The Signaling Adapter FRS-2 Competes with Shc for Binding to the Nerve Growth Factor Receptor TrkA

We have isolated a human cDNA for the signaling adapter molecule FRS-2/suc1-associated neurotrophic factor target and shown that it is tyrosine-phosphorylated in response to nerve growth factor (NGF) stimulation. Importantly, we demonstrate that the phosphotyrosine binding domain of FRS-2 directly binds the Trk receptors at the same phosphotyrosine residue that binds the signaling adapter Shc, suggesting a model in which competitive binding between FRS-2 and Shc regulates differentiation versusproliferation. Consistent with this model, FRS-2 binds Grb-2, Crk, the SH2 domain containing tyrosine phosphatase SH-PTP-2, the cyclin-dependent kinase substrate p13 suc1 , and the Src homology 3 (SH3) domain of Src, providing a functional link between TrkA, cell cycle, and multiple NGF signaling effectors. Importantly, overexpression of FRS-2 in cells expressing an NGF nonresponsive TrkA receptor mutant reconstitutes the ability of NGF to stop cell cycle progression and to stimulate neuronal differentiation.

The molecular mechanisms regulating proliferation versus differentiation in the developing nervous system depend, in part, on the availability of the neurotrophins (nerve growth factor (NGF), 1 brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3), and neurotrophin-4/5 (NT-4/5)) (1). Increasingly, it is apparent that the neurotrophins not only act as target-derived survival factors but can also regulate mitotic activity and promote differentiation of neuronal progenitors in both the peripheral (2,3) and central (4 -6) nervous systems. The biological effects of the neurotrophins are mediated by 2 classes of cell surface receptors: a low affinity p75 NTR receptor and a high affinity tyrosine kinase Trk receptor. As a low affinity receptor, p75 NTR modulates ligand binding (7,8) and some signaling pathways stimulated by TrkA (9,10) and, as a member of the Fas antigen family, stimulates sphingomyelin hydrolysis and ceramide synthesis leading to apoptosis in some cellular contexts (11)(12)(13).
In contrast, the Trk receptors are essential to proliferation, differentiation, and cell survival (14). TrkA is specifically activated by NGF, whereas TrkB and TrkC are the primary receptors for BDNF and NT-3, respectively (15,16). Upon stimulation, the Trk receptors activate numerous intracellular signaling molecules, including the Shc adapter proteins, Grb-2/Sos, Ras, Map kinase (MAPK), phosphatidylinositol 3-kinase, Src, phospholipase C␥-1, SH-PTP-1, SH-PTP-2, and the suc1associated neurotrophic factor target protein (SNT) (14). Phosphorylation of the Shc adapter proteins, and the concomitant recruitment of Grb-2/Sos, results in the Ras-dependent transient activation of MAPK (17), which is correlated with mitogenic and proliferative cell signaling. In contrast, phosphorylation of SNT and prolonged activation of MAPK correlate with neurotrophin-dependent cell cycle arrest and differentiation (18,19). Interestingly, prolonged activation of MAPK has recently been shown to be regulated by a parallel pathway involving Crk/C3G-dependent activation of a Ras-like GTPase, Rap-1 (17).
SNT has been considered a likely candidate to regulate a decision between cell cycle progression and cell cycle arrest based on its tyrosine phosphorylation by factors that induce differentiation (NGF, BDNF, NT-3, and fibroblast growth factor (FGF)) and by its ability to bind to the cyclin-dependent kinase substrate p13 suc1 . The role of SNT in neurotrophin signaling is underscored by the observation that two independent Trk receptor mutants that do not affect either mitogenic signaling and/or cell survival responses, do selectively affect the tyrosine phosphorylation of SNT and NGF-dependent cell cycle arrest/neuronal differentiation (20,21). The identity of the SNT proteins(s) has remained elusive until recently, when a murine FGFR signaling molecule, mFRS-2, was cloned and found to share properties comparable to SNT (22).
To better understand the role of FRS-2/SNT in neurotrophin signaling and cell cycle arrest, we have cloned and characterized human FRS-2. Amino acid sequence comparisons between murine and human FRS-2 indicate that FRS-2 has been evolutionarily conserved between mouse and human (96%). FRS-2 contains an amino-terminal myristylation site to facilitate tight association with the membrane, an amino-terminal phosphotyrosine binding (PTB) domain, four putative binding sites for Grb-2, two putative binding sites for the SH2 domaincontaining tyrosine phosphatase SH-PTP-2, and numerous potential binding sites (PXXP) for SH3 domain-containing proteins. We demonstrate that FRS-2 binds p13 suc1 and the SH3 domain of Src constitutively but shows NGF-dependent tyrosine phosphorylation and association with the SH2 domains of Grb2, Crk, and SH-PTP-2. Importantly, unlike signaling by the FGFR (22,23), we show that FRS-2 binds directly to neurotrophin-activated Trk receptors in a phosphotyrosine-dependent manner at a site that also regulates the binding of Shc. In vitro binding assays indicate that Shc and FRS-2 compete for binding to TrkA, suggesting a model in which competition between signaling proteins may regulate neurotrophin-dependent proliferation and/or differentiation. The role of FRS-2 expression in regulating cell cycle arrest and neuronal differentiation is further supported by the observation that FRS-2 overexpression in nnr5 cells expressing a differentiation minus TrkA receptor mutant (TrkAS3 cells) (21) reconstitutes NGF-dependent cell cycle arrest and neuronal differentiation.

MATERIALS AND METHODS
Antibodies and Growth Factors-Anti-phosphotyrosine antibody (PY20) and the horseradish peroxidase (HRP)-conjugated derivative of PY20 (RC20) were from Transduction Laboratories. Hybridoma 12CA5 cells (anti-HA) (24) and 9E10 (anti-c-Myc) (25) were grown as ascites tumors in 4 -6-week-old female Balb/c mice. Rabbit anti-SH-PTP-2 antibody was from Santa Cruz. HRP-goat anti-rabbit and HRP-goat anti-mouse secondary antibodies were from The Jackson Laboratory. p13 suc1 agarose was from Upstate Biotechnology or was the gift of D. Litchfield (University of Western Ontario). ␤-NGF was from Harlan Bioproducts for Science. BDNF and NT-3 were the gifts of A. A. Welcher (Amgen Inc.).
Cell Lines-The nnr5 cell line expressing hemagglutinin (HA)-tagged rat TrkA (B5 cells) and TrkAS3 mutant receptors (⌬ 493 IMENP 497 ; S3A3) have been previously described (21). B5 and S3A3 cells express approximately 100-fold higher levels of TrkA than PC12 cells. B5 cells were transfected with FRS-2/c-Myc by electroporation (26), and 3-4 days later, cells were stimulated with 100 ng/ml ␤-NGF (2-5 min) as described below. Untreated cells and ␤-NGF-treated cells (50 ng/ml, 5 min) were lysed in Nonidet P-40 lysis buffer, and equivalent amounts of clarified lysates were immunoprecipitated with 5 g of 9E10. Samples were analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting with anti-Tyr(P) (RC20) as described below. TrkAS3A3 cells were transfected with internal ribosome entry site (pIRES)/enhanced green fluorescent protein (EGFP)-FRS-2/Myc and cells were selected in 0.05 mg/ml G418 and 0.2 mg/ml hygromycin (Boehringer Mannheim) for 3-4 weeks. Clones positive for EGFP expression were identified by epifluorescence, picked, and assayed for FRS-2 expression by Western blotting as described below. Briefly, naive and 50 ng/ml ␤-NGF-treated cells (5 min) were lysed in Nonidet P-40 lysis buffer, and equivalent amounts of clarified lysates were immunoprecipitated with 5 g of 9E10. Samples were analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting with anti-Tyr(P) (RC20) as described below.
FRS-2 cDNA Cloning, Baculovirus, and Constructs-Using a degenerate sequence based on the published mouse FRS-2 protein sequence (22), we isolated a human genomic fragment from GenBank TM and designed a 5Ј primer (5Ј GAA GAA GCC ATG GGT AGC TGT TGT3Ј) and a 3Ј primer (5Ј TGC TGT CGA CGC TCA CAT GGG CAG ATC 3Ј). A 1.5-kb polymerase chain reaction (PCR) product was generated, subcloned into pCr2.1 (Invitrogen), and sequenced. hFRS-2 (residues 1-503) was cloned into pcDNA3.1/Myc-His (Invitrogen) inserting a human c-Myc tag at the carboxyl terminus. The rat TrkA K547A inactive kinase receptor was generated by PCR overlap mutagenesis and sequenced. HA-tagged rat TrkB and TrkC were generated by PCR overlap as described previously for TrkA (26). The hemagglutinin epitope was inserted just 3Ј to the signal sequence on rat TrkB and rat TrkC. HA-TrkB, HA-TrkC, and HA-TrkA K547A-expressing baculoviruses were generated as described (27). Glutathione S-transferase (GST)-hFRS-2 was generated by subcloning hFRS-2 as a 1.5-kb EcoRI/SalI fragment from pCr2.1/FRS-2 into pGEX-4T-2 (Amersham Pharmacia Biotech). The GST/hFRS-2(PTB) construct contains the amino-terminal 240 residues of hFRS-2. All other GST fusion constructs were the gifts of numerous investigators: Shc (J. McGlade, The Hospital for Sick Children, Toronto, Ontario, Canada); Grb-2, Crk, Ras-Gap, PLC␥-1, and v-Src (T. Pawson, Lunenfeld Research Institute, Toronto, Ontario, Canada); and Abl and Nck (P. P. DiFiore, European Institute of Oncology, Milan, Italy). GST fusion proteins were purified according to standard procedures. The FRS-2/c-Myc cDNA was also cloned into pIRES-EGFP (CLONTECH) in which transfected cells can be easily identified based on low levels of enhanced green fluorescent protein expression from an internal ribosomal initiation site.
Northern Blot-A rat multiple tissue Northern blot containing approximately 2 g of poly(A) ϩ RNA per lane (CLONTECH) was probed with the hFRS-2 cDNA that was radiolabeled with [␣-32 P]dCTP (Amersham Pharmacia Biotech) by random priming to a specific activity greater than 1 ϫ 10 9 cpm/g. The blot was washed to a final stringency of 42°C, 1ϫ SSC, 0.1% SDS. The blot was stripped and reprobed with actin (55°C, 0.1ϫ SSC). Densitometry was performed with a Bio-Rad imaging densitometer (Model GS-700) and Multi-Analyst software.
In Vitro Binding Assays-HA-tagged Trk receptors were expressed in baculovirus-infected insect High Five cells (21). Cells were stimulated (100 ng/ml ␤-NGF, BDNF, or NT-3) and lysed in Nonidet P-40 buffer as described (21). Protein concentrations of clarified lysates were determined with the Bio-Rad D C kit. The level of expressed Trk receptor was determined by titrating lysates by SDS-polyacrylamide gel electrophoresis and blotting with anti-HA. Lysates containing equal amounts of expressed Trk receptors or lysates prepared from hFRS-2-transfected B5 cells were used in binding assays containing 10 g of Sepharosebound GST proteins. Samples were assayed by Western blotting with RC20 (1:2500), anti-HA (1 g/ml), anti-c-Myc (2 g/ml), or anti-SH-PTP-2 (1:2000). HRP-goat anti-rabbit and HRP-goat anti-mouse were used at a dilution of 1:10,000. The chemiluminscence detection kit was from NEN Life Science Products.
For competition experiments, GST-hFRS-2(PTB)-and GST-PLC␥-1(SH2)-bound Sepharose was digested with 10 units of thrombin for 30 min at room temperature. The digests were washed twice with PBS containing protease inhibitors and 1 mM EDTA and pooled, and the eluate was used directly.
Neurite Response and 5-Bromo-2-deoxyuridine (BrdUrd) Assays-The neurite response assay was performed on poly D-lysine-coated 24-well clusters. Fresh neurotrophin was added every other day. On day 5, cells were assayed for changes in proliferation by measuring the incorporation of the thymidine analogue BrdUrd as described previously (21). The percentage of BrdUrd-positive cells was scored from 10 -12 independent frames. The S3-FRS-2 sample represents as average of two clones expressing FRS-2 (clones 34 and 40) and is thus an average of 20 independent frames.

Isolation of a Human FRS-2 cDNA and Northern Blot
Analysis-An hFRS-2 cDNA (1.5 kb) was isolated from fetal brain cDNA by PCR techniques with degenerate primers based on the protein sequence of mFRS-2 (22). The cDNA sequence predicts a protein with a primary sequence of 508 amino acids and a molecular mass of approximately 57 kDa. Amino acid comparison between hFRS-2 and mFRS-2 ( Fig. 1A) indicates that FRS-2 has been evolutionarily conserved between mouse and human (96% identity). hFRS-2, like mFRS-2, contains an amino-terminal myristylation signal (MGXXXS) and PTB domain, 15 potential sites of tyrosine phosphorylation, 4 Grb-2 binding sites, 2 potential binding sites for the tyrosine phosphatase SH-PTP-2, and putative binding sites for SH3 domaincontaining proteins (PXXP). Northern blot analysis of FRS-2 expression indicates low levels of two transcripts of approximately 7.2 and 7.5 kb in all tissues examined with slightly higher levels (2-5-fold) detected in brain, testes, and lung (Fig.  1B). The ratio of the two transcripts is comparable between tissues with slightly higher levels of the upper transcript present in lung. The difference between our 1.5-kb cDNA and the native transcripts represent untranslated regions of approximately 1.5 and 4.2 kb, respectively (Fig. 1B).
FRS-2 Is Tyrosine-phosphorylated by NGF and Binds p13 suc1 Constitutively-To facilitate transfection studies, hFRS-2 was tagged at the carboxyl terminus (c-Myc) without interfering with NGF-dependent tyrosine phosphorylation and/or its sig-naling properties. As shown in Fig. 2A, following transfection into NGF-responsive cells, hFRS-2/c-Myc (80 -90 kDa) was tyrosine-phosphorylated in response to NGF stimulation and was immunoprecipitated with the anti-Myc antibody (9E10) (top panel, compare lanes 2 and 4). Because hFRS-2/c-Myc predicts a protein of approximately 57 kDa, this suggests that FRS-2 is posttranslationally modified by mechanisms that include myristylation (22) and both serine/threonine and tyrosine phosphorylation. Similar to endogenous FRS-2, FRS-2/c-Myc bound p13 suc1 , although the transfected gene product migrated more slowly than the endogenous protein consistent with the addition of the c-Myc epitope (approximately 3 kDa) to the carboxyl terminus ( Fig. 2A, top panel, compare lanes 6 and 8). Stripping and re-probing the blot with anti-Myc antibodies indicated that hFRS-2/Myc can bind p13 suc1 constitutively and that the binding is not regulated by tyrosine phosphorylation (Fig. 2A, bottom panel, lanes 7 and 8). Moreover, more than one form of FRS-2 was consistently detected, indicating multiple forms of posttranslational modifications. Typically, only the highest molecular mass form of FRS-2 was tyrosine-phosphorylated in response to NGF.
From the primary sequence, it is also predicted that SH3 domain-containing proteins can interact with FRS-2 and facilitate alternate signaling pathways. To this end, we screened a panel of SH3 domains (Grb-2, Nck, Crk, Ras-Gap, c-Abl, PLC␥-1, and v-Src), as well as full-length Crk, for interaction with hFRS-2. As shown in Fig. 2D, the only SH3 domainmediated interaction observed was with v-Src. As predicted for SH3 domain interactions, the v-Src interaction occurred constitutively and was independent of NGF stimulation (lanes 19 and 20). The constitutive FRS-2/Src interaction is not surprising and may help explain the previous observation that v-Src expression in PC12 cells generates NGF-independent neurite outgrowth (34). Full-length Crk also interacts with hFRS-2, but because the binding is not SH3 domain-mediated (lanes 11,12), and the interaction is NGF-dependent, it likely involves the Crk SH2 domain.
FRS-2 Directly Binds the Neurotrophin Trk Receptors-As described above, hFRS-2 is modular in nature and contains an aminoterminal PTB domain that in many molecules mediates proteinprotein interactions involving phosphotyrosine. Whereas FRS-2 does not readily co-immunoprecipitate with either the FGFR (22) or TrkA (data not shown), FRS-2 has been shown to interact with the FGFR in vitro and in yeast (23). These observations suggest that FRS-2/receptor interactions are labile and/or weak in cells but that direct binding can be observed by alternative methods. These observations are reminiscent of weak interactions between the insulin receptor ␤-chain and its major signaling component IRS-1 in cells relative to readily detectable interactions detected in yeast (35).
Accordingly, we used in vitro binding assays and yeast interaction-trap assays to determine whether hFRS-2 could interact with TrkA and if so, where the site of interaction might be. HA-tagged TrkA, TrkB and TrkC were expressed in baculovirus-infected insect cells and lysates containing equivalent amounts of expressed Trk protein were used in in vitro binding experiments with purified GST fusion proteins. As shown in Fig. 3A, GST-hFRS-2(PTB), like the Shc(PTB) domain and the PLC␥-1(SH2) domain, effectively precipitates TrkA as well as TrkB and TrkC, indicating that hFRS-2 is a signaling component common to neurotrophin signaling by NGF, BDNF and NT-3. To assay direct interaction between TrkA and hFRS-2 in vivo, we utilized the yeast two-hybrid assay. Specifically, hFRS-2 was expressed as a fusion protein with the DNA binding domain of Gal 4 (pAS2-FRS-2; Trpϩ) and cells were transfected and tested for interaction with either p13 suc1 (pGAD-p13; Leuϩ) or the intracellular domain of either kinase-active TRKA or kinase-inactive TrkAK547A (pGAD-TrkA; Leu ϩ) in a standard two-hybrid assay (36). Untransfected yeast, yeast expressing pAS-FRS-2, and yeast co-expressing pAS and pGAD vectors were plated on nonselective medium (plate 1), sc me-dium minus Trp (plate 2), sc minus Leu and Trp (plate 3), and HALT medium containing 20 mM 3-amino triazole to assay two-hybrid interaction (plate 4). As shown in Fig. 3B, only functional Gal 4 transcription factor activity, and thus growth under selective conditions (HALT medium), was detected in yeast co-expressing hFRS-2 and either p13 suc1 or kinase-active TrkA (lanes 3 and 4 on plate 4). The interaction between hFRS-2 and p13 suc1 was constitutive, indicating that interaction occurred independently of hFRS-2 tyrosine phosphorylation and is consistent with the constitutive interaction observed in the transfection studies ( Fig. 2A). In contrast, hFRS-2-TrkA interaction in the yeast two-hybrid assay is tyrosine phosphorylation-dependent. Dimerization of the Gal 4 transcription factor is sufficient to stimulate the Trk kinase activity and to assay activity-dependent interactions (data not shown). Accordingly, despite a high level of TrkA ICD expression in yeast, the kinase inactive TrkA (K547A) does not interact with hFRS-2 (Fig. 3B, plate 4, lane 5).
To determine whether the FRS-2/TrkA interaction is also phosphorylation-dependent in the in vitro binding assays, baculoviruses expressing wild type TrkA and the kinase inactive TrkA were assayed for binding to GST-FRS-2(PTB) in comparison with GST-Shc(PTB) as a positive control. Shc interactions with TrkA have previously been shown to be phosphotyrosine-dependent (37). As shown in Fig. 4A, the TrkAK547A receptor expresses high levels of receptor (anti-HA blot) that is not catalytically active (anti-Tyr(P) blot). In vitro binding assays with purified GST fusion proteins indicated that the PTB domain of FRS-2 interacts with the kinase active TrkA, as does GST-Shc(PTB), and not the catalytically inactive form (Fig. 4B). Thus, as we observed in yeast (Fig. 3), FRS-2 interaction with TrkA is phosphotyrosine-dependent.
FRS-2 Binds TrkA at the Shc Binding Site, Tyr 499 -Thus far, we have demonstrated that hFRS-2 interacts with the Trk receptors through an amino-terminal PTB domain in a tyrosine phosphorylation-dependent manner. To map the binding site(s) on TrkA for hFRS-2(PTB) interaction, we took advantage of a series of rat TrkA receptor mutants that selectively affect specific signaling pathways. In this respect, NGF induces receptor dimerization and tyrosine phosphorylation at five intracellular tyrosine residues in rat TrkA (Tyr 499 , Tyr 679 , Tyr 683 , Tyr 684 , and Tyr 794 ). SHC and PLC␥-1 interact at phosphotyrosine residues Tyr 499 and Tyr 794 , respectively (Tyr 490 and Tyr 785 in human TrkA) through intrinsic PTB and SH2 domains. A total of four independent TrkA signaling mutants have been previously described and were used in the present study (see Table  I and Fig. 5A). Trk receptors were HA-tagged at the amino terminus and expressed in baculovirus-infected insect cells. Lysates containing comparable levels of expressed Trk receptors were used in in vitro binding assays with purified GST fusion proteins. As shown in Fig. 5B, GST fusion proteins corresponding to Shc(PTB), PLC␥-1(SH2), and hFRS-2(PTB) effectively precipitate wild type TrkA to levels significantly greater than GST alone. As predicted, the TrkAS9 mutant showed a loss of PLC␥-1 binding (Fig. 5B, lane 20) but retained both Shc(PTB) and hFRS-2(PTB) domain binding, indicating that tyrosine 794 is not involved in hFRS-2-TrkA binding. Analysis of the KFG minus TrkAS17 mutant indicates that it retained PLC␥-1 and Shc(PTB) binding but showed a reduction in hFRS-2(PTB) domain binding. This suggests that the KFG deletion reduced hFRS-2 binding but did not completely abolish the interaction. In contrast, both the TrkAS3 deletion mutant (⌬ 493 IMENP 497 ) and the Shc binding site mutant (Y499F) retained PLC␥-1 binding but showed a complete loss of both Shc(PTB) and hFRS-2(PTB) binding. Because Tyr 499 is still phosphorylated in the TrkAS3 mutant (21), these data indicate that both tyrosine 499 and the amino acids 5Ј of Tyr 499 are critical for hFRS-2(PTB)/TrkA binding and are consistent with the observation that residues 5Ј of the phosphotyrosine residue are also critical for high affinity Trk-Shc PTB domain binding (38). Importantly, we also found that a low, but detectable, amount of TrkA-FRS-2 binding could be recovered when very high concentrations of either TrkAS3 or TrkAS8 were used (data not shown), indicating that overexpression of either TrkA mutant can facilitate some protein-protein interaction.
FRS-2 and Shc Compete for Binding to TrkA at Tyr 499 -The observation that both Shc and hFRS-2 bind to the same phosphotyrosine residue on TrkA suggests a model in which Shc and FRS-2 may compete for binding to TrkA in vivo). Because Shc/Grb-2/Ras/Map kinase activation facilitates mitogenic signaling (39) and a decrease in FRS-2 phosphorylation correlates with an inability of TrkA to support NGF-dependent neurite outgrowth (20,21), this suggests that competition between Shc and FRS-2 for binding to Tyr 499 may, in part, regulate a cellular switch between cell cycle progression/mitogenesis and cell cycle arrest/differentiation.
To this end, competition between Shc and hFRS-2 binding to TrkA was assayed in vitro. GST-Shc (5 g) was used to precipitate baculovirus expressed HA-TrkA in the presence of increasing concentrations of soluble hFRS-2(PTB). As shown in Fig. 5C, hFRS-2 competes for binding to TrkA with an EC 50 requiring a 2-5-fold excess (between 10 and 25 g). Complete competition was obtained with a 20-fold excess (100 g). In comparison, increasing concentrations of the SH2 domain of PLC␥-1 or bovine serum albumin (up to 100 g) did not compete, indicating that the hFRS-2 competition both is specific and involves tyrosine 499.

FRS-2 Overexpression Reconstitutes NGF-dependent Neurite Outgrowth by a Differentiation Minus
TrkA Receptor Mutant-As shown above, FRS-2 showed a dramatic reduction in binding to both the Y499A TrkAS8 mutant and the ⌬ 493 IMENP 497 TrkAS3 deletion mutant, consistent with the decrease in stoichiometry of FRS-2/SNT tyrosine phosphorylation observed in TrkAS3 expressing nnr5 cells (21). To determine whether the reduced FRS-2/SNT tyrosine phosphorylation and receptor binding ability in TrkAS3 expressing nnr5 cells underscores the NGF nonresponsive phenotype, FRS-2 was ectopically expressed in TrkAS3A3 cells (21) on a vector that co-expresses EGFP from an internal ribosome entry site. Cells expressing EGFP were selected and assayed for NGF-dependent tyrosine phosphorylation of FRS-2, as well as an NGFdependent decrease in proliferation and a concomitant increase in neuronal differentiation. As shown in Fig. 6A, two independent clonal lines demonstrate high levels of NGF-induced tyrosine-phosphorylated FRS-2/Myc in comparison to either the parental S3A3 cells or cells expressing EGFP alone. To determine whether ectopic expression of FRS-2/Myc can stop cell cycle progression and facilitate neuronal differentiation, cells were grown for 5 days in the absence or presence of NGF and scored for neurite outgrowth. As shown in Fig. 6B, S3-FRS-2 expressing cells (clone 40 is shown, but a similar response was also found in clone 34; data not shown) demonstrated significant neurite outgrowth 5 days poststimulation with 100 ng/ml NGF (approximately 80 -90% of the culture) relative to the parental S3A3 cells or cells expressing EGFP (approximately 2%). Thus, FRS-2 overexpression in S3 cells reconstitutes NGFdependent differentiation that is morphologically similar to wild type TrkA-expressing nnr5 cells (B5). To address whether the rescue of process outgrowth was accompanied by a corresponding decrease in the mitotic index, S3A3 cells, S3 cells expressing FRS-2 or EGFP, and wild type TrkA-expressing B5 cells were assayed for changes in BrdUrd incorporation. As shown in Fig. 6C, B5 cells showed an NGF-dependent decrease in BrdUrd incorporation of approximately 70%, whereas the levels of BrdUrd incorporation in S3A3 cells and cells expressing EGFP were unaffected by NGF. In contrast, S3 cells coexpressing FRS-2/Myc showed an NGF-dependent decrease in BrdUrd incorporation of approximately 50%. DISCUSSION In summary, we have cloned the human FRS-2 adapter protein and begun to characterize its role in NGF-dependent signaling through TrkA. Although TrkA can stimulate both a mitogenic response and differentiation and cessation of cell cycle progression in nonneuronal and neuronal cells, respectively, the molecular mechanism(s) underlying this differential response is not well understood. Here, we provide evidence of a novel mechanism by which competitive binding between FRS-2 and Shc may regulate a cellular switch between cell cycle progression and cell cycle arrest/differentiation. Moreover, we have begun to address the mechanisms by which FRS-2 may regulate these processes. hFRS-2 is tyrosine-phosphorylated in response to NGF-stimulation and binds directly to TrkA through its amino-terminal PTB domain in a phosphotyrosine-  (37,52), and S9 (Y794F) (37). dependent manner. hFRS-2 also binds to neurotrophin-stimulated TrkB and TrkC receptors, indicating that it is probably involved in an analogous role in BDNF and NT-3 signaling as well. Recently, the PTB domain of FRS-2 has also been shown to bind the juxtamembrane region of the FGFR (23) by both yeast two-hybrid assays and in vitro binding assays using recombinant fusion proteins. In these assays, FRS-2 binding to the FGFR is independent of kinase activation and has been localized to a juxtamembrane region that lacks both asparagine and tyrosine residues (23). Although these data indicate that FRS-2 binding to the FGFR is constitutive, the possibility that FGF may regulate FRS-2 recruitment in vivo through an alternative mechanism has not yet been determined. Moreover, the precise residues within the FGFR juxtamembrane region that are essential to FRS-2 recruitment and activation have not yet been determined.
Thus, FRS-2 recruitment by the FGFR appears to be phosphotyrosine-independent, whereas we have shown that FRS-2 binding to TrkA requires both an active kinase and is phosphotyrosine-dependent. Tyrosine-phosphorylated FRS-2 binds the SH2 domains of Grb-2 and the tyrosine phosphatase SH-PTP-2. SH-PTP-2 is transiently tyrosine-phosphorylated in response to NGF stimulation, which correlates with an initial ability to bind FRS-2. Because SH-PTP-2 also binds Grb-2 through carboxyl-terminal phosphotyrosines (22,40), FRS-2 facilitates the Shc-independent recruitment of Grb-2 into the NGF-Trk signaling complex both directly and indirectly.
FRS-2 also binds the SH2 and SH3 domain-containing adapter protein Crk, in an NGF and phosphotyrosine-dependent manner. The relevance of this pathway to neuronal differentiation is exemplified by the fact that Crk/C3G/Rap1, but not Grb-2/Sos/Ras, appears to regulate prolonged MAPK activation (17) and by the fact that Crk overexpression induces constitutive PC12 cell differentiation (41). Consistent with these data, the Grb-2 binding sites on FRS-2 are dispensable for FGFinduced differentiation of PC12 cells (40). Importantly, the recent observation that prolonged MAPK activation stimulates expression of the cell cycle inhibitor p21 CIP/Waf-1 (42) provides an important functional link between FRS-2, Crk, cell cycle arrest, and neuronal differentiation. Collectively, these data highlight the functional differences between Shc and FRS-2dependent mechanisms to activate MAPK (Grb-2/Sos/Ras and Crk/C3G/Rap1) and emphasize the need to assay site-directed FRS-2 mutants incapable of binding Crk on NGF-dependent cell cycle arrest and neuronal differentiation.
By comparison, recent studies indicate that SH-PTP-2 activation and binding to FRS-2 are essential to FGF-induced differentiation of PC12 cells (40). Although the mechanism is not yet fully understood, given the arguments outlined above that Crk is essential to prolonged MAPK activation and neuronal differentiation, it is possible that SH-PTP-2 may inactivate an antagonist of MAPK activation, such as Ras-Gap or a MAPK phosphatase, and thereby positively regulate neuronal differentiation. Although overexpression of SH-PTP-2 in PC12 cells does not enhance NGF-dependent differentiation (43), as observed with FGF (40), overexpression of a catalytically inactive SH-PTP-2 decreases NGF-dependent differentiation (43), consistent with the model suggested above that inactivation of a MAPK antagonist, rather than recruitment of Grb-2, is essential to the role of SH-PTP-2 in neuronal differentiation.
FRS-2 also contains several potential binding sites for SH3 domain-containing proteins (PXXP). Of the several proteins that we tested, only the SH3 domain of Src bound FRS-2 constitutively in our in vitro binding assays. It has long been known that constitutive activation of Src generates NGF-independent neuronal differentiation (34) and that the activation of Src precedes the activation of Ras in NGF-dependent TrkA signaling (44). Our demonstration that FRS-2 binds Src in vitro provides the first biochemical evidence of how the Trk receptors recruit Src into their signaling cascades in vivo and further supports the importance of Trk-dependent binding of FRS-2 to the process of NGF-dependent neuronal differentiation.
By taking advantage of a series of specific TrkA signaling mutants, we demonstrate that the FRS-2 binding site on TrkA involves tyrosine 499, as well as the amino acids just 5Ј of this residue. Interestingly, these same residues (around the NPXY motif) also bind the PTB domain of Shc (37,38,45). The possibility that more than one protein could bind to TrkA at this site was previously suggested (21) and is consistent with the observation that Shc, IRS-1, and IRS-2 compete for binding to the NPXY motif of the ␤-chain of the insulin and insulin-like growth factor-1 receptors (46 -48). Overlapping binding sites have also been demonstrated for Nck and PI-3 kinase (p85) binding to the platelet-derived growth factor receptor (49). Thus, the possibility that Shc and FRS-2 can compete for binding to the Trk receptors is not unprecedented. Importantly, we provide direct evidence that Shc and FRS-2 do, in fact, compete for binding to TrkA in vitro. Using a constant amount of GST-Shc(PTB), we found that soluble FRS-2 was able to compete with Shc for binding to TrkA with an EC 50 requiring an approximately 2-5-fold excess of FRS-2. Collectively, these data provide a novel model by which competition between Shc and FRS-2 may occur in vivo and, possibly, regulate a switch between cell cycle progression/mitogenesis and cell cycle arrest/differentiation.
Although overexpression of many components of the Shc/ MAPK pathway (Shc, MAPK kinase (MEK), and MAPK) (50,51) is sufficient to stimulate constitutive neurite outgrowth, this does not imply that Shc-dependent activation of MAPK is the physiologically relevant pathway correlated with prolonged MAPK activation and neuronal differentiation in vivo. Rather, overexpression studies simply saturate one pathway at the expense of another, and if driving prolonged MAPK stimulation is required for neuronal differentiation, then it is perhaps not surprising that overstimulation of Shc-dependent MAPK can drive constitutive neurite outgrowth. Importantly, however, FRS-2 overexpression in nnr5 cells expressing mutant TrkA S3 (⌬ 493 IMENP 497 ) receptors (S3A3 cells) (21) reconstitutes NGFdependent differentiation and cessation of cell cycle progression. TrkAS3 receptors retain the ability to phosphorylate Shc, as well as mitogenic signaling in transfected fibroblasts, yet show a dramatic reduction in the stoichiometric phosphorylation of FRS-2/SNT, as well as the ability to support neurite outgrowth in S3A3 cells (21). Because high concentrations of mutant TrkAS3 receptors can stimulate low levels of FRS-2 binding in vitro (discussed above), it is likely that stable overexpression of FRS-2 in TrkAS3 cells either reconstitutes sufficient TrkA binding or permits sufficiently high levels of FRS-2 tyrosine phosphorylation to facilitate NGF-dependent cell cycle arrest and differentiation. The role of FRS-2 in NGF-dependent differentiation is also supported by mutant Trk receptors in which Tyr 499 is replaced with phenylalanine (TrkAS8). Cells expressing TRK Y499F do not stimulate NGF-dependent Shc tyrosine phosphorylation and/or Shc/Grb-2 binding (21,37). Moreover, Trk Y499F receptors show a reduction in the stoichiometry of NGF-dependent FRS-2 phosphorylation (data not shown). However, as described above for TrkAS3, we have found that overexpression of Trk Y499F receptors can stimulate both FRS-2 phosphorylation and low levels of neuronal differentiation, which may account for the observations by Obermeier et al. (52) and Stephens et al. (37) in which cells expressing high levels of Y499F Trk receptors retain NGF-dependent SNT phosphorylation, as well as varying degrees of neurite outgrowth. The important relevance of the FRS-2 overexpression studies described here are that they reconstitute NGF-dependent neuronal differentiation, suggesting that FRS-2 may be the physiologically relevant substrate essential to NGF-dependent mitotic arrest. The competition between Shc and FRS-2 for binding to NGF-activated TrkA suggests a novel mechanism by which proliferation and differentiation may be regulated in response to neurotrophin stimulation.