The adaptor protein Shc (Src homology and collagen-containing protein) plays an important role in the activation of signalling pathways downstream of RTKs (receptor tyrosine kinases) regulating diverse cellular functions, such as differentiation, adhesion, migration and mitogenesis. Despite being phosphorylated downstream of members of the FGFR (fibroblast growth factor receptor) family, a direct interaction of Shc with this receptor family has not been described to date. Various studies have suggested potential binding sites for the Shc PTB domain (phosphotyrosine-binding domain) and/or the SH2 (Src homology 2) domain on FGFR1, but no interaction of full-length Shc with these sites has been reported in vivo. In the present study, we investigated the importance of the SH2 domain and the PTB domain in recruitment of Shc to FGFR2(IIIc) to characterize the interaction of these two proteins. Confocal microscopy revealed extensive co-localization of Shc with FGFR2. The PTB domain was identified as the critical component of Shc which mediates membrane localization. Results from FLIM (fluorescence lifetime imaging microscopy) revealed that the interaction between Shc and FGFR2 is indirect, suggesting that the adaptor protein forms part of a signalling complex containing the receptor. We identified the non-RTK Src as a protein which potentially mediates the formation of such a ternary complex. Although an interaction between Src and Shc has been described previously, in the present study we implicate the Shc SH2 domain as a novel mediator of this association. The recruitment of Shc to FGFR2 via an indirect mechanism provides new insight into the regulation of protein assembly and activation of various signalling pathways downstream of this RTK.
- fibroblast growth factor receptor 2 (FGFR2)
- fluorescence lifetime imaging microscopy (FLIM)
- Förster resonance energy transfer (FRET)
- Src homology and collagen containing protein (Shc)
- tyrosine kinase
FGFRs [FGF (fibroblast growth factor) receptors] are members of a RTK (receptor tyrosine kinase) family that plays an important role in the regulation of a multitude of cellular functions, such as apoptosis, proliferation, migration, differentiation and survival (reviewed in ). The FGFR family comprises four independent genes (FGFR1–4) that can be alternatively spliced to create numerous receptor isoforms. The different FGFRs are highly homologous, with certain regions, such as the kinase domain, demonstrating sequence identity of up to 90% [1,2]. Differences in the extracellular domain regulate the specific activation of individual FGFR isoforms by different FGFs. Upon ligand binding in the presence of auxiliary glycosaminoglycans, such as heparan sulfate, autophosphorylation occurs on tyrosine residues in the intracellular region. These phosphorylation events facilitate the recruitment of numerous signalling proteins which subsequently activate various pathways downstream of the FGFRs [3,4]. FGFR1 has been used primarily as a model system to elucidate signal transduction events from this RTK family. Seven tyrosine residues (largely conserved among FGFR2–4) have been shown to be phosphorylated in the intracellular region of FGFR1 upon ligand binding . However, despite the sequence homology and the idea that the main difference between FGFR isoforms is the level of phosphorylation, but not the type of signalling proteins phosphorylated , several studies have shown that subtle differences exist in the signalling events emanating from the different FGFR isoforms [7–9]. These differences may be important in the regulation of signalling specificity from the different FGFR isoforms, and highlight the importance of understanding the way in which different signalling proteins are recruited to various FGFR isoforms.
Despite various attempts, few proteins that associate directly with FGFR1 have been identified to date. Of the seven major sites which are subject to autophosphorylation, only PLCγ (phospholipase Cγ) binding to Tyr766 and CrkII (CT10 sarcoma oncogene cellular homologue II) binding to Tyr463 have been described [3,10]. FRS2 (FGFR substrate 2) is known to associate with the receptor constitutively and independently from tyrosine phosphorylation in the juxtamembrane region . The proteins Sef (similar expression to FGF genes) and Grb (growth-factor-receptor-bound protein) 14 have also been shown to bind to FGFR1, but their binding sites have not been described [11,12]. Grb2 is able to bind to the EGFR [EGF (epidermal growth factor) receptor] directly via its SH2 (Src homology 2) domain, but no such interaction could be detected in the case of FGFR1 .
In mammals, the Shc (Src homology and collagen-containing protein) family comprises three different forms of this adaptor protein, namely ShcA, ShcB [also known as Sck (Shc-like protein)] and ShcC, of which the latter two are predominantly expressed in the brain [14–16]. ShcA (most commonly, and in the present study is referred to as Shc) is ubiquitously expressed and itself exists as three isoforms, p66 Shc (66 kDa Shc), p52 Shc (52 kDa Shc) and p46 Shc (46 kDa Shc), where the numerical values reflect their relative molecular masses. These are encoded by two different transcripts generated from the same gene by use of alternative 5′ exons . The p52 Shc and the p46 Shc isoforms are produced from the same transcript by the use of alternative start sites. Shc is an adaptor protein involved in signalling downstream of various RTKs. It contains a CH1 domain (collagen homology 1 domain), a region that is rich in proline and glycine residues, but has no resemblance to the collagen fold, flanked by an SH2 domain at the C-terminus and a PTB domain (phosphotyrosine-binding domain) at the N-terminus (reviewed in ). Whereas the SH2 domain mediates the association with cell-surface receptors, such as the T-cell receptor and EGFR , the PTB domain has been shown to interact with a wider variety of RTKs, including EGFR, insulin receptor, HER2 (human EGFR 2)/neu and TrkA (tyrosine kinase neurotrophin receptor A) [20,21]. The similarity of the Shc PTB domain fold to that of PH domains (pleckstrin homology domains) means that, in addition to mediating binding to tyrosine-phosphorylated motifs, it is able to bind to acidic phospholipids (such as phosphatidylinositols) . The CH1 domain contains tyrosine phosphorylation sites, Tyr239/Tyr240 and Tyr317, which are phosphorylated upon growth factor receptor activation and consequently form docking sites for the SH2 domain of the adaptor protein Grb2. The best known role of Shc is the recruitment of the Grb2–Sos (Son of sevenless) complex to activated RTKs and the subsequent activation of the Ras/Raf/MEK [MAPK (mitogen-activated protein kinase)/ERK (extracellular-signal-regulated kinase) kinase]/ERK1/2 pathway . However, it has also been shown to be involved in the activation of other downstream signalling pathways. The phosphorylation sites in the CH1 domain have been shown to perform pleiotropic and non-redundant roles in JNK (c-Jun N-terminal kinase) and p38 MAPK activation, cell death and c-myc transcription .
Numerous studies have shown that Shc is phosphorylated in response to FGF stimulation of cells expressing FGFR1, FGFR3 or FGFR4 [5,6,8,13,24,25]. It has been suggested previously that a possible binding site for the Shc PTB domain may be present on FGFR1, since a phosphopeptide based on the sequence surrounding Tyr730 was able to bind to Shc and block the mitogenic function of FGFR1 . On the other hand, synthetic peptides corresponding to the sequences surrounding phosphorylated Tyr766, Tyr730 and Tyr558 of FGFR1 can bind to the isolated Shc SH2 domain in vitro [13,27]. Despite the presence of potential binding sites for the Shc SH2 and/or PTB domains on FGFR1 (which are conserved among other FGFR isoforms), co-precipitation of Shc was only demonstrated in the case of p66 Shc with FGFR3 and under non-physiological conditions with FGFR1 in mammalian cells expressing v-Src (the virally encoded form of Src) [28,29]. The latter finding indicated that a direct interaction between the receptor and Shc is possible and that a Shc-binding site is present on FGFR1. Furthermore, the Shc SH2 domain has been shown to block ERK2 activation downstream of the Pleurodeles FGFR1 (which is homologous with human FGFR1) in Xenopus laevis oocytes [29a]. Although this does not necessarily indicate direct binding between the two proteins, it suggests that such an event would occur via the SH2 domain. Despite such evidence of the involvement of Shc in signal transduction from various FGFR family members and the identification of potential binding sites for Shc on FGFRs, it is clear that the method by which this adaptor protein is recruited to this receptor family in vivo needs to be investigated further. Identification of the Shc-binding site in vivo and the importance of the different Shc domains in recruitment to FGFRs would provide important details about the activation of effectors, such as the different members of the MAPK family downstream of this family of RTKs.
In the present study, we chose to focus on the interaction of Shc (p52 and p46 isoforms) with FGFR2, because signalling from this member of the FGFR family has been poorly investigated to date. Additionally, identification of a Shc-binding site on one FGFR isoform may provide important insight into the recruitment of this adaptor to the entire FGFR family. In the present study, we report that Shc co-localizes with FGFR2 at the plasma membrane and in intracellular membrane compartments and that it is associated with the receptor in a manner that allows co-precipitation of the two proteins. We highlight the requirement for the Shc PTB domain in its membrane recruitment and the necessary context of the full-length protein (i.e. the presence of both the SH2 and PTB domains) for stable association with the receptor. FLIM (fluorescence lifetime imaging microscopy) revealed that, despite co-localization and co-precipitation of Shc with FGFR2, these two proteins do not interact directly in vivo. Thus Shc must be recruited to a multimeric protein complex upon FGFR2 activation, rather than utilizing any of the previously described binding sites on the receptor directly. We identified the tyrosine kinase Src as a binding partner of Shc in cells expressing FGFR2, which may play an important role in regulating the assembly of the ternary complex involving Shc and FGFR2. Shc has been shown to interact with various RTKs, including EGFR, insulin receptor and TrkA receptor. Our observations of a lack of direct association of Shc with FGFR2 highlight an important alternative mechanism for the recruitment of signalling proteins to RTKs to finely control the specific activation of various downstream signalling pathways. The fact that Shc is phosphorylated following FGFR activation, but does not directly associate with them, despite the presence of binding sites, highlights the fact that the recruitment of adaptor proteins into specific multiprotein signalling assemblies downstream of different receptors plays an important part in regulating signalling specificity.
MATERIALS AND METHODS
FGF9 was from R&D Systems and heparan sulfate was from Sigma–Aldrich. The anti-GFP (green fluorescent protein) antibody was purchased from Rockland Immunochemicals, the anti-Shc antibodies were from Upstate and BD Transduction Laboratories, and the anti-Src antibody was from Cell Signaling Technologies. Secondary horseradish peroxidase-conjugated secondary antibodies were obtained from Sigma–Aldrich.
cDNA encoding FGFR2(IIIc) was a gift from Professor John Heath (School of Biosciences, University of Birmingham, Edgbaston, Birmingham, U.K.). It was PCR-amplified and cloned in-frame into the EGFP-N2 vector (Clontech). The vectors encoding mRFP [monomeric RFP (red fluorescent protein)] used to generate N-terminal or C-terminal RFP-fusion proteins were a gift from Professor Tony Ng (Randall Division of Cell and Molecular Biophysics, School of Biomedical and Health Sciences, King's College London, London, U.K.). cDNA encoding the p52 isoform of human ShcA (amino acids 17–472) was a gift from Professor Kodi Ravichandran (Department of Microbiology, University of Virginia, Charlottesville, VA, U.S.A.). Full-length Shc was PCR-amplified and cloned into the RFP-C vector to create a C-terminally RFP-tagged protein. The Shc ΔPTB (Shc lacking the PTB domain) construct (see Figure 3) was created by PCR amplification with a separate forward primer and was cloned into the RFP-C vector. Shc R401A, Shc R175Q and Shc Triple (Shc R112Q/K116A/K139A) mutant constructs were created using the QuikChange® site-directed mutagenesis kit (Stratagene) following the manufacturer's instructions, using the C-terminally RFP-tagged Shc as a template. The Shc SH2 domain [as a GST (glutathione transferase)-tagged fusion protein in the pGEX-2T vector] was a gift from Professor Cosima Baldari (Department of Evolutionary Biology, University of Siena, Siena, Italy) and was subcloned into the RFP-N vector to create an N-terminally tagged fusion protein. The Shc PTB domain was PCR-amplified using appropriate primers and cloned in-frame into the RFP-C vector or into the pGEX4T-2 GST-expression vector (Amersham Biosciences).
GST fusion proteins of the SH2 and PTB domains and full-length Shc (cloned into the pGEX4T-2 expression vector) were created by expression in the BL21(DE3) Escherichia coli strain and purified using GST-Bind Resin (Novagen).
HEK-293T cells (human embryonic kidney cells expressing the large T-antigen of simian virus 40) were cultured in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) fetal bovine serum. Prior to stimulation, cells were serum-starved for 18 h in serum-free DMEM. Cells were transfected using Lipofectamine™ 2000 (Invitrogen) according to the manufacturer's instructions. Stable clones expressing the FGFR2–GFP construct were isolated by dilution cloning in 96-well plates and screening for fluorescence.
PC12 cells were cultured in DMEM supplemented with 5% (v/v) fetal bovine serum and 10% (v/v) horse serum. Prior to stimulation, cells were serum-starved for 18 h in DMEM with 0.1% horse serum. PC12 cells expressing FGFR2–GFP were generated as described previously . Stable cells expressing FGFR2–GFP and Shc–RFP were created by additional transfection with cDNA encoding C-terminally RFP-tagged Shc and selecting for stably transfected cells by dilution cloning in 96-well plates and screening for fluorescence.
For immunoprecipitation experiments, cells were serum-starved for 18 h and stimulated with 10 ng/ml FGF9 in the presence of an additional 1 μg/ml heparan sulfate for 15 min and lysed in 20 mM Tris/HCl (pH 7.5), 138 mM NaCl, 1 mM EGTA, 20 mM 2-glycerophosphate, 10% (w/v) glycerol, 1 mM sodium orthovanadate and 20 mM sodium fluoride, supplemented with 1% protease inhibitor cocktail III (Calbiochem). Whole-cell lysates (2 mg of total protein) were subjected to pulldown experiments using GST fusion proteins or immunoprecipitation with appropriate antibodies as indicated. Immunoprecipitates were captured on Protein A or Protein G beads, subjected to four washes with lysis buffer, followed by SDS/PAGE and Western blotting with the appropriate antibodies as indicated.
Cells were seeded on to glass coverslips (coated with poly-D-lysine in the case of PC12 cells) and allowed to adhere overnight. Following serum starvation, cells were stimulated with 10 ng/ml FGF9 in the presence of additional heparan sulfate (1 μg/ml) for 15 min. Following a wash in PBS, cells were fixed in 4% (w/v) paraformaldehyde in PBS (pH 8.0). Coverslips were washed in PBS and mounted on to glass coverslides on a drop of mounting medium [0.1% 1,4-phenylenediamine (anti-fade agent; Fluka) in 50% (v/v) glycerol in PBS (pH 8.0)]. Slides were analysed using a Leica TCS SP2 system with a ×63 oil-immersion objective lens. GFP was excited at 488 nm using an argon visible-light laser and its emission was detected using a 514/10 nm band selection. Mid-sections of the cells were chosen to avoid interference from cell attachment to the slides. Fluorescence images were collected using a photomultiplier tube interfaced to an Intel Pentium II system running the Leica TCS NT control software. The images presented are representative of 10–15 fields of view analysed per independent experiment. Fluorescent intensity graphs were created using the Leica LCS software. Peaks represent the pixel intensity of GFP (green) and RFP (red) respectively along the points of an arbitrarily drawn line.
FLIM analysis was carried out using a Leica TCS SP2 inverted microscope set-up with a ×63 water-immersion objective lens, which was adapted for TCSPC (time-correlated single-photon counting) FLIM with a Becker and Hickl SPC 830 card using 64 or 256 time channels in a 3 GHz, Pentium IV, 1 GB RAM computer. The samples were excited using a femtosecond titanium sapphire laser (Coherent Mira, repetition rate of 76 MHz) that was pumped by a 6.5 W solid-state laser (Coherent Verdi V6). Images were obtained with a line-scan speed of 200 Hz. Two-photon excitation was carried out using a wavelength of 900 nm and fluorescence was detected through a 525±25 nm interference filter using a cooled PMC100-01 detector (Becker and Hickl, based on a Hamamatsu H5772P-01 photomultiplier). The fluorescence decays obtained were fitted using a single exponential decay model with Becker and Hickl SPCImage software v2.8.3. GFP fluorescence lifetimes were portrayed in false colour maps. The images shown are representative of at least five independent fields of view analysed. Changes in overall GFP lifetime in the entire field of view were represented graphically in the form of lifetime histograms.
Co-precipitation of Shc and FGFR2
To assess the co-localization and interaction of Shc and FGFR2, we created cell lines expressing a GFP-tagged version of FGFR2(IIIc). We showed previously that this receptor is active and responds to FGF9 stimulation [30,31], indicating that GFP tagging does not interfere with receptor activity. Immunoprecipitation experiments revealed that Shc (p52 and p46 isoforms) and FGFR2 could be co-precipitated from both HEK-293T and PC12 cells expressing the GFP-tagged version of FGFR2 (Figure 1). Co-precipitation of Shc with the receptor was somewhat reduced in PC12 cells compared with HEK-293T cells. This is most likely to be the result of cell-specific differences in protein expression, particularly of transfected FGFR2–GFP. The observed co-precipitation indicates an interaction between FGFR2 and Shc.
Co-precipitation of Shc and FGFR2 did not increase significantly following stimulation with FGF9. An explanation for this observation could be the fact that FGFR2 is already highly phosphorylated in the unstimulated state, and its level of phosphorylation only increases minimally upon FGF9 stimulation [30,31]. Association of Shc with this basally phosphorylated receptor may already take place in the absence of FGF stimulation. FGF binding to the receptor would, however, be required to induce the activation of downstream signalling pathways, since pathways such as the ERK1/2 pathway are not activated in unstimulated cells [30,31].
Interestingly, this observation also highlights potential differences in the way in which proteins are recruited to various FGFR isoforms, since co-precipitation of Shc with FGFR1 under physiologically relevant conditions has not been observed previously [5,13]. We therefore chose to investigate in more detail the association of Shc with FGFR2 and its role in intracellular signalling from this receptor.
Co-localization of Shc with FGFR2 on the plasma membrane and intracellular membraneous compartments
To confirm the role of Shc in FGFR2 signal transduction suggested by the immunoprecipitation experiments, we chose to investigate the cellular localization of Shc in relation to FGFR2. Intracellular co-localization of two proteins does not indicate their direct interaction, but does highlight the recruitment of two proteins to the same cellular compartment and the involvement of adaptor proteins in signalling from a certain receptor. In conjunction with co-precipitation experiments, co-localization may yield important insights into the spatial and temporal regulation of intracellular protein–protein interactions.
HEK-293T cells expressing FGFR2–GFP were transfected with mRFP-tagged Shc (encoding the p52 and p46 isoforms, see Figure 3B) and cells were stimulated with FGF9. Phosphorylation of mRFP-tagged Shc occurred in a manner comparable with untagged Shc, which indicates that the mRFP tag does not interfere with Shc functionality (results not shown). Strong co-localization of Shc with FGFR2 at the plasma membrane was observed in HEK-293T cells [Figure 2A, yellow colour on overlay images as well as overlapping RFP (red) and GFP (green) peaks in the fluorescence intensity graphs]. Co-localization on the plasma membrane occurred in both the unstimulated and FGF9-stimulated states, which is in agreement with the co-immunoprecipitation experiments. In the absence of FGFR2, Shc was not significantly localized to the plasma membrane (Figure 2B), which confirms that the observed membrane localization of Shc occurs as a result of an association with FGFR2.
In PC12 cells, a similar co-localization pattern was observed (Figure 2A); however, Shc membrane localization increased strongly upon FGF9 stimulation compared with HEK-293T cells (increased yellow colour in the overlay image and greater overlap of the red and green peaks in fluorescence intensity graphs). Despite slight cell-specific differences, extensive co-localization between Shc and FGFR2 was observed. This further indicates the involvement of this adaptor protein in FGFR2 signalling and raises the question of whether a direct binding site for Shc on the receptor can be identified and which domain of Shc is involved in mediating the recruitment of Shc to FGFR2.
The Shc PTB domain is required for its membrane localization
Shc possesses two domains that could potentially mediate association with phosphorylated tyrosine motifs on FGFR2: the N-terminal PTB domain and the C-terminal SH2 domain (Figure 3). Both domains have been shown to mediate the association of Shc with other RTKs, such as the EGFR, PDGFR (platelet-derived growth factor receptor) and insulin receptor [20,32,33]. Several potential binding sites for either the Shc SH2 domain or the Shc PTB domain on FGFR1 have been identified in vitro [13,26,27]. These sites are conserved between FGFR1 and FGFR2 and could mediate the recruitment of Shc to FGFR2 and hence result in the observed co-localization and co-immunoprecipitation of Shc with this receptor. To assess whether the PTB domain or the SH2 domain or indeed phosphorylation on tyrosine residues in the CH1 domain were required for Shc recruitment to the membrane in response to the expression and activation of FGFR2, we created three Shc constructs that were missing a functional SH2 domain (Shc R401A) or the PTB domain (Shc ΔPTB) respectively, or in which all three tyrosine residues in the CH1 domain were mutated to phenylalanine [Shc 3F (Shc Y239F/Y240F/Y317F)] (see Figure 3 for details of all constructs used and expression of RFP-tagged constructs in HEK-293T cells). Western blotting revealed the integrity of all RFP fusion proteins (Figure 3B). All three cDNA constructs were transiently transfected into HEK-293T cells and co-localization with FGFR2 was observed both in the absence and in the presence of FGF9 (Figure 4).
Mutation of the arginine residue critical for phosphotyrosine binding by the SH2 domain (Shc R401A) did not affect the localization of Shc to the plasma membrane (Figure 4). The pattern of co-localization with the receptor was unaltered from that of full-length Shc. Similarly, mutation of Tyr239, Tyr240 and Tyr317 to phenylalanine (Shc 3F) did not affect co-localization of Shc with FGFR2, as highlighted by the fluorescence intensity graphs (Figure 4). These observations indicate that the Shc SH2 domain and tyrosine phosphorylation in the CH1 domain are not required for the co-localization of Shc with FGFR2.
In contrast, Shc ΔPTB was unable to localize to the plasma membrane (Figure 4). This indicates that the PTB domain plays a major role in mediating the co-localization of Shc with FGFR2 at the plasma membrane. However, the resolution of confocal microscopically assessed protein co-localization does not allow the assessment of direct protein–protein interactions in vivo. Thus, despite the requirement of the PTB domain for the co-localization of Shc with FGFR2, an involvement of the SH2 and CH1 domains in the association with the receptor cannot be ruled out by the present study.
The Shc PTB domain can bind to tyrosine-phosphorylated motifs on target proteins, but is also able to bind to phospholipids [34,35]. This dual-binding capacity means that removal of the Shc PTB domain as a whole does not discriminate between membrane localization as a result of phospholipid binding or an interaction with FGFR2. We therefore chose to individually mutate the binding sites for phospholipids (Shc Triple) and phosphotyrosine (Shc R175Q)  (see Figure 3 for details). As expected, the mutant protein was still able to bind to phospholipids (Shc R175Q) and was membrane localized in a fashion comparable with full-length Shc (Figure 4). Previous studies have shown that the ability of the PTB domain to bind to both phospholipids and phosphotyrosine is important for its localization to the membrane . In agreement with this, Shc lacking the ability to bind phospholipids (Shc Triple) was also able to co-localize with FGFR2 upon FGF9 stimulation (Figure 4). This observation indicates that binding of tyrosine-phosphorylated proteins plays a role in Shc recruitment to the membrane in addition to its ability to bind phospholipids via the PTB domain.
Requirement of functional SH2 and PTB domains for recruitment to FGFR2
Since the co-localization of two proteins does not necessarily infer their direct interaction, we chose to utilize various Shc mutants (see Figure 3) in order to assess which of the Shc domains was important for mediating its co-precipitation with FGFR2. HEK-293T cells stably expressing FGFR2–GFP were transiently transfected with equal amounts of DNA encoding Shc, Shc 3F, Shc ΔPTB or Shc R401A as for co-localization experiments and stimulated with FGF9. FGFR2 was immunoprecipitated with an anti-GFP antibody to ensure isolation of only the exogenously expressed FGFR2 (Figure 5). Co-localization of Shc lacking its three tyrosine phosphorylation sites in the CH1 domain was unaltered from that of full-length Shc (Figure 4). Likewise, Shc 3F co-precipitated with FGFR2 to a similar extent as full-length Shc (Figure 5, doublets represent the p46 and p52 isoforms), which supports the notion that phosphorylation in the CH1 domain is not required for Shc to associate with this receptor.
However, both the Shc R401A and the Shc ΔPTB constructs, lacking a functional SH2 and PTB domain respectively, were observed to co-precipitate with the receptor to a significantly lesser extent than full-length Shc and could only be observed by prolonged exposure of the Western blot. This highlights a requirement for both the SH2 and the PTB domain for effective association of Shc with FGFR2. The Shc ΔPTB mutant protein is probably unable to associate with FGFR2 because the lack of the PTB domain impairs its recruitment to the plasma membrane and hence to the receptor itself. Interestingly, Shc R401A (containing a non-functional SH2 domain) was also unable to associate with FGFR2 to a great extent, despite its co-localization being unaltered from that of full-length Shc (Figure 4). Thus, despite the ability to mediate the efficient co-localization of Shc with FGFR2, the presence of the PTB domain alone is not sufficient to result in stable complex formation with the receptor. Taking into account these results, it appears that the PTB domain is required to bring the SH2 domain into the vicinity of FGFR2, where the latter then stabilizes and/or mediates the association with FGFR2. In the absence of the PTB domain, the lack of Shc membrane association would be expected to prevent the stable association of Shc with FGFR2.
No direct protein–protein interaction takes place between FGFR2 and Shc
Neither pulldown methods, such as immunoprecipitation, nor confocal microscopy analysis of protein co-localization can indicate conclusively whether two proteins interact directly or whether they are found in a ternary complex (i.e. are associated indirectly). Since both the PTB and the SH2 domain seemed to be required for the association with FGFR2, we chose to use FLIM to determine unequivocally whether a direct association between Shc and FGFR2 occurs and whether the SH2 or the PTB domain or indeed both mediate this interaction. When an acceptor fluorophore, such as RFP, is within a close enough distance (typically between 1–10 nm) to a donor fluorophore (GFP), this will result in FRET (Förster resonance energy transfer) and a subsequent shortening of the donor fluorescence lifetime [36,37]. By measuring the differences in donor lifetime in the absence or presence of an acceptor fluorophore, a direct interaction between two fluorescently tagged proteins can be assessed within a cell. FLIM has been used successfully previously to determine direct protein–protein interactions in vivo and is therefore a suitable technique to characterize the interaction between Shc and FGFR2 .
The average fluorescence lifetime of GFP in the context of the FGFR2–GFP construct was approx. 2 ns, as indicated by the peak on the lifetime histogram (Figure 6A). This value is similar to that of GFP alone (Figure 6A). The slightly wider peak in cells expressing FGFR2–GFP compared with GFP alone indicates that there is a larger spread of GFP molecules with shorter or longer lifetimes, which may be the result of varying conditions in the environment of GFP when located near the plasma membrane . The expression of mRFP in cells also expressing GFP did not result in a decrease in GFP lifetime (Figure 6A). This illustrates that FRET between the two fluorophores does not occur unless they are positioned proximal to each other as a result of a direct protein–protein interaction between two tagged proteins.
Immunoprecipitation and co-localization experiments suggested that the PTB domain was necessary to recruit Shc to the membrane, but that the SH2 domain was a major component required for stable complex formation with FGFR2. To determine whether either of these domains is able to mediate the direct association of Shc with the receptor, cells were transiently transfected with constructs encoding the RFP-tagged Shc SH2 or PTB domains, or full-length Shc and stimulated with FGF9 or left unstimulated.
Surprisingly, FLIM analysis revealed that Shc and FGFR2 do not associate directly in vivo (Figure 6B). Neither expression of the SH2 or PTB domains on their own nor of full-length Shc induced a significant change in the overall GFP lifetime (Figure 6B). The peaks in the lifetime histograms remained centred around 2 ns, which indicates that the GFP molecules on the receptor and the RFP molecules on the Shc constructs were not positioned proximal to each other as a result of a direct interaction between the tagged proteins. A small left-shift of the peak was observed in FGF9-stimulated cells expressing the RFP-tagged SH2 domain. However, since full-length Shc did not show a similar pattern, this is likely to be the result of overexpression of the SH2 domain and non-specific binding to tyrosine-phosphorylated sites on FGFR2. The fact that no changes in GFP lifetime could be observed indicates that Shc does not bind directly to the receptor, despite the presence of various potential binding sites. The association of these two proteins must therefore occur indirectly via ternary complex formation.
Src is a candidate protein mediating the interaction of Shc with FGFR2
Since FLIM revealed that no direct interaction takes place between Shc and FGFR2, we decided to probe for candidate proteins that could mediate ternary complex formation between these two proteins. Shc immunoprecipitation and pulldown experiments using GST-tagged SH2 and PTB domains were performed. Although proteins such as Grb2, PLCγ and IRS4 (insulin receptor substrate 4) were observed in Shc immunoprecipitates, we were unable to identify any of these proteins in pulldown experiments using the isolated domains (results not shown). Their presence in Shc immunoprecipitates was therefore likely to be the result of ternary complex formation. We further tested the interaction of Shc with FRS2, CrkII, SH2-B (SH2-containing signalling protein B) and c-Cbl, but were unable to show binding of any of these proteins to Shc.
Interestingly, Shc could also be co-immunoprecipitated with the non-RTK Src (results not shown). Pulldown experiments using full-length Shc also revealed an association between these two proteins (Figure 7B). The role of Src in FGFR signalling is somewhat controversial , but a direct interaction between the kinase and FGFR1 has been shown in vitro and a previous study has shown that Src plays an important role in FGFR1 activation and signalling dynamics [41,42]. The latter study also showed that Src co-localizes with FGFR1. Additionally, we observed co-immunoprecipitation of Src with FGFR2 (Figure 7C), which further indicates that it may play a role in the recruitment of Shc to FGFR2 and hence in the localization of Shc to the plasma membrane in cells expressing this receptor. Pulldown experiments using GST-tagged Shc SH2 and PTB domains respectively revealed that the association between Shc and Src is mediated via the SH2 domain (Figure 7A). This observation correlates with results highlighting the requirement for a functional Shc SH2 domain for ternary complex formation between Shc and the receptor (Figure 5). Co-precipitation of FGFR2 with the Shc SH2 domain was observed only in the GST pulldown assay following ligand stimulation (Figure 7A). This provides further evidence for the role of Src in ternary complex formation. Hence the SH2 domain and the presence of Src play an important role in mediating the association of Shc with the receptor.
Previous studies have shown that Shc is able to bind Src via its PTB domain and results in activation of Src kinase activity [43,44]. We failed to observe a significant association of the Shc PTB domain with Src (Figure 7A). Differences in the cell and receptor systems used in the present work compared with previous studies may account for the differences observed. An interaction of the Shc SH2 domain with Src has not been described previously. The fact that this interaction could also be observed in unstimulated cells raises the possibility that it is not phosphorylation dependent and is in agreement with the observation that Shc and FGFR2 were co-precipitated in unstimulated cells. Shc has been shown previously to bind to a protein called mPAL (mouse protein expressed in activated lymphocytes) via its SH2 domain in a non-phosphorylation-dependent manner . A similar mechanism may be at work in the case of Shc and Src. Alternatively, the Shc SH2 domain could bind to one of the Src autoregulatory tyrosine phosphorylation sites (of which Tyr527 would be expected to be phosphorylated in resting cells). Such a binding event may subsequently play a role in regulating kinase activity. Regardless of the precise manner of the interaction, our results highlight Src as a candidate protein mediating the indirect interaction of Shc with FGFR2. Such a mechanism of recruitment opens up the possibility of interesting new investigations into the precise role of Shc in FGFR2 signalling and its contribution to the activation of different downstream signalling pathways.
Although the adaptor protein Shc has been implicated as a major component in the recruitment of Grb2 to numerous RTKs, its precise involvement in FGFR signalling has remained somewhat of a mystery. The present study has revealed that Shc co-localizes intracellularly with FGFR2, and that these two proteins can be co-precipitated. This observation contrasts with the fact that co-precipitation of these Shc isoforms could not be observed in the case of FGFR1 or FGFR3 [28,29]. Additionally, it indicates that, despite a high degree of sequence homology between these three FGFR isoforms, differences exist in the way in which Shc is recruited to various members of this growth factor receptor family. Any such differences in the recruitment of Shc to various FGFR isoforms could play an important role in the activation of different pathways downstream of the individual receptors within this family. This is further supported by previous work that highlighted differences in the affinity of FRS2 for FGFR1 and FGFR3 . The differential recruitment of the same signalling proteins to individual members of this RTK family would be expected to regulate the generation of specific downstream signals. By expressing different FGFR isoforms at their surface, cells could therefore fine-tune their response to stimulation by FGFs not only by presenting different extracellular domains with varying specificity for FGF isoforms, but also by generating slightly altered intracellular signalling pathways.
It was somewhat unexpected to find that, despite intensive co-localization and co-precipitation of Shc and FGFR2, no direct interaction between these two proteins could be observed using a high-resolution FRET technique (Figure 6). It is particularly interesting to note that no interaction was observed despite the presence of various potential binding sites for Shc on FGFR2 [13,26,27]. The indirect association of these two proteins must be mediated via the formation of a ternary signalling complex involving at least one other protein. We identified the non-RTK Src as a mediator of the indirect interaction between FGFR2 and Shc. Although a specific binding site for Src on FGFR2 has not been described, previous work has shown that the Src SH2 domain is able to bind to FGFR1 in vitro . Further support for the fact that Src plays an important role in mediating the association of Shc with FGFR2 comes from a previous study which showed that Shc could be co-precipitated with FGFR1 in cells overexpressing v-Src . This observation strengthens the proposal that Src plays a role in the recruitment of Shc to a ternary protein complex involving FGFR2, and further highlights differences in the way proteins are recruited to individual members of the FGFR family.
The Shc PTB domain was required for membrane localization (Figure 4). However, the addition of the SH2 domain was necessary to obtain stable ternary complex formation with FGFR2 (Figure 5). Previous studies have shown a lack of localization of the Shc ΔPTB construct to the membranous fraction, whereas full-length Shc was associated with the membrane fractions of unstimulated T-cells, BaF cells and COS cells . The results of the present study are consistent with these previous reports. The ability of the PTB domain to bind phospholipids was sufficient to mediate the extensive co-localization of Shc with FGFR2 at the plasma membrane (Figure 4, Shc R175Q). When considered in conjunction with the fact that the lack of a functional SH2 domain greatly reduced the ability of Shc to co-precipitate with the receptor (Figure 5), a model emerges in which the PTB domain mediates membrane localization and brings the SH2 domain into the vicinity of FGFR2. This would then allow the SH2 domain to interact with proteins, such as Src, mediating the formation of the ternary signalling complex between Shc and FGFR2. We have shown previously that the Shc SH2 domain is occluded until phosphorylation in the CH1 domain occurs . Recruitment of Shc to the membrane via the PTB domain would ensure its localization to a site where phosphorylation and consequently interaction with other proteins via the SH2 domain can occur.
In a search for candidate proteins mediating the formation of a ternary complex involving FGFR2 and Shc, we probed for the interaction of the Shc PTB and/or SH2 domains with numerous proteins known to bind to FGFR1 or to at least be involved in FGFR signalling. We identified the non-RTK Src as an interacting partner of the Shc SH2 domain. This observation is in agreement with the fact that the Shc SH2 domain was required for the stable association of this adaptor protein with FGFR2 (Figure 5). Since Src is able to phosphorylate Shc, the interaction of these proteins is likely to play a role in the regulation of Shc phosphorylation and hence the activation of downstream signalling pathways. A previous study showed that binding of phosphatidylinositol 4,5-bisphosphate to the Shc PTB domain stimulates its phosphorylation by Src . This observation supports our model for the FGFR2 system, in which the Shc PTB domain binds to phospholipids in the plasma membrane, whereas the SH2 domain mediates the association of Shc with Src. This could subsequently facilitate the phosphorylation of Shc by this kinase. Further work would need to be carried out to investigate whether Shc phosphorylation downstream of FGFR2 does indeed occur via Src and not by the receptor itself. Additionally, the fact that Shc co-localization was not altered upon removal of the ability of Shc to bind phospholipids via the PTB domain suggests that the PTB domain could be involved in further stabilizing this ternary complex via interactions with other unidentified proteins. This is supported by the fact that the PTB domain is able to bind both phospholipids and phosphoproteins at the same time .
Shc has been shown to bind directly to numerous RTKs. Therefore the observation that the association of Shc with FGFR2 is indirect despite co-precipitation and the presence of potential Shc-binding sites on the receptor was somewhat unexpected. However, the association of Shc with Src, and hence indirectly with FGFR2, may highlight roles for Shc in FGFR2 signalling in addition to Grb2–Sos complex recruitment. Further investigation of the way in which Shc and Src interact and the potential effects on the activity of the kinase and/or the recruitment of other proteins involved in signalling from FGFR2 are required to describe in detail the effects that this novel method of Shc recruitment has on the activation of downstream signalling pathways.
Our results show that, despite its ability to bind to numerous RTKs, Shc does not interact directly with FGFR2. Taking into consideration the fact that previous studies were unable to show co-precipitation of p52/p46 Shc and FGFR1 or FGFR3 under physiological conditions, it is likely that, in vivo, FGFRs in general do not present a binding site for Shc. In addition to highlighting differences in protein recruitment to different members of the FGFR family, this observation also reveals an important way in which signalling specificity downstream of different RTKs can be achieved. Shc has been shown to bind directly to RTKs, such as EGFR, PDGFR, TrkA and the insulin receptor [20,21]. The exclusion of full-length Shc binding to FGFR2 despite the presence of potential binding sites indicates that different RTKs are able to recruit downstream adaptor proteins in a very specific and regulated fashion. Such precise recruitment of proteins to the activated receptor would then be expected to lead to the activation of a specific combination of downstream pathways to relay information from the cell surface to the interior of the cell. Although further investigation is necessary to elucidate the exact mechanism by which Shc is recruited to FGFR2 via Src, the fact that Shc was recruited to this receptor in an indirect fashion requiring both functional SH2 and PTB domains indicates that various receptors may indeed assemble highly individual signalling complexes to specifically activate a host of downstream signalling pathways.
J. E. L. is a Wellcome Trust Senior Research Fellow. A. C. S. received the Countess of Lisburne Scholarship for doctoral research.
Abbreviations: CH1 domain, collagen homology 1 domain; CrkII, CT10 sarcoma oncogene cellular homologue II; DMEM, Dulbecco's modified Eagle's medium; EGF, epidermal growth factor; EGFR, EGF receptor; ERK, extracellular-signal-regulated kinase; FGF, fibroblast growth factor; FGFR, FGF receptor; FLIM, fluorescence lifetime imaging microscopy; FRET, Förster resonance energy transfer; FRS2, FGFR substrate 2; GFP, green fluorescent protein; Grb, growth-factor-receptor-bound protein; GST, glutathione transferase; HEK-293T cell, human embryonic kidney cell expressing the large T-antigen of simian virus 40; MAPK, mitogen-activated protein kinase; PDGFR, platelet-derived growth factor receptor; PLCγ, phospholipase Cγ; PTB domain, phosphotyrosine-binding domain; RFP, red fluorescent protein; mRFP, monomeric RFP; RTK, receptor tyrosine kinase; SH2, Src homology 2; Shc, Src homology and collagen-containing protein; p46 Shc, 46 kDa Shc; p52 Shc, 52 kDa Shc; p66 Shc, 66 kDa Shc; Shc3F, Shc Y239F/Y240F/Y317F; Shc, ΔPTB, Shc lacking the PTB domain; Shc Triple, Shc R112Q/K116A/K139A; Sos, Son of sevenless; TrkA, tyrosine kinase neurotrophin receptor A
- © The Authors Journal compilation © 2008 Biochemical Society