Interaction of Fibroblast Growth Factor Receptor 3 and the Adapter Protein SH2-B: A Role in Stat5 Activation

upon FGFR3-mediated signaling through SH2-B b . Our data thus suggest that the adapter protein SH2-B b may represent an important signaling molecule that mediates downstream biological effects of FGFR3 activation.


SUMMARY
Fibroblast growth factor receptor 3 (FGFR3) influences a diverse array of biological processes, including cell growth, differentiation and migration. Activating mutations in FGFR3 are associated with multiple myeloma, cervical carcinoma and bladder cancer. To identify proteins that interact with FGFR3 and which may mediate FGFR3-dependent signaling, a yeast two-hybrid screen was employed using the cytoplasmic kinase domain of FGFR3 as the bait.
We identified the adapter protein SH2-B as an FGFR3-interacting protein.
Coimmunoprecipitation experiments demonstrate binding of the SH2-Bβ isoform to FGFR3 in 293T cells. Tyrosine phosphorylation of SH2-Bβ was observed when coexpressed with activated FGFR3 mutants, such as the weakly activated mutant N540K or the strongly activated mutant K650E, both associated with human developmental syndromes. The extent of tyrosine phosphorylation of SH2-Bβ correlates with receptor activation, suggesting that FGFR3 activation mediates tyrosine phosphorylation of SH2-Bβ. Furthermore, two tyrosine phosphorylation sites of FGFR3, Y724 and Y760, are required for optimal binding of the Src homology-2 (SH2) domain of SH2-Bβ. We also demonstrate the phosphorylation and nuclear translocation of Stat5 by activated FGFR3, which increases in response to overexpression of SH2-Bβ. Taken together, our results identify SH2-Bβ as a novel FGFR3 binding partner which mediates signal transduction.

INTRODUCTION
Fibroblast growth factor receptors (FGFRs) are receptor tyrosine kinases (RTKs) that integrate many different intercellular signals affecting cell growth, differentiation, migration, wound healing and angiogenesis, depending on the target cell type and developmental stage (1,2). The FGFR family comprises of four structurally related members: FGFR1, FGFR2, FGFR3 and FGFR4, exhibiting three extracellular immunoglobulin-like (Ig) domains, a single transmembrane domain and a split intracellular tyrosine kinase domain (3,4,5). Mutations in FGFRs, which may be either familial or spontaneous in origin, are responsible for a large number of human developmental disorders including skeletal dwarfism and craniosynostosis syndromes (6,7,8). Translocations and mutations affecting members of the FGFR family are also importantly associated with several human cancers (6,(9)(10)(11)(12).
FGFR3 plays a particularly important role in skeletal development (13)(14)(15). Disruption of murine FGFR3 produces severe and progressive bone dysplasia with enhanced endochondral bone growth, suggesting that FGFR3 mediates the negative regulation of bone growth (16,17).
Mutations in FGFR3 are directly responsible for human dwarfism syndromes, including hypochondroplasia, achondroplasia and thanatophoric dysplasia (TD) (6). Several of the mutations that cause these syndromes reside within the FGFR3 kinase domain, and result in varying degrees of constitutive receptor activation. The N540K substitution, located proximal to the split tyrosine kinase domain, underlies the mild skeletal dwarfism hypochondroplasia, and confers weak constitutive activation (15,18). At the other end of the spectrum, the K650E substitution located within the activation loop of the kinase domain, relieves the normal requirement for regulatory phosphorylation at Y647 and Y648 and leads to profound constitutive kinase activation in comparison to wild-type FGFR3 (19). This mutation causes thanatophoric dysplasia type II (TDII), a neonatal lethal dwarfism syndrome (6,19). Recently, a different activating substitution at this same position, K650M, has been associated with the syndrome SADDAN, or Severe Achondroplasia with Delayed Development and Acanthosis Nigricans (12).
Abnormal activation of FGFR3 as a result of somatic mutation has been reported in conjunction with several human cancers, including multiple myeloma, cervical carcinoma, and bladder carcinoma (7,8,20). The specific FGFR3 mutations involved include K650E and K650M in the kinase domain, or R248C, S249C, G370C and Y373C, in the extracellular domain. All of these mutations identified in human neoplasia have been previously described as activating mutations associated with TDI, TDII or SADDAN (6,19,21,22).
These phosphotyrosine residues provide specific binding sites for signaling proteins containing Src homology 2 (SH2) domains or phosphotyrosine-binding (PTB) domains (26,27). For example, Y766 in FGFR1 has been shown to interact with phospholipase C-γ (PLC-γ) (28). The adapter protein fibroblast growth factor receptor substrate 2 (FRS2) has also been shown to associate with FGFR1 (29)(30)(31). Activation of FGFR1 leads to tyrosine phosphorylation of FRS2 at several sites, leading to recruitment of Grb2 (32). Besides PLC-γ and FRS2, little is known about substrates of FGFRs that lead to mitogenesis and differentiation.
Given the importance of understanding FGFR3-mediated signaling, both for human developmental syndromes and also for those human cancers where FGFR3 activation has been observed, we wished to identify novel FGFR3-interacting proteins that may represent important substrates for downstream signaling. Towards this end, we employed a yeast two-hybrid screen in which the bait was the kinase domain from either wild-type FGFR3, or from an activated mutant. Using the weakly activated N540K mutant as the bait, we were able to identify four candidate binding proteins that interact with FGFR3, one of which is the adapter protein SH2-B.
In this study, we have identified and characterized SH2-B as an FGFR3 binding partner.
We also demonstrate that activated FGFR3 can directly phosphorylate SH2-Bβ. In addition, we show that expression of SH2-Bβ together with activated FGFR3 increases Stat5B phosphorylation in 293T cells. Stat5B was also observed to relocalize exclusively to the nucleus upon FGFR3-mediated signaling through SH2-Bβ. Our data thus suggest that the adapter protein SH2-Bβ may represent an important signaling molecule that mediates downstream biological effects of FGFR3 activation.
Yeast Two-Hybrid Screen -A yeast two-hybrid screen was performed according to previously published protocols (50,51). The two-hybrid plasmids pBTM116, pVP16 and LexA-  Liquid Culture β-Galactosidase Assay -Yeast were cotransformed with the indicated plasmids and grown on His↑ plates for 3-4 days at 30°C. Liquid culture β-galactosidase assays were then performed according to the protocol provided by Clontech using o-Nitrophenyl β-Dgalactopyranoside (ONPG) (Sigma). Each reaction was carried out at 30°C until the sample became yellow. Samples that did not develop a yellow color were stopped at the end of the fourth hour. The absorbance of each sample was measured at 420 nm, and β-galactosidase activity was calculated using the following formula: β-galactosidase units = 1000 x OD 420 /(t x V x OD 600 ), where t is elapsed time (min) of incubation, V is 0.1 ml x concentration factor, and OD 600 is the absorbance of 1 ml culture at 600 nm. The data obtained were the results of five independent experiments.
Immunofluorescence -2C4 cells were cultured in 10% heat-inactivated FBS containing 80 µg/ml G418 in 5% CO 2 . Cells were plated into 60 mm plates containing glass coverslips at a density of 5 x 10 5 . The next day, the cells were transfected with a total of 2.1 µg DNA of the indicated constructs using Effectene (Qiagen) according to the manufacturer's directions. DNA ratios for the triple transfection were: 0.1 µg GFP-Stat5B, 0.5 µg FGFR3, and 1.5 µg myc-SH2-Bβ. pcDNA3 was added to make up the difference in the single and double transfections.
Twenty-four hours post-transfection, the coverslips were fixed with 3% paraformaldehyde. To visualize protein localization, the cells were permeabilized with 0.5% Triton-X, rinsed with PBS, and blocked with 3% BSA for 30 min. The cells were incubated with FGFR3 antibody (1:500) for 1 hour, washed, then incubated with 1:500 rhodamine-conjugated anti-rabbit secondary antibody for 45 min. After washing with PBS, the coverslips were mounted onto glass slides with 90% glycerol in 0.1 M Tris-HCl (pH 8.5) plus phenylenediamine to prevent fading.
Cells were photographed using a Nikon Microphot-FXA microscope with a Hamamatsu C5810 camera.

RESULTS
Identification of an FGFR3-Interacting Protein -A yeast two-hybrid screen was employed to identify potential substrates of FGFR3. To construct the bait for the two-hybrid screen, the entire intracellular domain of FGFR3, was fused to the LexA DNA-binding domain.
These constructs were used as two-hybrid baits and screened against a 9.5 dpc mouse embryonic cDNA library. Cotransformed yeast were plated on His↑ plates, and the activation of  (Table 1). PLC-γ, which was also isolated from the screen, was used as a positive control since it has been shown to bind to Y766 of FGFR1, which corresponds to Y760 in FGFR3 (28). This yeast two-hybrid interaction demonstrates for the first time a direct interaction between FGFR3 and PLC-γ. The empty pVP16 vector was used as an additional negative control. Thus, we demonstrate a novel interaction between the SH2 domain of SH2-B and FGFR3.    Since both FGFR3 and SH2-Bβ migrate between 116 kDa and 97.5 kDa on 10% SDS-PAGE, we wanted to rule out the possibility that the bands shown in the top panel of Figure 5A were tyrosine phosphorylated FGFR3. To accomplish this, we utilized a truncated R3-K650E construct which contained only the intracellular domain and a myristylation signal at the Nterminus (myr-R3-K650E) to properly localize the protein to the plasma membrane, and which we have extensively characterized in previous studies (22,48). This construct runs at ~50 kDa.  Figure 5B shows that myc-SH2-Bβ was expressed in the appropriate samples (lanes 2 and 4). Identification of SH2-B Binding Sites in FGFR3 -Seven autophosphorylation sites in FGFR1 have been described previously (23,28). Based on the sequence alignment with FGFR1, the potential autophosphorylation sites (Y577, Y647, Y648, Y724, and Y760) in FGFR3 were identified. In addition, Y770 in FGFR3 is conserved throughout FGFRs, suggesting that it may be a potential autophosphorylation site as well. A schematic diagram of the intracellular domain of FGFR3 with six potential autophosphorylation sites is shown in Figure 7A. Y647 and Y648 are located in the activation loop, and their phosphorylation is involved in the conformational changes that accompany receptor activation (49). Previously we showed that substitution of all non-activation loop Y residues with F caused FGFR3 to be inactive (49). We therefore focused on residues Y724, Y760, and Y770. To determine the tyrosine residue(s) of FGFR3 required for SH2-B binding, we generated a series of LexA-R3-N540K mutants containing phosphorylation-site mutations in which the tyrosine residues were mutated to phenylalanine.
LexA-R3-N540K was used as the template for these mutants ( Figure 7B), since this derivative was used as bait in the initial yeast two-hybrid screen.
Each mutant was tested against the SH2-B region isolated from the screen by βgalactosidase liquid assay (Table 2). PLC-γ was used as a positive control due to its previously characterized interaction with FGFR1 (28). Among the mutants, LexA-770F exhibited the strongest interaction with SH2-B (Table 2). This mutant lacks Y770, but retains Y724 and Y760. This suggests that one or the other, or both, of these residues is important for the interaction with SH2-B. The LexA-724F mutant, which retains Y760 and Y770 but lacks Y724, showed a decrease in binding. In addition, the LexA-760F mutant, which retains Y724 and Y770 but lacks Y760, also exhibited a significantly decreased interaction. These data indicate that Y724 and Y760 are important for SH2-B binding.
The results presented in Table 2 also demonstrate that LexA-R3-N540K exhibits significant interaction with PLC-γ. Interestingly, when Y770 was removed by mutation, the resulting mutant LexA-770F exhibited increased interaction with PLC-γ, and to a lesser extent with SH2-B. This observation could suggest that Y770 plays a negative regulatory role in FGFR3 signaling, at least with regard to signaling through the effector proteins SH2-B and PLC-γ. When GFP-Stat5B (47) was cotransfected with wild-type FGFR3, Stat5B was predominantly cytoplasmic (Figure 9, panel e). Cotransfection of GFP-Stat5B with the activating mutant, R3-K650E, resulted in an increase in nuclear localization of Stat5B (Figure 9, panel h). When SH2-Bβ was also transfected with R3-K650E and GFP-Stat5B, Stat5B relocalized completely into the nucleus (Figure 9, panel k). In the presence of the mutant SH2-Bβ(R555E), R3-K650E and GFP-Stat5B, however, Stat5B was localized to the cytoplasm ( Figure  9, panel n). These data correlate with the increase in phosphorylation observed in Figure 8 upon coexpression of activated FGFR3 and SH2-Bβ, and a decrease in phosphorylation upon coexpression of activated FGFR3 and SH2-Bβ (R555E).

DISCUSSION
Numerous skeletal and developmental disorders have been shown to result from mutations in FGFRs (6,56,57). Mutations in FGFR3 result in many human disorders, including TDI, TDII, SADDAN, and dwarfism (6,21). More recently, FGFR3 has been linked to cancers such as multiple myeloma and bladder and cervical carcinoma (7,8). Unlike other RTKs, few of the immediate downstream signals of FGFR3 have been identified. In this study, we have identified SH2-B as an FGFR3 binding protein using the yeast two-hybrid screen.
Coimmunoprecipitation experiments in 293T cells between FGFR3 and SH2-B confirmed this interaction. We have also found by β-galactosidase liquid assay that Y724 and Y760 in FGFR3 interact with the SH2 domain of SH2-B. In addition, activated FGFR3 promotes binding to SH2-Bβ and stimulates tyrosine phosphorylation of SH2-Bβ. Furthermore, the kinase domain of FGFR3-K650E can directly phosphorylate SH2-B in vitro. These results suggest that SH2-B is a direct substrate of FGFR3.
Previously, PLC-γ represented the only SH2 domain-containing binding partner for FGFRs (28). The SH2-B clone isolated from the yeast two-hybrid screen primarily contains the SH2 domain ( Figure 1B), demonstrating that SH2-B binds to FGFR3 via its SH2 domain.
Consistent with this, the SH2 domain alone was sufficient to bind to activated FGFR3 in mammalian cells. Since SH2 domains have been found to bind to phosphotyrosine residues (58), it is likely that SH2-B interacts with the phosphotyrosine residue(s) of FGFR3.
Coimmunoprecipitation experiments demonstrated that binding of the SH2-Bβ isoform to FGFR3 correlated with receptor activation, which further supports that the interaction is phosphotyrosine dependent. This also corresponds to the fact that a N540K mutation leads to a milder form of activated FGFR3 than a K650E mutation. Interestingly, our data indicate that full-length SH2-Bβ can also weakly associate with wild-type FGFR3 (Figure 4). One possible explanation for this observation is the existence of an additional low-affinity binding site(s) in SH2-B for FGFR3. In fact, SH2-Bβ has been shown to bind to tyrosine phosphorylated JAK2 not only via its SH2 domain but also by its N-terminal region (amino acids 1 to 555); however, this binding is not phosphotyrosine-dependent (59).
We also show that both Y724 and Y760 of FGFR3 are required for interaction with the SH2 domain of SH2-B. Mutating these residues to phenylalanine impairs the association between FGFR3 and the SH2 domain. Previous studies from our laboratory have shown that a derivative of FGFR3 containing all conserved tyrosine residues stimulated transformation, Stat activation, and phosphatidylinositol (PI) 3-kinase activation (49). Substitution of all nonactivation loop tyrosine residues with phenylalanine rendered the FGFR3 derivative inactive; however, the addition of Y724 restored its ability to stimulate the above signaling pathways (49).
Data from our two-hybrid screen also indicate that the SH2 domain of the p85 regulatory subunit of PI 3-kinase interacts with FGFR3 (data not shown). It is possible that the SH2 domains of p85 and SH-2B compete for binding of Y724 in FGFR3, resulting in the activation of different signaling pathways. We have also previously shown that both Y724 and Y760 were required for maximal Stat activation (49). Similarly, both tyrosine residues may be necessary to mediate the signaling events carried out by SH2-B in response to FGFR3 activation.
Nonetheless, the possibility that the SH2 domain interacts with only Y724 or Y760 should not be excluded.
We also demonstrate by an in vitro kinase assay that the SH2 domain of SH2-Bβ can be directly phosphorylated by activated FGFR3. Since SH2-Bβ has 9 tyrosine residues, some of these residues may be directly phosphorylated by FGFR3 while others may be phosphorylated by non-receptor tyrosine kinases such as JAK2 (35). Among these residues, Y439, Y494, and Y624 were predicted to be potential tyrosine phosphorylation sites (33). The Y624 residue in particular is located within a YVPS motif, which is a putative phosphorylation site by PDGF receptor (33).
Since the SH2-B clone isolated from the screen contains the entire SH2 domain as well as the Y624 residue, it would be interesting to determine if FGFR3 specifically phosphorylates Y624.
There are two other tyrosine residues, Y525 and Y564, in the SH2 domain of SH2-Bβ which may also serve as phosphorylation sites. Future studies will determine which residue(s) is specifically phosphorylated by FGFR3. Identification of the tyrosine residues phosphorylated by FGFR3 will be important in studying the signaling pathway(s) mediated by SH2-Bβ as they will suggest molecular mechanisms of recruitment of signaling proteins by FGFR3 activation via specific tyrosine phosphorylation.
In this study, we have characterized a novel interaction between FGFR3 and SH2-Bβ and we have demonstrated that FGFR3 activation results in the tyrosine phosphorylation of SH2-Bβ.
We also demonstrate one mechanism by which FGFR3 mediates downstream signaling. In 293T cells, we observe phosphorylation of Stat5 in cells transfected with activated FGFR3. This phosphorylation increases in response to expression of SH2-Bβ. We also demonstrate the nuclear translocalization of Stat5B when both activated FGFR3 and SH2-Bβ are coexpressed.
When the mutant SH2-Bβ(R555E) was coexpressed with activated FGFR3, Stat5B becomes inactivated and is predominantly cytoplasmic. SH2-B therefore appears to represent a novel and biologically relevant substrate of FGFR3.