41,42). Numb binds to the ␣-adaptin subunit of the AP-2 complex (clathrin adaptor protein 2) in both Drosophila and mammalian cells (42,43) and is required for asymmetric localization of Drosophila ␣-adaptin (43). Numb also interacts with EH domain-containing proteins Eps15 and the EHD/RME-1 family of endocytic proteins (41,42). In mammalian cells, knockdown of Numb using RNA interference or overexpression of Numb or short fragments of Numb interferes with transmembrane protein endocytosis and trafficking through an unknown mechanism (41,42,44,45). It has been proposed that Numb suppression of Notch signaling is mediated through AP-2 recruitment and regulation of Notch endocytosis (43,46,47), although to date there is no direct evidence to support this model. Previously, we showed that Numb also binds to the Su(Dx) orthologue Itch and cooperates with Itch to enhance the ubiquitination of membrane-associated Notch1 (32), suggesting that Numb might function in a post-endocytic compartment to regulate Notch trafficking.
Here we investigate the role of Numb in the ligand-independent endocytosis and trafficking of Notch1. We show that overexpression of Numb promotes Notch1 trafficking and degradation, whereas depletion of Numb facilitates Notch1 recycling. Numb mutants defective for binding the E3 2 ligase, Itch, or the endocytic proteins ␣-adaptin, Eps15 and EHD, fail to promote Notch1 degradation. Together our data suggest that Numb suppresses Notch1 signaling by regulating post-endocytic sorting events that lead to Notch1 degradation.

MATERIALS AND METHODS
Antibodies and Constructs-Full-length mouse Notch1 cDNA (FL-Notch1), containing a single internal Myc epitope tag cloned into the HindIII site between amino acid residues Ala 2292 and Ser 2293 , was subcloned into pcDNA1(ϩ) vector. All Numb mutant constructs were generated from Numbp66 cDNA and cloned into a pEF vector. Numb⌬PTBC was generated using PCR-amplified fragments representing amino acids Met 1 to Lys 85 and Lys 174 to the stop codon and fused in-frame to the pEF parental vector. Numb⌬C has a deletion of the sequence coding for the 41 amino acids at the carboxyl terminus. All cDNAs generated by PCR were verified by sequencing in both directions. Itch cDNAs were the kind gift from G. Gish and T. Pawson. HA-tagged Notch1 cDNA (pBOS HA-N1) has been described previously (15). Rabbit anti-NotchIC (06-808, Upstate Biotechnology Inc., Lake Placid, NY) and rabbit anti-NotchEC (06-809, Upstate Biotechnology Inc.) antibodies were used for immunoblots. Rabbit anti-intracellular domain of Notch (2421, Cellular Signaling), goat anti-NotchC20 (sc-6015, Santa Cruz Biotechnology, Santa Cruz, CA), anti-Myc (9E10 ascites), and rabbit anti-HRS antibodies (kind gift of S. Urbe) (48) were used for immunostaining. Anti-Numb was generated in rabbits with a synthetic peptide to the carboxyl terminus as described previously (30).
Cell Culture and Transfections-HEK293T cells were grown in Dulbecco's modified Eagle's medium (DMEM; Wisent) supplemented with 10% fetal bovine serum (FBS) and transfected with Lipofectamine reagent (Invitrogen) in Opti-MEM (Invitrogen) according to the manufacturer's instructions. Pooled transfected cells were replated in individual 6-cm dishes for time course experiments. C2C12 cells were grown in DMEM supplemented with 10% FBS and 5% calf serum. C2C12 cells that stably express full-length, HA-tagged Notch1 were grown in 10% FBS and 5% calf serum supplemented with 2 g/ml puromycin. C2C12 cells were transfected with Lipofectamine 2000 reagent (Invitrogen) using 3 g of cDNA per 6-well dish and 12 l of Lipofectamine 2000. Equal numbers of pooled transfected cells were replated in individual 6-cm dishes. Cells were used 18 to 24 h post-transfection.
For siRNA silencing experiments, two 21-nucleotide siRNA oligomers (Dharmacon Research) were designed to regions corresponding to the Numb coding sequence (GCACCUGC-CCAGUGGAUCC) and 3Ј-untranslated region sequence (GUAGCACAUUGCAACAACA). The Scramble II duplex from Dharmacon (D-001205-20) was used as a negative control for siRNA activity. Only C2C12 cells were used for siRNA experiments because significant knockdown of endogenous Numb could not be achieved in HEK293T cells. Overexpression or depletion of Numb was confirmed by immunoblotting cell lysates with anti-Numb.
Surface Biotinylation and Trafficking Assays-Transfected cells grown to 70% confluency were cooled on ice for 30 min, washed with cold PBS, and then labeled with EZ-Link NHS-SSbiotin (Pierce) in biotinylation buffer (154 mM NaCl, 10 mM HEPES, 3 mM KCl, 1 mM MgCl 2 , 0.1 mM CaCl 2 , 10 mM glucose, pH 7.6) for 1 h at 4°C. After two washes with cold PBS, cells were incubated with DMEM containing 10% FBS and 100 mM glycine for 5 min on ice to quench unconjugated biotin and washed several times with cold PBS. One sample plate was removed and lysed in PLC lysis buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1.5 mM MgCl 2 , 1% Triton X-100, 1 mM EGTA, 10 mM sodium pyrophosphate, 100 mM sodium fluoride containing COMPLETE protease inhibitor tablets (Roche Applied Science)) to measure total surface biotinylation. A second sample plate was also removed to determine the efficiency of removal of biotinyl groups on the cell surface by three 20-min treatments with 50 mM 2-mercaptoethanesulfonic acid (MeSNa) in pre-chilled stripping buffer (100 mM NaCl, 50 mM Tris-HCl, 1 mM MgCl 2 , 0.1 mM CaCl 2 , pH 8.6) at 4°C. The remaining samples were then incubated at 37°C in pre-warmed DMEM containing 10% FBS over 30 min. At indicated time points, cells were washed in cold PBS and stripped of surface biotin in MeSNa stripping buffer (as described above). Cells were then lysed and protein lysates quantitated. Equal amounts of protein lysate were mixed with immobilized streptavidin-Sepharose beads (Pierce) overnight at 4°C to isolate biotinylated proteins and then washed three times in Nonidet P-40 wash buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 2 mM EGTA, 10% glycerol, 1.5 mM CaCl 2 , 1% Nonidet P-40). The bed volume was removed with a 30-gauge needle before resuspending in SDS-Laemmli sample buffer. Recovered biotinylated proteins were analyzed by SDS-PAGE and Western blot using indicated antibodies. In all experiments, whole cell lysates were analyzed by SDS-PAGE and Western blot to examine FL-Notch and Numb expression.
For internalization experiments, cells were labeled with biotin as described above and incubated at 18°C in DMEM containing 10% FBS and 25 mM HEPES, pH 7.6. Shifting cells to 18°C results in an accumulation of endocytosed proteins in early or sorting endosomes in many cell types (49 -52). At indicated time points cells were MeSNA-stripped, lysed in PLC lysis buffer, and intracellular biotinylated Notch examined using streptavidin-Sepharose beads as described above.
For recycling assays, biotinylated cells were incubated at 18°C for 2 h in DMEM containing 25 mM HEPES, pH 7.6, and 10% FBS to accumulate internalized biotinylated surface proteins, and then MeSNa-stripped as described above. After washing with cold PBS, cells were shifted to 37°C to resume trafficking. At the indicated time points, one set of cells was stripped of surface biotin and lysed, and a second set was incubated in stripping buffer without MeSNA and then lysed. Intracellular pools of biotinylated protein were examined as described above.
For protein quantification, protein bands were visualized by enhanced chemiluminescence (Pierce) and signals quantified by densitometric scanning using an Alpha Innotech Fluorochem 8000 imaging system. Data were acquired using the Alpha Innotech Fluorochem 8000 software and exported to Excel (Microsoft) for analysis. All statistical analysis comparisons were done with Student's t test in Excel.
Immunofluorescent Microscopy-For subcellular localization studies, HEK293T and C2C12 cells were plated onto poly-Lornithine-coated glass coverslips 18 h before use. Cells were fixed in 2% paraformaldehyde in PBS containing Ca 2ϩ and 30 mM sucrose for 20 min at room temperature, washed with 100 mM glycine in PBS/Ca 2ϩ for 10 min, and then permeabilized for 10 min with either 0.2% Triton X-100 or 0.05% saponin in PBS/ Ca 2ϩ . After several washes in PBS, cells were blocked with 3% normal donkey serum in PBS/Ca 2ϩ for 30 min at room temperature and then incubated with primary antibodies diluted in blocking serum for 30 min at 37°C. Cells were washed five times for 5 min in PBS/Ca 2ϩ and primary antibodies indirectly labeled for 30 min at 37°C as follows: Cy5-conjugated donkey anti-rabbit IgG (1:500) and Cy3-conjugated donkey anti-mouse IgG (1:500) from Jackson ImmunoResearch Laboratories, and Alexa488-conjugated donkey anti-mouse (1:500; Molecular Probes, Eugene, OR). Cells were washed with PBS and mounted in DAKO mounting medium. Cells were examined with a Zeiss Axiovert 200 microscope equipped with a Hamamatsu Orca AG CCD camera and spinning disk confocal scan head. Volocity software was used to acquire and analyze images. Figures were made using Adobe Photoshop digital image software (San Jose, CA). To label the late endosomal/lysosomal cell compartment, cells were incubated overnight in normal growth media containing 10 g/ml FITC-conjugated 70-kDa dextran (Molecular Probes) followed by a 2-h chase period in dextran-free media containing anti-HA.
Antibody Uptake and Recycling Assay-Cells expressing HA-N1, grown on polyornithine-coated coverslips (ϳ50% confluent), were incubated with Alexa488-conjugated anti-HA (1:1000) in growth media for 30 min. After cooling on ice and washing with cold growth media, anti-mouse IgG-binding sites on cell surface Alexa488 anti-HA were blocked by incubation with excess unlabeled anti-mouse (1:100; Jackson Immuno-Research) for 15 min at 4°C. Cells were washed and then rewarmed to 37°C in growth media between 0 and 30 min. Alexa488 anti-HA that had re-inserted into the plasma membrane during this incubation period was identified by incubation with Cy3-labeled anti-mouse (1:500) at 4°C for 15 min followed by washing and fixation in 2% paraformaldehyde. Cells were imaged using a 25ϫ water objective (0.8 NA) and spinning disk confocal microscope as described above. The total cellular, endocytosed HA-Notch1 (Alexa488 anti-HA fluorescence intensity) and cell surface Alexa488 anti-HA labeled HA-N1 (Cy3-␣ mouse fluorescence intensity) were quantified using the measurement functions of Volocity imaging software.

Notch1 Is Constitutively Internalized and Recycled-To
examine constitutive Notch1 receptor trafficking, we measured the internalization and subsequent trafficking of surface-labeled endogenous Notch1 in C2C12 cells. Subconfluent C2C12 cells were surface-labeled with biotin at 4°C and then incubated at 37°C to allow internalization and post-endocytic trafficking of biotinylated Notch1. At specific time points, the cells were treated with MeSNa to remove any biotin label on proteins remaining at the cell surface. Intracellular biotinylated Notch1 protein was recovered with streptavidin beads and quantified by Western blot analysis using anti-Notch1 antibodies. Biotinylation of both subunits of the Notch1 heterodimer was observed confirming that the processed Notch1 receptor is present at the plasma membrane (Fig. 1A, upper panels). Following incubation at 37°C, a steady increase in intracellular biotinylated Notch1 was observed (Fig. 1A, lower panel) indicating that endogenous Notch1 is constitutively internalized from the cell surface. A similar analysis of HEK293T cells expressing the full-length Notch1 receptor (FL-Notch1) was also performed. In HEK293T cells, both subunits of the Notch1 heterodimer were biotinylated at the plasma membrane indicating that ectopically expressed Notch1 is processed and present at the cell surface (Fig. 1B, upper and lower panels, respectively). Similar to endogenous Notch1 in C2C12 cells, biotinylated FL-Notch1 was endocytosed and accumulated intracellularly in 293T cells (Fig. 1B) demonstrating that both endogenous and ectopically expressed Notch1 are constitutively trafficked. To examine post-endocytic trafficking of Notch1, HEK293T cells expressing FL-Notch1 were labeled with biotin at 4°C and then incubated at 18°C to allow internalization of biotinylated FL-Notch1 from the plasma membrane under conditions that block recycling to the cell surface. Cells were then treated with MeSNa to remove biotin at the cell surface and returned to 37°C to resume trafficking. At the indicated time points, cells were either lysed to examine the total amount of biotinylated protein or treated a second time with MeSNa before lysis to remove biotin from proteins that had returned to the cell surface. A steady loss in the intracellular pool of biotinylated FL-Notch1, with almost complete loss after 30 min, was observed (Fig. 1C, compare lane 1 and lanes 2-4, lower panel) suggesting that internalized FL-Notch1 was either being returned to the cell surface or degraded. To assess FL-Notch1 degradation, the rate of change in the total amount of biotinylated FL-Notch1 was measured in cells shifted to 37°C but not stripped of surface biotin a second time. Under these experimental conditions, the total amount of biotinylated FL-Notch1 remained relatively unchanged at 5 min (Fig. 1C, compare lane 1 and lane 2, upper panel) suggesting that the loss of biotinylated FL-Notch1 from the intracellular pool at 5 min is the result of recycling to the cell surface. However, by 30 min the total cellular, biotinylated FL-Notch1 had decreased suggesting that a portion of biotinylated FL-Notch1 is targeted for degradation at later times. A similar trend was observed when an antibody against the extracellular domain of the FL-Notch heterodimer (Fig. 1D) was used, indicating that the intact receptor is constitutively trafficked.
We used immunocytochemistry and antibody uptake experiments to further investigate the constitutive internalization and trafficking of Notch1 toward a degradative pathway or recycling to the plasma membrane. HEK293T cells were transiently transfected with pBOS HA-N1, which encodes fulllength Notch1 with an extracellular amino-terminal HA tag (15). Cells were surface-labeled with Alexa488-conjugated anti-HA (Al488␣HA) while at 4°C to prevent endocytosis ( Fig.  2A, left panel). When the cells were warmed to 37°C for 60 min, surface-labeled HA-Notch1 was greatly reduced with a concomitant labeling of many intracellular vesicles ( Fig. 2A, right  panel). To characterize the HA-Notch1-containing vesicles, we preincubated cells overnight with the fluid-phase marker FITC-dextran followed by anti-HA for 2 h at 37°C. Under these conditions, some surface-labeled HA-N1 colocalized with FITC-dextran positive vesicles (Fig. 2B). This suggests that constitutively endocytosed Notch1 is sorted to late endosomes and degraded in a lysosomal compartment. In addition, when HEK293T cells expressing HA-N1 were incubated with Al488␣HA at 37°C for 2 h, fixed and immunostained with anti-HRS, a marker for late endosomes, we observed colocalization between the labeled HA-Notch1 and HRS (Fig. 2C).
To examine whether Notch1 is also recycled back to the plasma membrane, we used C2C12 cells that stably express HA-N1 (C2C12 HA-N1 (15)). C2C12 HA-N1 cells were incubated with Al488␣HA at 37°C to allow uptake of labeled HA-N1 into intracellular vesicles, and any remaining extracellular Al488␣HA bound to the cell surface was blocked to subsequent binding of Cy3-conjugated anti-mouse (Cy3-antimouse) by incubation with excess unlabeled anti-mouse IgG at 4°C. Cells were re-warmed to 37°C to allow trafficking to continue for 10, 20, or 30 min, and we detected the re-appearance of Al488␣HA-labeled HA-N1 at the plasma membrane by incubating nonpermeabilized cells with Cy3 anti-mouse, which was measured using spinning disk confocal microscopy (Fig. 3). A steady increase in Cy3-anti-mouse labeling of the cell surface HA-N1 was observed from 0 to 30 min, although total HA-N1 remained constant, indicating that constitutively internalized antibody labeled HA-N1 is also recycled (Fig. 3).
Numb Regulates Notch1 Intracellular Trafficking-The variability in HA-N1 expression levels in cell lines precluded a quantitative analysis of Notch antibody uptake and trafficking events in FIGURE 1. Notch1 is constitutively trafficked independent of ligand. A, subconfluent C2C12 cells expressing endogenous Notch1 receptor were labeled with biotin at 4°C. After 1 h, one plate of cells was lysed to examine total surface biotinlabeled Notch1 (Tot), and a second was treated with MeSNa (0Ј) to strip the biotin label. Cells were then incubated at 37°C and at the indicated times stripped of biotin remaining at the cell surface. Intracellular pools of biotinylated Notch1 were recovered from whole cell lysates using streptavidin beads and analyzed by Western blot using antisera specific for the carboxyl terminus of Notch1, anti-NotchIC, or for the extracellular epidermal growth factor-like repeat region of Notch1, anti-NotchEC, as indicated. IB, immunoblot. B, HEK293T cells transfected with FL-Notch1 were labeled with biotin, and intracellular pools of biotin-labeled FL-Notch1 were examined as described above. C and D, FL-Notch1 is recycled. HEK293T cells expressing FL-Notch1 were biotinylated, incubated at 18°C for 2 h to internalize biotin-labeled FL-Notch1, and then treated with MeSNa to remove surface biotin. Cells were then returned to 37°C to resume trafficking. One set of cells was treated a second time with MeSNa followed by lysis (intracellular, lower panel), although the second set was lysed without MeSNa treatment (surface ϩ IC, intracellular; upper panel). Pools of biotinylated FL-Notch1 were recovered with streptavidin beads and analyzed by Western blot using antisera specific for either the intracellular (C) or extracellular domain (D) of Notch1. the presence of excess Numb or Numb depletion. Therefore, we used a biochemical approach to examine the effects of Numb on constitutive Notch1 receptor trafficking. We overexpressed p66 Numb in C2C12 cells and measured intracellular accumulation of surface-biotinylated, endogenous Notch1 over 30 min (Fig. 4A). Overexpression of Numb resulted in a reduction of the internalized pool of biotinylated Notch1 that accumulated after 30 min (Fig. 4A) compared with control cells. Similarly, when Numb and FL-Notch1 were coexpressed in HEK293T cells, the amount of intracellular biotinylated FL-Notch1 protein that accumulated after 30 min of internalization (Fig. 4B, lower panel) was decreased compared with FL-Notch1 alone (Fig. 4B, upper  panel). Under identical conditions, overexpression of Numb had no effect on the intracellular accumulation of surface-biotinylated epidermal growth factor receptor (EGFR) or transferrin receptor ( Fig. 4C and data not shown) indicating that Numb overexpression has a specific effect on Notch1 trafficking.
To assess Notch1 receptor trafficking in the absence of Numb, C2C12 cells were transfected with Numb-specific RNA duplexes or with control scrambled RNA duplexes. In Numb siRNA transfected cells, accumulation of internalized, biotinylated Notch1 was markedly reduced compared with cells transfected with scrambled siRNA, suggesting that depletion of Numb protein also alters the trafficking of Notch1 (Fig. 4D). Together these data suggest that altering the protein levels of Numb within the cell disrupts constitutive steady state trafficking of Notch1.
Numb Endocytic Motifs and Interaction with Itch Are Required to Regulate Notch1 Trafficking-To determine the regions of Numb required for modulating Notch1 trafficking, we examined the effects of the Numb mutant (Nb⌬C) lacking the last 41 amino acids of Numb, which includes the NPF and DPF motifs required for Numb binding FIGURE 2. Surface-labeled Notch1 is constitutively endocytosed into vesicles that colocalize with FITCdextran and HRS. A, C2C12 cells that stably express HA-tagged Notch1 were surface-labeled with Al488anti-HA at 4°C (left panel), washed, and incubated at 37°C for 60 min to allow for endocytosis (right panel). B, HEK293T cells transiently expressing HA-Notch1 were incubated for 18 h with 1 mg/ml FITC-dextran (70,000 molecular weight; Molecular Probes) in normal growth media at 37°C. The cells were washed and chased with media containing anti-HA for 2 h, fixed, permeabilized, and then stained with Cy3-␣-mouse to identify ␣-HAlabeled Notch1 and with rabbit ␣-NbC (Cy5-␣-rabbit secondary) to mark the plasma membrane. C, HEK293T cells transiently expressing HA-Notch1 were incubated with Alexa488-␣-HA (green) at 37°C for 1 h. Following fixation with 2% paraformaldehyde, cells were permeabilized and immunostained with rabbit ␣HRS (Cy5labeled donkey ␣-rabbit) and guinea pig ␣NbC (Cy3 labeled donkey anti-guinea pig). Images were captured using a Zeiss Axiovert 200 microscope equipped with a Hamamatsu Orca AG CCD camera and spinning disk confocal scan head. Images were processed using Volocity software, and figures were prepared using Photoshop. Scale bars represent 10 m. Images are representative of three independent experiments.
to ␣-adaptin, Eps15, and EHD4 (41,42), and the Nb⌬PTBC mutant, which lacks 88 amino acids within the carboxyl-terminal region of the PTB domain required for binding to the E3 ligase, Itch (32) (Fig. 5A). FL-Notch1 was cotransfected with wild type Numb, Nb⌬C, or Nb⌬PTBC, and the accumulation of intracellular biotinylated FL-Notch1 was examined. Although wild type Numb (NbWT) decreased the amount of biotinylated FL-Notch1 within intracellular pools compared with control cells (Fig. 5B, top and middle panels), expression of Nb⌬C had no effect on the amount of biotinylated FL-Notch1 at the cell . Surface-labeled Notch1 is constitutively recycled to the plasma membrane. A, C2C12 HA-N1 cells were incubated with Al488anti-HA (green) in growth media at 37°C for 30 min to accumulate surface-labeled HA-Notch1 in intracellular vesicles (0 min No Block). Cy3-labeled anti-mouse-binding sites on surface Al488anti-HA were blocked with excess unlabeled anti-mouse IgG at 4°C (0 min). Cells were then re-warmed to 37°C to allow for continued trafficking of internalized HA-Notch1 (10 min and 20 min). Prior to fixation, all cells were incubated with Cy3-labeled anti-mouse (red) to visualize unblocked, Al488-anti-HA tagged, and HA-Notch1 at the cell surface. Cells were imaged using a ϫ25 water objective and spinning disk confocal microscope as described under "Materials and Methods." Scale bars represent 20 m. Images are representative of three independent experiments. B, average total cellular fluorescence intensity in each channel was measured in 45-55 cells for each condition shown in A. The mean intensities for each condition are shown, allowing comparison of the changes in the total labeled HA-Notch1 (green bars, left) and surface-labeled HA-Notch1 (red bars, right).
surface or accumulating within intracellular pools (Fig. 5B, bottom panel). This suggests that interaction with endocytic proteins ␣-adaptin, Eps15, and/or EHD/Rme1 is required for Numb to regulate Notch1 trafficking. Overexpression of Nb⌬PTBC also failed to cause a reduction in intracellular pools of Notch1 and, in contrast to wild type Numb, resulted in an increased accumulation of intracellular biotinylated FL-Notch1 compared with control cells (Fig. 5C). These observations indicate a distinct role for the Numb PTB domain in regulating intracellular FL-Notch1 trafficking. Previously we have shown that Nb⌬PTBC is defective for binding to Itch, and others have demonstrated a role for Itch in Notch1 trafficking to the lysosome for degradation (32,53,54). Together these data suggest that Numb-mediated sorting of Notch1 also requires Notch1 ubiquitylation. In support of this, we observed that overexpression of wild type Itch promoted depletion of intracellular pools of biotinylated FL-Notch1 (Fig. 5D) similar to that observed with Numb overexpression. Expression of a ligase-dead mutant form of Itch, ItchC830A, had a reduced effect on FL-Notch1 trafficking but did cause a reduction in intracellular pools at 30 min.
Numb Promotes Post-endocytic Trafficking and Degradation of Notch1-The reduction in the pool of internalized Notch1 observed could reflect changes in either the rate of internalization, degradation, or recycling. To determine the step in the endocytic pathway influenced by Numb expression levels, we first examined Notch1 internalization. HEK293T cells expressing FL-Notch1 were biotinylated and then incubated at 18°C to allow for internalization but not the subsequent intracellular trafficking of biotinylated FL-Notch1. A steady accumulation of intracellular biotinylated Notch1 was observed over 4 h (Fig.  6A, upper panel). Overexpression of Numb had no effect on FL-Notch1 internalization compared with control cells (Fig.  6A, lower panel). Similarly, when Numb protein levels were depleted in C2C12 cells using Numb-specific RNA interference duplexes, no change in Notch1 internalization was observed compared with control (Fig. 6B).
We next examined the effect of Numb on post-endocytic trafficking. Using a procedure similar to that described in Fig.  1C, biotinylated FL-Notch1 was allowed to accumulate in HEK293T cells, with or without overexpressed Numb. Following the removal of surface biotin with MeSNa, we measured changes in the total pool of biotinylated Notch1 (intracellular plus surface Notch1), and we compared this to changes in the intracellular pool of biotinylated FL-Notch1 remaining over 30 min. In cells expressing FL-Notch1 alone, a steady decrease in the amount of intracellular biotinylated FL-Notch1 was observed in the MeSNa-treated cells over 30 min (Fig. 7A, compare lane 1 with lanes 2-4, lower panel), whereas the total amount of biotinylated FL-Notch1 remained relatively stable at 5 and 15 min and then decreased after 30 min (Fig. 7A, lane 1  with lanes 2-4, upper panel). Because the total pool of biotinylated FL-Notch1 remained constant at 5 and 15 min, indicating that minimal degradation had occurred, the decrease in the intracellular biotinylated FL-Notch1 (in the MeSNa treated samples) likely represents loss because of receptor recycling back to the cell surface. Coexpression of FL-Notch1 and Numb resulted in an accelerated loss of both the intracellular pool of biotinylated FL-Notch1 (Fig. 7B, compare lane 1 with lanes  2-4, lower panels) and the total biotinylated FL-Notch1 (Fig.  7B, compare lane 1 with lanes 2-4, upper panels) compared with cells expressing FL-Notch1 alone, indicating that the loss A, C2C12 cells expressing endogenous Notch1 receptor transfected with empty vector or Numb were labeled with biotin at 4°C for 1 h, allowed to traffic at 37°C, and the intracellular pools of biotin-labeled Notch1 (biotin-N1) examined using streptavidin beads and Western blot analysis as described previously. IB, immunoblot; Tfxn, transfection. B, HEK293T cells were transfected with FL-Notch1 alone or FL-Notch1 and Numb and the trafficking of the FL-Notch1 receptor examined. C, Numb does not alter trafficking of EGFR. HEK293T were cotransfected with EGFR alone or with Numb, and accumulation of intracellular biotin-labeled EGFR examined as described above. D, loss of Numb protein disrupts intracellular trafficking of Notch1. C2C12 cells were transfected with Numb-specific RNA duplexes or control scrambled RNA duplexes, and the trafficking of endogenous Notch1 receptor was examined 48 h post-transfection as described above. Whole cell lysates were examined for Numb protein expression using antisera specific for Numb. Scram, Scrambled siRNA. Representative Western blots are shown. Films from three independent experiments were scanned and quantitated using an Alpha Innotech Fluorochem 8000 imaging system (see "Material and Methods"), and results are shown in graphs (right). Data are expressed relative to total (Tot) surface-biotinylated Notch1 (mean Ϯ S.D.; n ϭ 3). Significance was determined by Student's t test; **, p Ͻ 0.01; *, p Ͻ 0.05 compared with control. HEK293T cells expressing FL-Notch1 and wild type Itch or a ligase-dead mutant were biotinylated, and trafficking of FL-Notch1 was examined. Whole cell lysates were examined for equal expression of ItchWT and ItchC830A. Representative Western blots are shown. Films from three independent experiments were scanned and quantitated using an Alpha Innotech Fluorochem 8000 imaging system (see "Materials and Methods") and results are shown in the graphs (right). Data are expressed relative to total (Tot) surface-biotinylated Notch1 (mean Ϯ S.D.; n ϭ 3). Significance was determined by Student's t test; **, p Ͻ 0.01; *, p Ͻ 0.05 compared with control.
of intracellular biotinylated FL-Notch1 under these conditions is primarily because of degradation of FL-Notch1. These results imply that overexpression of Numb reduces intracellular pools of biotinylated Notch1 by promoting Notch1 degradation.
Next we examined the effect of Numb depletion on postendocytic trafficking of endogenous Notch1. In C2C12 cells expressing a nonspecific scrambled RNA duplex, the total amount of biotinylated Notch1 remained steady after 5 and 15 min (Fig. 7C, compare lane 1 with lanes 2-4, upper panel), whereas the intracellular pool of biotinylated endogenous Notch1 protein was steadily lost over 30 min (Fig. 7C, compare  lane 1 with lanes 2-4, lower panel), similar to the trend observed in Fig. 7A. Numb depletion by siRNA resulted in an accelerated loss of the accumulated biotinylated Notch1 at early time points (Fig. 7D, compare the lower panels of Fig. 7, C  and D). However, the total amount of biotinylated Notch1 remained steady after 5 and 15 min (Fig. 7, C and D, compare  lane 1 with lane 2-4, upper panels) suggesting that the accelerated loss of intracellular biotinylated Notch1 in Numb depleted cells is a consequence of Notch1 recycling rather than degradation. Together these results support a role for Numb in promoting the routing of Notch1 from a constitutive recycling pathway into an endocytic compartment that leads to degradation.

DISCUSSION
Accumulating evidence suggests that entry of Notch1 into the endosomal pathway and subsequent trafficking is intimately linked to downstream signaling outcomes (reviewed in Refs. [55][56][57]. In Drosophila, disruption of the endocytic pathway can alter Notch localization and activity. In addition, proteolysis of the ligand-activated receptor by ␥-secretase requires HEK293T cells expressing FL-Notch1 with or without overexpressed Numb were biotinylated at 4°C and then incubated at 18°C. Total biotin-labeled FL-Notch1 (biotin-N1) at the cell surface (Tot) and the efficiency of MeSNa treatment (0Ј) was monitored as described previously. Biotin remaining at the cell surface was stripped at the indicated time points, and intracellular pools of biotin-labeled FL-Notch1 were recovered using streptavidin beads and analyzed by Western blot. Tfxn, transfection; IB, immunoblot. B, loss of Numb protein using Numb-specific siRNA had no effect on Notch1 internalization compared with C2C12 cells expressing scrambled siRNA duplexes. Representative Western blots are shown. Numb overexpression or depletion was confirmed by immunoblotting of whole cell lysates with anti-Numb. Films from three independent experiments were scanned and quantitated using an Alpha Innotech Fluorochem 8000 imaging system (see "Materials and Methods") and results are shown in graphs (right). Data are expressed relative to total (Tot) surface-biotinylated Notch1 (mean Ϯ S.D.; n ϭ 3). Significance was determined by Student's t test. Cells were then returned to 37°C to resume trafficking. One set of cells was treated a second time with MeSNa followed by lysis to examine intracellular pools of biotinylated FL-Notch1 (IC, lower panels), whereas the second set was lysed without MeSNa treatment to examine total pools of biotinylated FL-Notch1 remaining (Surf ϩ IC, upper panels). Pools of biotinylated FL-Notch1 were recovered with streptavidin beads and analyzed by Western blot using antisera specific for Notch1. Tfxn, transfection; IB, immunoblot. C and D, knockdown of Numb protein using Numb-specific RNA interference increases FL-Notch1 recycling back to the plasma membrane. C2C12 cells expressing a scrambled control siRNA duplex (C) or Numb-specific siRNA duplex (D) were analyzed as described above for intracellular and total pools of biotinylated FL-Notch1. Whole cell lysates were quantified and analyzed for Numb expression. Scram, scrambled siRNA; IC, intracellular; Surf ϩ IC, surface and intracellular. Representative Western blots are shown. Films from three independent experiments were scanned and quantitated using an Alpha Innotech Fluorochem 8000 imaging system (see "Materials and Methods"), and results are shown in graphs (right). Data are expressed relative to total (Tot) surface-biotinylated Notch1 (mean Ϯ S.D.; n ϭ 3). Significance was determined by Student's t test; **, p Ͻ 0.01;*, p Ͻ 0.05 compared with respective controls shown in A and C.
both mono-ubiquitination and clathrin-mediated endocytosis (20,58). Notch signaling is also affected by the availability of the receptor at the plasma membrane to interact with ligand, which can be influenced by both its rate of removal from the membrane and its insertion. In this study, we investigated the role of Numb in the constitutive endocytosis and intracellular trafficking of mammalian Notch1. Our data demonstrate that, although Numb has no effect on the constitutive endocytosis of Notch1 from the plasma membrane, Numb positively regulates late sorting events in the endocytic pathway such that Notch1 is trafficked to the late endosome/lysosomal pathway. Our results are consistent with a model in which Numb promotes the rerouting of Notch1 from a constitutive recycling pathway into a late endosomal compartment leading to degradation. We and others have shown that Notch1 is continuously internalized from the plasma membrane into endocytic sorting compartments where it is recycled back to the cell surface or sorted to the late endosome and degraded (53). Similar observations of constitutive internalization of Notch have been described in Drosophila (24,59). Internalization of Notch into sorting compartments functions as an important decision point for regulation of downstream signaling events (20,(23)(24)(25). Our results strongly support a role for Numb as a mediator of these trafficking decisions. In the presence of Numb, Notch may be diverted from returning to the plasma membrane where it can interact with ligand, and instead is routed through the late endosome to the lysosome where it is degraded. Numb therefore could regulate the pool of activation-competent Notch. Although this model implies that Numb regulates the steady state trafficking of Notch, upstream of activation events, we cannot exclude the possibility that Numb might also influence the trafficking of ligand-bound Notch.
In Drosophila, Numb is required for asymmetric localization of the ␣-adaptin subunit of AP-2 in dividing sensory organ precursor cells (43), and both vertebrate and Drosophila Numb bind ␣-adaptin. In mammalian cells ␣-adaptin is required for Numb localization in cortical membrane patches (60). Given these interactions with the AP-2 complex, it has been proposed that Numb regulates cell fate decisions through asymmetric segregation of the endocytic machinery, which in turn facilitates Notch receptor internalization (47). Our data suggest that Numb directs the intracellular routing of the Notch1 receptor through endocytic sorting compartments and late endosomes, and this function is dependent on the presence of both the ␣-adaptin and EH domain binding sites. This suggests that the Numb interaction with AP-2 is important for the post-endocytic trafficking of Notch1. In keeping with this notion, a role for AP-2 in the post-endocytic trafficking of cargo through a clathrin-independent pathway has recently been described (61).
Trafficking of receptors for lysosomal degradation is a well characterized mode of receptor down-regulation (reviewed in Refs. 62,63). Alternatively, membrane proteins can be routed through an endocytic sorting compartment where they undergo complex internal cycling between the recycling endosomes and late endosomes before being returned to the cell surface or shuttled into a degradative pathway. The EH domain-containing protein, EHD/RME-1, functions to posi-tively regulate recycling and promotes the exit of cargo from endocytic recycling compartment (64 -70). Previously, we reported a conserved interaction between Numb and EHD family members and demonstrated that Numb knockdown disrupted the post-endocytic trafficking of the interleukin-2 ␣ receptor (Tac) (41). A recent study in C. elegans examined the function of RME-1 and the Numb orthologue, NUM-1A, and provided evidence that NUM-1A inhibits endosomal recycling by negatively regulating RME-1 (71). This suggests that the enhanced Notch1 recycling that we observed in Numb-depleted cells may reflect the loss of EHD inhibition. Further examination of the Numb-EHD interaction and how this may affect Notch1 recycling will be important in understanding the mechanism by which Numb influences endocytic sorting.
Our observations are also congruent with the previously described function of two Itch-related E3 ligases, dNedd4 and Suppressor of Deltex, Su(Dx), that regulate endosomal sorting of Notch in the absence of ligand activation (24,25). Overexpression of these E3 ligases did not affect Notch internalization but did promote the sorting of Notch into vesicular compartments positive for Hrs and Rab7. Similarly, the C. elegans orthologue, WWP1, is important for degradation but not internalization of LIN12/Notch (23). More recently, the mammalian orthologue of Su(Dx), AIP4/Itch, was also shown to be involved in controlling post-endocytic degradation of constitutively internalized Notch1 via a process that requires ubiquitination (53). These studies suggest that ubiquitination of Notch is required for a post-endocytic sorting event leading to Notch degradation. Consistent with these observations, we found that a form of Numb that does not interact with Itch causes accumulation rather than degradation of internalized Notch1. This suggests that without the Numb-Itch interaction, Notch1 is routed into an endocytic sorting compartment but then fails to transit to the late endosome and multivesicular body, a step that is thought to require receptor ubiquitination (24,25). A recent study demonstrated the important role of ubiquitination at this late endocytic step in preventing ligand-independent activation of Notch (72).
This study provides evidence that Numb may function as an antagonist of endocytic recycling in mammalian cells, and others have demonstrated such a role in C. elegans. However, a role for Numb in regulating post-endocytic trafficking in Drosophila has not yet been demonstrated. Genetic interaction between the Drosophila EHD orthologue Past1 and the Notch pathway has recently been reported, and we previously showed that Drosophila Numb interacts with Past1 (41,73). These observations suggest that Drosophila Numb is likely to have a similar postendocytic function. In Drosophila, Numb binds to and regulates the subcellular distribution of another transmembrane protein, Sanpodo, that functions at the cell surface as an activator of Notch-dependent cell fate selection (74,75). Therefore, Drosophila Numb could influence Notch activity indirectly through regulation of Sanpodo trafficking.
The role of Numb in post-endocytic trafficking likely extends to transmembrane cargo proteins in addition to Notch1. A recent study examining the intracellular trafficking and processing of the amyloid precursor protein, a molecule that undergoes similar proteolytic processing to Notch1, showed that overexpression of Numb influenced amyloid precursor protein endosomal degradation (76). Interestingly, this effect was dependent on the presence of the alternatively spliced insert within the PTB domain of Numb (the same isoform of Numb used in this study). PC12 cells, which overexpressed forms of Numb having the PTB insert, showed decreased intracellular accumulation of amyloid precursor protein. This effect was abrogated by treatment with inhibitors of the lysosomal, degradative pathway but unaffected by inhibition of the proteosomal pathway for degradation.
Numb has also been shown to influence endocytosis of other transmembrane proteins, including ␤1-integrin, and the neuronal cell adhesion molecule L1 (44,45). These studies demonstrated effects of Numb on the intracellular accumulation of internalized molecules; however, post-endocytic events such as recycling and degradation were not assessed. Based on our observations we suggest that Numb may also influence the constitutive trafficking of ␤1-integrins and L1 toward recycling or degradation. Further studies are required to identify additional Numb-specific cargo proteins and determine the mechanism by which Numb regulates their post-endocytic trafficking.