Functional Analysis of the Numb Phosphotyrosine-binding Domain Using Site-directed Mutagenesis*

The Numb protein is involved in cell fate determination during Drosophila neural development. Numb has a protein domain homologous to the phosphotyrosine-binding domain (PTB) in the adaptor protein Shc. In Shc, this domain interacts with specific phosphotyrosine containing motifs on receptor tyrosine kinases and other signaling molecules. Residues N-terminal to the phosphotyrosine are also crucial for phosphopeptide binding to the Shc PTB domain. Several amino acid residues in Shc have been implicated by site-directed mutagenesis to be critical for Shc binding to receptor tyrosine kinases. We have generated homologous mutations in Numb to test whether, in vivo, these changes affect Numb function during Drosophila sensory organ development. Two independent amino acid changes that interfere with Shc binding to phosphotyrosine residues do not affect Numb activity in vivo. In contrast, a mutation shown to abrogate the ability of the Shc PTB domain to bind residues upstream of the phosphotyrosine virtually eliminates Numb function. Similar results were observedin vitro by examining the binding of the Numb PTB domain to proteins from Schneider S2 cells. Our data confirm the importance of the PTB domain for Numb function but strongly suggest that the Numb PTB domain is not involved in phosphotyrosine-dependent interactions.

During the development of an organism, different cell fates can be acquired through both intrinsic or extrinsic mechanisms. In an intrinsic mechanism, a certain factor in the cell is asymmetrically distributed within the mother cell, such that when the mother cell undergoes cytokinesis, the factor is distributed differentially to the two daughter cells. Conversely, in an extrinsic mechanism, the daughter cell receives signals from neighboring cells (or its own sibling cell) to acquire a certain cell fate. The Numb protein represents a key point in linking both intrinsic and extrinsic means of assigning cell fates during asymmetric divisions in the Drosophila peripheral nervous system (Refs. 1 and 2; for reviews see Refs. [3][4][5]. In Drosophila, Numb is necessary for the asymmetric division of cells in neural lineages (1, 6 -8). In the case of simple external sensory organ lineages, the sensory organ precursor undergoes two sets of divisions as follows: the first set produces two daughter cells (IIa and IIb), and the second set produces a bristle and socket cell from the IIa cell and a neuron and glial cell from the IIb cell (Fig. 1A). Numb protein has been shown to be localized to one side of the sensory organ precursor prior to division (1). After cell division it is then segregated asymmetrically to only the IIb cell. In numb mutants, the IIb daughter cell is transformed to a IIa cell fate resulting in two sets of bristle and socket cells (1). In numb mosaics, occasionally four socket cells are observed suggesting that Numb is also needed during the second division (Fig. 1B). Conversely, when Numb protein is overexpressed, the IIa cell is transformed into a IIb cell, resulting in two sets of neuron and glial cells (see Fig. 1C, balding; Ref. 1). When Numb is overexpressed after the first cell division, only the second stage of the lineage is affected, resulting in a transformation of a socket to a bristle cell (see Fig. 1D, twinning) or a glia into a neuron (1). Genetic studies have indicated that numb functions to antagonize signaling by the Notch receptor (2,9). Lack of Notch function leads to an increase in neuronal cells in the Drosophila peripheral nervous system, whereas activated Notch leads to a decrease in cells adopting the neuronal fate (2,10). A reduction in Notch signaling partially suppresses the numb mutant phenotype leading to the generation of some neuronal cells. These data suggest that numb functions to inhibit Notch signaling and thus promotes the neuronal phenotype. Furthermore, there is evidence that Notch binds directly to the Numb PTB 1 domain (2).
The phosphotyrosine binding (PTB) domain of Numb is critical for Numb function but not for the asymmetric localization of Numb (11). The PTB domain was first described in the protein, Shc, and was subsequently identified by sequence homology to be present in several other proteins including Numb (12)(13)(14)(15)(16). The PTB domain of Numb and Shc are 20% identical at the amino acid level ( Fig. 2A, see Refs. 12 and 17), suggesting that the Numb PTB domain may act similarly to the Shc PTB domain in mediating signal transduction events by binding to the NPXpY motif (18). In Shc, the PTB domain binds a ⌿XNPXpY motif on receptor tyrosine kinases, where ⌿ is a hydrophobic residue, X is any amino acid residue, N is asparagine, P is proline, and pY is phosphotyrosine. Several individual amino acid residues in Shc have been implicated by sitedirected mutagenesis to be critical for Shc binding to the * This research was supported by Public Health Service Grant NS29119 (to R. B.), Public Health Service National Research Service award, and Chemical and Hearing Senses training grant (to L. Y.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Both authors contributed equally to this work and should be considered joint co-authors.
‡ ⌿XNPXpY motif (17,19,20). These include residues involved in binding the phosphotyrosine as well as residues necessary for contacting the amino acids upstream of the phosphotyrosine in the ⌿XNPXpY peptide. An important question is whether all or only some of the PTB domains are involved in phosphotyrosinedependent interactions (21)(22)(23)(24)(25)(26). In an attempt to address whether or not the Numb PTB domain is involved in phosphotyrosine binding, we generated point mutants in the Numb PTB domain analogous to those generated in Shc (19,21). Two of these mutations affect conserved residues involved in the binding of the Shc PTB domain to phosphotyrosine. These are serine 148 and arginine 171 of Numb. Another mutation targets phenylalanine 195 of Numb. In Shc, this phenylalanine binds to the hydrophobic residue Ϫ5 and the asparagine Ϫ3 to the phosphotyrosine without directly interacting with phosphotyrosine (17). Finally, a fourth mutation, phenylalanine to leucine 149 of Numb, which results in reduced binding affinity of the Shc PTB domain, was also tested.
We have examined how these mutations affect Numb function in vivo both during Drosophila sensory organ development and mesodermal development. The phenylalanine to valine mutation (amino acid residue 195) is able to virtually eliminate the gain-of-Numb function when overexpressed in transgenic flies using the UAS-GAL4 system (27). Altering the residues in Numb homologous to those involved in Shc phosphotyrosine binding has little or no effect on the lineage transformation activity of Numb overexpression in vivo. Another mutation, phenylalanine to leucine (position 149), which causes Shc to bind with a reduced affinity in vitro, exhibits an intermediate phenotype in vivo. In binding studies using Drosophila cell culture, we detect several proteins that bind to the wild-type PTB domain and to the S148A and R171Q mutants but not to the F195V mutant of Numb. These data with site-directed mutagenesis confirm that the PTB domain is crucial for Numb function but strongly suggest that phosphotyrosine is not required for Numb PTB domain function in vivo.

DNA Constructs and Mutagenesis-Wild-type
Numb PTB domain GST fusion protein constructs were made by subcloning a polymerase chain reaction-generated fragment of numb DNA spanning the 966 -1490-base pair region of numb cDNA into the pGSTag vector (28) in the EcoRI site. PTB domain point mutants were generated by polymerase chain reaction-based site-directed mutagenesis from the wild-type pG-STag-Numb PTB domain.
The full-length numb cDNA (760 -2800 base pairs) was subcloned into the pUAST vector in the KpnI site. Site-directed mutagenesis of full-length numb was done by using Transformer Site-directed Mutagenesis kit (CLONTECH) or U.S.E. Mutagenesis kit (Amersham Pharmacia Biotech). A Myc epitope tag was attached to the C-terminal end of the Numb coding region. All constructs were sequenced using Sequenase version 2.0 (U. S. Biochemical Corp.).
Western Blot of Embryo Lysates-To assay for the expression of both wild-type and Numb protein, a Western blot was performed. The UAS-Numb flies (wild-type and mutant) were crossed to the da G32 -GAL4 driver (29). Embryos from this cross were collected from 0 to 8 h at 25°C or 0 to 16 h at 18°C. They were then washed in NaCl/Triton X (0.7% NaCl, 0.04% Triton X-100), dechorionated in 50% bleach in NaCl/Triton X, and then washed again in NaCl/Triton X, followed by a wash in PBT (0.1 M NaPO 4 , 0.3% deoxycholate, 0.5% Triton X-100). The embryos were then lysed in RIPA buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1.5 mM MgCl 2 with 1 mM EGTA, 1% sodium deoxycholate, 0.1% SDS, 10% glycerol, and 1% Triton X-100) with 1 mM phenylmethylsulfonyl fluoride, 10 mg/ml aprotinin, 10 mg/ml leupeptin, 100 mM sodium fluoride, 200 mM sodium vanadate, and 10 mM tetrasodium pyrophosphate by grinding using a Dounce homogenizer. Lysates were then subjected to electrophoresis and transferred to nitrocellulose membrane. The membrane was then blotted with anti-Numb antibody (1).
Cell Culture-Notch expressing Schneider S2 cells (30) were grown at room temperature in Schneider cell media (Life Technologies, Inc.), supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, 100 mg/ml streptomycin, and expression main-tained with 200 mg/ml hygromycin B. Notch expression was induced by heat shock at 37°C for 30 min followed by recovery at room temperature for 2 h. The cells were then lysed in lysis buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1.5 mM MgCl 2 with 1 mM EGTA, 10% glycerol, 1% Triton X-100, 10 mg/ml aprotinin, 10 mg/ml leupeptin, and 1 M phenylmethylsulfonyl fluoride). PC12 cells overexpressing TrkA receptor were stimulated with EGF or NGF for 5 min at 37°C as described (12,18). Cells were lysed as described above and incubated with GST fusion proteins.
Methionine Labeling of Schneider Cell Lysate-Wild-type Schneider cells were first grown to 90% confluency. The cells were then placed in methionine-free Dulbecco's modified Eagle's medium supplemented with 5% heat-treated fetal bovine serum. Cells were labeled for 4 h in the same media with [ 35 S]methionine (80 Ci/ml; Easytag Express Labeling Mix, Life Science Products). Cells were then rinsed with cold phosphate-buffered saline and lysed in lysis buffer with proteases and phosphatase inhibitors.
GST Fusion Protein Pull-down Assay-GST (glutathione S-transferase) fusion proteins were expressed and bound to glutathione-agarose beads using standard protocols (13). GST fusion proteins of the various Numb constructs were quantitated by SDS-PAGE, and the equal amounts were incubated with radiolabeled Schneider cell lysates for 90 min at 4°C. The beads were then washed three times with lysis buffer, boiled in 1ϫ sample buffer, and separated by SDS-PAGE. The gel was then fixed and treated with Amplify (Amersham Pharmacia Biotech) prior to drying and subsequent autoradiography. For Western blotting, the proteins were transferred to nitrocellulose and blotted with antibody directed against the intracellular domain of Notch (C17.9C6; kindly provided by S. Artavanis-Tsakonas) or anti-phosphotyrosine (12).
Creation of UAS-Numb Transgenic Flies-The UAS-Numb constructs were purified on a CsCl gradient and then coprecipitated with the p25.7wc⌬2-3 transposase helper plasmid (31) at a concentration of 500 g/ml UAS-Numb and 100 g/ml p25.7wc⌬2-3 in injection buffer (5 mM KCl, 0.1 mM NaH 2 PO 4 , pH 6.8). The DNA was then injected in w Oregon R flies. At least four independent lines were generated for each construct (see Table I). The flies were grown on standard yeast-glucose media at 25°C, unless otherwise noted.
The wild-type and mutant UAS-Numb flies were crossed to several different GAL4 driver lines that were kindly donated by several different laboratories including that of Gerhard Technau , and 71B-GAL4) exhibited a mix of twinning and bristle loss over much of the fly body, including the wing margin, to varying degrees (data not shown). The phenotype was restricted to the body bristles, with an essentially wild-type wing margin, in UAS-Numb;160-GAL4 flies. Conversely, one GAL4 driver, C96-GAL4 (32), when crossed to UAS-Numb, exhibited a moderate degree of twinning of the wing margin bristles but an essentially wild-type body bristle phenotype (see Fig.  3B). To assay the effect of wild-type and mutant Numb on mesoderm development, a double GAL4 line of twist-GAL4;24B-GAL4 (Refs. 27 and 33) was used (this line was constructed and kindly donated by Wendy Lockwood).
To assay for the relative strength of the different mutations, several independent lines of the UAS-Numb wild-type and mutant flies were crossed to 189-GAL4 flies. The number of missing macrochaetae were scored and tabulated in Table I. Rarely, a twinned macrochaetae was observed with this driver, and these were scored the same as a missing macrochaetae.
Immunohistochemical Staining-To examine the effects of the UAS-Numb constructs (wild-type and mutant) during embryonic development, the UAS-Numb flies were crossed to a da G32 -GAL4 line (29) for assaying neuronal development and twist-GAL4;24B-GAL4 for assaying mesodermal development. The embryos from this cross were then collected, dechorionated, and fixed as described previously (34). The neurons were then visualized using the 22C10 monoclonal antibody (35) at a concentration of 1:50. Cardiac and muscle cells were visualized with an anti-Evenskipped antibody (36) used at a concentration of 1:10,000. Horseradish peroxidase-conjugated goat anti-mouse IgG or anti-rabbit IgG (Bio-Rad) was used as the secondary antibody at a concentration of 1:200. Imaginal discs for immunohistochemical staining were prepared and stained as described by Patel (37). Anti-Myc antibody (9E10) was used at a 1:50 dilution to visualize transgenic Numb overexpression. Horseradish peroxidase-conjugated goat antimouse IgG (Bio-Rad) was used as the secondary antibody at a concentration of 1:200.

Numb PTB Domain Function in Vivo-
Since we wanted to determine if the Numb PTB domain functions in vivo in a fashion analogous to the Shc PTB domain, we decided to use Numb overexpression within the Drosophila peripheral nervous system as our assay. The UAS-GAL4 system (27) was used so that Numb could be induced specifically in ectodermal tissues, either during embryonic or pupal development. The full-length wild-type Numb cDNA, tagged with a Myc epitope, was subcloned into the pUAST vector (27), and several transgenic lines were generated (see Fig. 2B; Table I). As expected, many different phenotypes, corresponding to those observed by Rhyu et al. (1), were observed when the wild-type Numb was overexpressed under the control of different GAL4 drivers, presumably because the different GAL4 drivers are turning on Numb expression at different points in the neural lineages (for details about some of the GAL4 drivers tested see "Materials and Methods"). When Numb is overexpressed before the first division of the sensory organ precursor, no bristles or sockets are formed, because the corresponding cells are transformed into neurons and glial cells in a simple external sensory organ lineage (Fig. 1C, balding). However, when Numb protein is overexpressed only after the first division in the sensory organ precursor lineage, a transformation of sockets into bristles is observed (Fig. 1D, twinning). An example of the twinning phenotype can be seen in Fig. 3B, where overexpression of wild-type UAS-Numb driven by the C96-GAL4 driver (32) resulted in twinning of bristles on the wing margin. One of the most striking phenotypes observed was seen in the progeny of 189-GAL4 crossed to UAS-Numb, in which most of the bristles on the body were missing compared with wild type (see Fig. 3, A and B, and Table I). We chose to use this GAL4 driver for our assay, since any reduction in Numb activity might be expected to produce a less severe phenotype, intermediate between the extensive baldness observed with wild-type Numb overexpression and the normal bristle pattern observed in wild-type flies.
The PTB domain of Shc has been extensively characterized in vitro using site-directed mutagenesis (17,19,20). Several different amino acids have been identified as being critical for Shc binding to its receptor targets. An alignment of the human Shc PTB domain and the Drosophila Numb PTB domain is shown in Fig. 2A. The amino acids indicated by asterisks in Fig.  2A have been identified as being crucial for Shc PTB domain binding and are conserved in the Numb PTB domain. The Shc F198V mutation inhibits binding of the Shc PTB domain to a number of receptor tyrosine kinase targets (19,38,39). A homologous mutation was made in the Drosophila Numb PTB domain, F195V, within a full-length Numb construct in the pUAST vector (Fig. 2). Four transgenic lines were made from this construct and crossed to both embryonic and imaginal GAL4 driver lines (see "Materials and Methods"). The progeny from these crosses appeared almost indistinguishable from wild type (Fig. 3E, compare with Fig. 3A). There is strong evidence to suggest that this mutant Numb protein is being expressed at high levels. In Western blot analysis, a higher level of Numb protein is seen in the progeny of da G32 -GAL4 crossed to UAS-Numb relative to the amount observed in wildtype flies, suggesting that the Numb F195V mutant protein is strongly induced (Fig. 4A). Furthermore, by using immunohistochemical staining with an antibody that recognizes the Myc epitope tag on the transgenic Numb protein, strong expression of both wild-type and Numb F195V was observed in larval third instar wing discs (Fig. 4B). Finally, a very low number of duplicated bristles and/or missing bristles (3.3% macrochaetae per fly compared with 0.5% in wild type) is observed with Numb F195V overexpression (driven by certain GAL4 drivers, especially T80-GAL4, but also 1445-GAL4 and 30A-GAL4). This suggests that although the mutant Numb protein is being highly expressed, it has virtually no activity.
Two other sets of mutations, S151A and R175Q, have been shown to specifically interfere with the binding of the Shc PTB domain to phosphotyrosine residues (17,19). Homologous mutations were made in the full-length Numb protein (S148A and R171Q), which was then subcloned into the pUAST vector for the creation of transgenic lines (Fig. 2B). When these Numb mutants are overexpressed in the UAS-GAL4 system, the overexpression phenotypes are indistinguishable from that of wildtype Numb protein, exhibiting an almost complete balding phenotype (Fig. 3, C and D; Table I).
Finally, a mutation, F149L, which is analogous to the F152L mutation in Shc, was tested. Shc PTB domains with this mutation bind less well than the wild-type PTB domain but better than the F198V mutant (19). Similarly, this mutation in Numb yielded an intermediate bristle phenotype in transgenic flies. When using 189-GAL4 to drive expression, these flies exhibited a moderate degree of bristle loss ( Fig. 3F; Table I), as opposed to the very extensive bristle loss observed with wild-type and S148A and R171Q Numb overexpression (Fig. 3, B-D; Table I).
We also wanted to determine if these mutations had the same effect on Numb function during embryonic PNS development and during mesodermal development. Compared with wild-type embryos, excess neurons were observed when UAS-Numb flies were crossed to da G32 -GAL4 flies, which drives expression throughout the ectoderm (29). As in the adult, the S148A and R171Q mutations gave a phenotype similar to overexpression of wild-type Numb, whereas overexpression of the F195V mutant gave a wild-type phenotype (Fig. 5, A-C). Since Numb appears to influence cell fate decisions in mesoderm as well, 2 we wanted to see if the mutations that we were examining in neural development have a similar effect during mesoderm development. To assay for this, the UAS-Numb wild-type and mutant lines were crossed to the twist-GAL4;24B-GAL4 line (27,33), which drives Numb expression throughout the 2 M. Park, L. Yaich, and R. Bodmer, unpublished observations.
Numb PTB Domain Function in Vitro-To confirm our results in vivo, we analyzed the in vitro binding ability of the Numb PTB domain. Wild-type and mutant Numb PTB domains were generated as a GST fusion protein in bacteria. These proteins were purified on glutathione-agarose and used as an affinity matrix to bind proteins from [ 35 S]methionine-labeled Schneider S2 cell lysate. After binding, the beads were washed with lysis buffer containing 1% Triton X-100. The bound proteins were then separated by SDS-PAGE and exposed for autoradiography. Several proteins could be seen to bind to the The IIb cell is transformed into a IIa cell, which can lead to two sets of socket and hair cells. In numb mosaics, the second stage of the lineage is affected as well (as shown here), resulting in four socket cells. C, when Numb protein is overexpressed only early in the lineage, the IIa cell is transformed into a IIb cell, resulting in two sets of glia and neurons. D, when Numb is overexpressed only after the first cell division, the IIa and IIb cells form normally, but one of their progeny is transformed from a socket to a bristle and/or a glia into a neuron, respectively.

FIG. 2. Sequence alignment and UAS Numb constructs.
A, amino acid sequence alignment of Shc and Numb PTB domains. Amino acids with asterisks represent residues mutated in the UAS and GST Numb constructs. B, schematic representation of UAS Numb constructs. The full-length wild-type numb cDNA was subcloned into the pUAST vector (27). Four independent constructs (F195V, S148A, R171Q, and F149L) were made with site-directed mutations that had been shown to affect Shc PTB domain binding activity. A Myc epitope tag was attached to the C-terminal end of the Numb coding region in each of the constructs. These constructs were then used to make transgenic lines to assay Numb activity in vivo.
Numb PTB domain but not to the beads containing GST alone. Similarly, specific proteins could be detected that bound to wild-type Numb PTB domain but not to the F195V mutant (see arrows Fig. 6A). However, the S148A and R171Q mutations of the Numb PTB domain did not affect this binding. The effect of the F195V mutation to abrogate binding of the Numb PTB domain correlated closely with the ability of this mutation to impair Numb gain-of-function in vivo. In contrast, the S148A and R171Q mutations had no effect on binding in vitro or Numb function in vivo. Although we identified the binding of the Numb PTB domain to methionine-labeled proteins, none of these proteins appear to contain phosphotyrosine. In the Schneider S2 lysate we used for these experiments, we found that many of the proteins contained phosphotyrosine as indicated by phosphotyrosine immunoblotting. However, none of these tyrosine-phosphorylated proteins bound to GST alone or to the GST Numb PTB domain (Fig. 6B). Furthermore we tested the ability of the Numb PTB domain to bind tyrosine-phosphorylated proteins from growth factor-stimulated PC12 cells that overexpress the TrkA receptor (Fig. 6C, NGF receptor). Cells were stimulated with EGF or NGF leading to the tyrosine phosphorylation of many proteins in the total lysates. Lysates were incubated with GST alone, GST-Numb PTB domain, or GST-Shc PTB and bound proteins detected by antiphosphotyrosine immunoblotting. Although the activated growth factor receptors (EGF receptor and TrkA) bound to the Shc PTB domain, no tyrosine-phosphorylated proteins bound to the Numb PTB domain.
One of the proteins that bound to the Numb PTB domain in Fig. 6A had a molecular mass greater than 200 kDa. We wondered if this protein might be Notch as it had previously been suggested that the Numb PTB domain can bind Notch (2). We utilized S2 cells that express full-length Notch under the control of the heat shock promoter (30). After induction, S2 cells were lysed in the same lysis buffer used in the [ 35 S]methionine experiments in Fig. 6A. However, in this experiment Notch was FIG. 3. Wild-type and mutant Numb overexpression phenotypes. A, the bristle pattern on an Oregon R wild-type fly notum. AЈ, a close-up of the bristles on an Oregon R wild-type wing margin. B, overexpression of wild-type Numb. Wildtype and mutant UAS-Numb is driven by the 189-GAL4 line in this figure. Most of the macrochaetae and microchaetae are missing. Rarely, a few twinned bristles may be observed in these flies. BЈ, a close-up of twinned bristles resulting from overexpression of wild-type Numb driven by the C96-GAL4 driver (compare with AЈ; note the absence of sockets). C, overexpression of S148A Numb. D, overexpression of R171Q Numb. S148A and R171Q are believed to affect the binding of the PTB domain to phosphotyrosine residues. Evaluation of the strength of the phenotype is based on number of bristles missing or duplicated. Although there are some differences in the number of bristles affected (missing or duplicated) between individual flies (likely due to slight variations in the level and/or timing of GAL4 expression from cell to cell; see differences between thoraxes in B, C, and D), overall the phenotypes of S148A Numb and of R171Q Numb overexpression are virtually indistinguishable from wild-type Numb overexpression, and the average degree of cell transformations is comparable. To verify this further, several independent lines were assayed and scored for each mutant Numb overexpression experiment (see Table I). E, overexpression of F195V Numb. The homologous mutation in the Shc PTB domain abolishes binding to many receptor tyrosine kinase targets. These flies look indistinguishable from wild type (A), despite evidence that the mutant Numb protein is being expressed (see Fig. 4). All the macrochaetae and microchaetae are present in their normal numbers and locations. F, overexpression of F149L Numb. This mutation results in a reduced binding affinity for the Shc PTB domain. Likewise, when the homologous mutation is tested in this Numb overexpression assay, an intermediate phenotype is observed. Many macrochaetae and microchaetae are missing but to a much lesser extent than observed with wild-type Numb overexpression (see also  Table I). overexpressed and analyzed by a sensitive immunoblotting technique using anti-Notch antibodies. Notch overexpression after induction was easily detected by immunoblotting cell lysates with anti-Notch antibodies (Fig. 6D, lanes 1 and 2). To determine if the Numb PTB domain could bind Notch, we immobilized the Numb PTB domain on glutathione beads and added lysates from cells with and without induction of Notch overexpression. We used 10 times the amount of lysates used for immunoblotting in the 1st two lanes of Fig. 6D with an excess of Numb PTB domain on glutathione beads. Despite  (35). The dorsal cluster is indicated by the parenthesis, and the lateral chordotonal cluster is highlighted with the arrowhead. B, overexpression of the S148A UAS-Numb mutant driven by da G32 -GAL4. This phenotype is indistinguishable from overexpression of wild-type Numb (data not shown). There are many more neurons than observed in the wild-type embryo, as shown here for the lateral chordotonal cluster (arrowhead, compare with A) and the dorsal cluster (bracket, compare with A). C, overexpression of the F195V UAS-Numb mutant driven by da G32 -GAL4. The phenotype of this mutant is essentially wild-type (a dorsal cluster is bracketed; compare with A). D, a wild-type embryo stained with anti-Eve (36) showing a double row of strongly Eve-expressing pericardial cells (arrowhead) at the dorsal midline flanked by segmentally repeated muscles with weaker Eve expression (arrow points to the nuclei of a muscle syncytium). E, overexpression of R171Q UAS-Numb driven by twist-GAL4;24B-GAL4 and stained with anti-Eve. This phenotype is indistinguishable from overexpression of wild-type Numb. 2 Most of the Eve-expressing pericardial cells are missing (although a few are still present, see arrowheads), but the Eve-expressing dorsal muscles are still present. F, overexpression of F195V UAS-Numb driven by twist-GAL4;24B-GAL4 and stained with anti-Eve. The pattern of Eve-stained pericardial and dorsal muscle cells is indistinguishable from wild-type (compare with D).
using this large amount of lysate that contained overexpressed Notch, we were unable to detect Notch binding by the Numb PTB domain. This suggests that Notch may not be a high affinity binding partner for the Numb PTB domain. DISCUSSION The PTB domain of Shc has been shown to be critical for binding to a number of receptor tyrosine kinases. This domain is well conserved in the Drosophila Numb protein (12,17), suggesting that there may also be some conservation of biochemical activity. A number of amino acid residues have been identified in Shc as being critical for its ability to bind to NPXpY motifs (17,19,20). Our data demonstrate that the mutation F195V severely impairs the function of the Numb PTB domain in vitro and in vivo. The phenylalanine mutated in this experiment is conserved in almost all members of the PTB domain family (19). These results confirm the utility of using this mutation to probe PTB domain function. Although our data are the first to use point mutants to disrupt Numb PTB domain function, the work of others has also pointed to a crucial role for this domain in Numb function. Frise et al. (11) showed that deletion of the PTB domain impaired Numb function in vivo, and Verdi et al. (23) showed that overexpression of rat Numb PTB domain alone affected neural development in neural crest derived MONC-1 cells.
Numb has been shown to be genetically upstream of Notch and is proposed to inhibit Notch signaling (2). In addition, yeast two-hybrid studies and protein binding assays show that Notch and Numb are capable of a direct physical interaction and that this interaction is probably mediated by the PTB domain of Numb (2). However, in our studies we were not able to detect high affinity interactions between Numb and Notch under conditions where we could detect the binding of several labeled proteins to the Numb PTB domain. Further work will be necessary to determine if the genetic interaction between Numb and Notch rely on direct interactions or occurs via secondary proteins.
A central question that this study helps address is whether phosphotyrosine is required for PTB domain interactions. Original studies of the Shc and insulin receptor substrate-1 PTB domains showed that phosphotyrosine is crucial for the interactions of these PTB domains and their binding part-FIG. 6. In vitro binding of the Numb PTB domain to proteins from Schneider S2 and PC12 cells. A, Schneider cells were metabolically labeled with [ 35 S]methionine before being subjected to GST pull-down assay with various GST fusion protein constructs as indicated. Arrows indicating protein bands that are found only in coprecipitation with the Numb wild-type, SA, and RQ constructs but not with the Numb FV. B, Western blot assay with anti-phosphotyrosine antibody. GST Numb PTB domain or GST alone were immobilized on beads and incubated with lysates from Schneider cells. The precipitates were separated by SDS-PAGE, transferred to nitrocellulose, and blotted against anti-phosphotyrosine antibody. C, PC12 cells overexpressing TrkA (18) were either left unstimulated (O) or stimulated with EGF (E), 200 ng/ml, or NGF (N), 50 ng/ml, for 5 min at 37°C and then the cells lysed as described (12,18). Lysates were then incubated with GST fusion proteins immobilized on beads. The precipitates were washed and together with total lysates were separated by SDS-PAGE, transferred to nitrocellulose, and blotted with anti-phosphotyrosine antibody. The strong bands in the Numb PTB lanes represent the Numb GST fusion protein. Arrows show the EGF receptor (EGFR) or the NGF receptor (TrkA) binding to the Shc PTB domain. D, Schneider cells bearing an inducible Notch gene (30) were grown to 90% confluency before heat shock induction. The lysates were than incubated with Numb PTB GST fusion protein. Precipitates were separated by SDS-PAGE, transferred to nitrocellulose, and blotted for Notch. The first two lanes are total lysates and the next two lanes are Numb PTB GST fusion protein precipitates.
ners. Affinity between these PTB domains and their phosphotyrosine containing binding partners falls to undetectable levels when phosphotyrosine is omitted (40). Residues crucial for phosphotyrosine binding by the Shc PTB domain are conserved in several members of the PTB domain family including Numb (17). This has led to the suggestion that many members of this family might be involved in phosphotyrosine-dependent interactions. However, mutation of the conserved residues in Numb, which are involved in phosphotyrosine recognition by Shc, had no significant effect on Numb activity in vivo. Furthermore, we could detect binding of the Numb PTB domain independent of phosphotyrosine in vitro (Fig. 6). This finding is supported by earlier data with the X11 and FE65 PTB domains that demonstrated binding to amyloid precursor protein independently of tyrosine phosphorylation (21,26). In the case of X11 this binding occurs to a ⌿XNPXY motif on amyloid precursor protein (21). In agreement, recent studies have identified a ⌿XNPXY containing protein that binds to the Numb PTB domain independently of tyrosine phosphorylation (41). 3 This indicates that the Numb PTB domain can also bind a non-phosphorylated NPXY motif. This speculation is supported by our finding that the F195V mutant of the Numb PTB domain eliminates Numb function. This phenylalanine residue in the Shc PTB domain is crucial for interactions with upstream elements of the ⌿XNPXPY motif and might serve a similar role in Numb. In contrast, however, Pawson and co-workers (25) have suggested that both phosphotyrosine-and non-phosphotyrosine-dependent interactions might occur between the Numb PTB domain and target peptides. By using in vitro peptide selection these authors identified the motif YIGPY⌿ as a target for the Numb PTB domain. This peptide can bind with slightly higher affinity when the second tyrosine is phosphorylated. These studies, however, did not identify a target protein for Numb that might contain such a motif. Further evidence identifying the exact protein involved in Numb function will clarify these issues. The powerful genetic and biochemical studies now being performed on Numb should allow the identification of the physiological Numb PTB domain partner(s) and their mode of interaction. Our data predict that this interaction will be crucial for Numb function, but independent of phosphotyrosine binding.