Identification of Y589 and Y599 in the juxtamembrane domain of Flt3 as ligand-induced autophosphorylation sites involved in binding of Src family kinases and the protein tyrosine phosphatase SHP2

Fms-like tyrosine kinase 3 (Flt3) functions as a growth factor receptor and is expressed primarily in multipotential hematopoietic stem cells and progenitors as well as in placenta, gonads, and brain. Together with its activating ligand Flt3 ligand (FL) it is a crucial player in assuring normal function of stem cells and the immune system.^1-3  Moreover, approximately 30% to 35% of patients with acute myeloid leukemia (AML) carry a mutation in Flt3, rendering Flt3 the most frequently mutated gene in AML.^4,5  Flt3 mutations are generally grouped into 2 classes: point mutations in the vicinity of codon 835 or 842 within the tyrosine kinase domain (TKD); or internal tandem duplications (ITDs) of varying lengths within the juxtamembrane domain of Flt3, which sterically represses the intrinsic kinase activity of Flt3 in the absence of ligand.^4,6  Both classes of mutations result in constitutive activation of Flt3 but distinct signaling and transforming capacities.^7-10  Although debatable as prognostic markers by themselves, ITD and TKD mutations are correlated with poor prognostic features for AML patients, suggesting Flt3 or one of its downstream effectors as potential therapeutic targets.^11-15 

Flt3 constitutes, together with the receptor for stem cell factor (c-Kit), the receptors for platelet-derived growth factors (PDG-FRs), and colony-stimulating factor-1 (CSF-1), the type III family of receptor tyrosine kinases (RTKs).⁵  Type III RTKs share a common modular structure consisting of 5 extracellular Ig-like domains, a short transmembrane stretch, a juxtamembrane region followed by a bipartite kinase domain interrupted by the kinase insert, and the carboxyterminal tail.¹⁶  Ligand binding causes receptor dimerization, kinase activation, and transphosphorylation of RTK on multiple tyrosine residues.¹⁷  These autophosphorylated tyrosine residues together with 3 to 6 adjacent amino acids form high-affinity docking sites for relay molecules possessing either phosphotyrosine binding (PTB) or Src homology 2 (SH2) domains.¹⁸  Upon relocation to the receptor, these signaling or adapter molecules become activated in either a phosphorylation-dependent or -independent manner and are thereby capable of transducing the signal downstream. Relay molecules reported to be recruited and/or activated upon Flt3 activation include the p85 subunit of PI3K, Ras-GAP, PLC-γ, Vav, SHIP1, SHP2, ShcA, Grb2, Cbl, and Src family kinases (SFKs) as well as Stat5.^19-26  Whereas the immediate signaling steps following ligand binding (ie, binding of signaling molecules to autophosphorylated tyrosines) are well studied in the c-Kit, PDGFRs, and CSF-1 receptor systems,¹⁶  not a single in vivo autophosphorylation site of Flt3 or its respective binding partners have been reported. Most ITD isoforms carry a duplication containing at least one of the juxtamembrane tyrosine residues Y589, Y591, Y597, or Y599, which could theoretically add to the aberrant signal relay from the autoactivated receptor.²⁷ 

Here we report that Y572, Y589, Y591, and Y599 of Flt3 are phosphorylated in vivo in Flt3-expressing cells following ligand stimulation. Furthermore, we demonstrate that phosphorylated Y589 and Y599 are docking sites for signaling intermediates including SFKs. pY589 has a negative regulatory effect most likely due to recruitment and activation of SFKs via pY589. In contrast, pY599 appeared to be one of the major tyrosine residues with a stimulatory function upon FL binding. Mutation of tyrosine 599 to a phenylalanine residue decreased FL-mediated proliferation and/or survival upon interleukin-3 (IL-3) withdrawal in Flt3-32D cells. Apart from a possible contribution of SFKs to the Y599F phenotype, we provide data suggesting that the protein tyrosine phosphatase SHP2 plays a role. In Y599F-Flt3-32D cells, SHP2 binding to Flt3 is dramatically reduced, as is ligand-stimulated tyrosine phosphorylation of SHP2. Furthermore, silencing of SHP2 impaired Erk phosphorylation in ligand-stimulated wild-type Flt3 (WT-Flt3) cells, which is consistent with other reports stressing the importance of SHP2 for Erk activation downstream of RTKs.²⁸  Interestingly, mutated hyperactivated SHP2 has been found in some forms of leukemia including Flt3-ITD-positive AML. ^29-32 

pMSCV-neo vector containing human Flt3 cDNA was a kind gift from Dr G. Gilliland (Brigham and Women's Hospital, Harvard Medical School, Boston, MA). For transfections of adherent cells, polymerase chain reaction (PCR)-amplified cDNA of human Flt3 was subcloned into KpnI-BamHI restriction sites of pcDNA 3.1 expression vector (Invitrogen, Carlsbad, CA). Murine SHP2 SMARTpool siRNA duplexes (proprietary target sequences) were obtained from Dharmacon (Lafayette, CO). Recombinant human FL and murine IL-3 were purchased from Prospec Tany (Rehovot, Israel). The anti-Flt3 antibody was raised against a synthetic peptide corresponding to amino acids (aa's) 740-757 of human Flt3 and affinity purified as described.³³  Peptide-specific antibodies against individual tyrosine phosphorylation sites in Flt3 were raised by immunizing rabbits with synthetic peptides corresponding to aa's 583-597 (CTGSSDNE(p)YF-(p)YVDFRE; pY589/pY591) and aa's 591-605 (CYVDFRE(p)YE(p)YDLK-WEF; pY597/pY599) of human Flt3 and affinity purified as described.³⁴  Specificity of the antibodies was confirmed in immunoblots with WT and the respective Y→F mutant of Flt3. A rabbit antiserum recognizing Erk-2 has been described elsewhere.³⁵  Anti-pErk (p42/p44) antibody was from Cell Signaling Technology (Beverly, MA). Anti-SHP2, anti-Fgr, anti-Lyn, anti-Hck, and anti-PY99 antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). HRP-coupled secondary antimouse and antirabbit antibodies were from Pierce (Rockford, IL). GST fusion proteins containing the SH2 domains of either Src or SHP2 were kind gifts from Dr T. Pawson (Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, ON, Canada).

To mutate specific tyrosines to phenylalanines, the QuikChange mutagenesis XL kit (Stratagene, La Jolla, CA) was employed according to the manufacturer's instructions. All mutations were subsequently confirmed by sequencing.

Porcine aortic endothelial (PAE) cells³⁶  were cultured in Ham F-12 medium, whereas Cos-1 and EcoPack cells were cultured in Dulbecco modified essential medium (DMEM; Gibco Invitrogen, Carlsbad, CA). Media were supplemented with 10% FCS, 100 units/mL penicillin, and 100 μg/mL streptomycin. Promyeloid 32D cells (German Collection of Microorganisms and Cell Cultures [DSMZ], Braunschweig, Germany) were maintained in RPMI-1640 medium plus 10% heat-inactivated FCS, 10 ng/mL recombinant murine IL-3, 100 units/mL penicillin, and 100 μg/mL streptomycin.

The 32D cells were starved by withdrawing growth factors for 4 hours. All other cell lines were starved by maintaining them overnight in medium with only 0.5% serum. Flt3 was stimulated with 100 ng/mL FL for the indicated periods of time unless stated otherwise.

Inhibitor concentrations for the appropriate experiments were 100 μg/mL cycloheximide (Sigma-Aldrich, St Louis, MO), 10 μM PP1, or 15 μM PP2 (Calbiochem, EMD Biosciences, Darmstadt, Germany), respectively.

The 32D cells were metabolically labeled by incubation in methionine/cysteine-free DMEM medium (Gibco Invitrogen) with 100 μCi (3.7 MBq) Promix (Amersham Biosciences, Uppsala, Sweden) at 37°C for 4 hours before lysis and further processing.

Adherent cells were transfected using Lipofectamine/PLUS or Lipofectamine 2000 reagents (Invitrogen) according to the manufacturer's instructions. In order to establish 32D cells (no endogenous Flt3) stably expressing WT or mutant Flt3, packaging EcoPack cells were transfected with the corresponding pMSCV-Flt3 constructs, and virus-containing supernatants were collected 72 hours after transfection. Retroviral infection of 32D cells was followed by a 2-week selection in 0.2 mg/mL G418. For all experiments, pools of retrovirally infected cells were used in order to average out possible interclonal variation.

Silencing of SHP2 in Flt3-32D cells was achieved by electroporation (300V, 1500 μF) in a Gene PulserII (Bio-Rad, Hercules, CA) in the presence of 100 nM SHP2-siRNA and incubation for another 72 hours before experiments were performed.

Flt3-32D transfectants were labeled with PE-anti-human CD135 or PE-mouse IgG₁ κ (BD Biosciences Pharmingen, Heidelberg, Germany) as isotype control according to standard protocols. Flow cytometry was performed with a FACSort instrument (BD Biosciences, San Jose, CA).

To assess the number of living Flt3-32D cells upon IL-3 withdrawal in the absence and presence of FL, cells were washed twice in IL-3-free medium, seeded in 96-well plates (30 000 cells/100 μL IL-3-free medium and well), and treated with different concentrations of FL for 48 hours. Then an MTT assay (Sigma-Aldrich) was performed according to the manufacturer's instructions. Triplicate samples from 3 individual experiments were subjected to analysis of variance (ANOVA; GraphPad Prism, 4.0; San Diego, CA).

Immunoprecipitation, GST pulldown, and Western blotting were conducted as described elsewhere.^34,35  Immunodetection was performed by enhanced chemiluminescence using Super Signal Dura reagent (Pierce) and a CCD camera (LAS-3000; Fujifilm, Tokyo, Japan). Bands were quantified by MultiGauge software (Fujifilm).

Peptide synthesis and purification were performed as described.³⁷  An additional cysteine residue was added in the amino terminus to allow coupling to KLH using m-maleinimido-benzoyl-N-hydroxy-succinimide ester (MBS) coupling. The following peptides were synthesized: Y589 (CTGSSDNEYFYVDFRE, corresponding to aa's 583-597 of human Flt3), pY589 (CTGSSDNE(p)YFYVDFRE), Y599 (CYVDFREYEYDLKWEF, corresponding to aa's 591-605 of human Flt3), and pY599 (CYVD-FREYE(p)YDLKWEF). As control for specificity, a peptide corresponding to the PI3-kinase binding site in human c-Kit, pY721, was used (CSDST-NE(p)YMDMKPGV). Peptides were dissolved at 1 mg/mL in DMSO and immobilized on Affigel-10 (Bio-Rad) according to the manufacturer's instructions. Slurry (100 μL) was incubated in the cold for 2 hours with precleared 32D cell lysates. For competition experiments, 0.5 mg/mL of the indicated soluble peptide was included to the reaction mixture. Peptide-bound proteins were either subjected to autoradiography (³⁵S-labeled cells) or processed for immunoblotting.

In vitro kinase activity of Flt3 and Lyn were conducted essentially as described in Voytyuk et al³⁴  and Lennartsson et al.³⁵  After SDS-polyacrylamide gel electrophoresis (SDS-PAGE) separation and electroblotting, radioactive proteins were localized on a PhosphorImager (FLA3000; Fujifilm) and densitometrically quantified by MultiGauge software when needed.

Flt3-PAE cells were ³²P labeled in phosphate-free DMEM (Gibco Invitrogen) supplemented with 0.5% dialyzed FCS, 10 mM HEPES, and 2 to 6 mCi (74-222 MBq) ³²P-orthophosphate (PBS-43; Amersham) for 6 to 7 hours at 37°C and subsequently treated as described elsewhere.³³ 

In order to investigate autophosphorylated tyrosine residues in the juxtamembrane region of Flt3 (aa's 572-603) we labeled Flt3-PAE cells with [³²P]orthophosphate, isolated Flt3 after ligand stimulation, and performed 2-dimensional phosphopeptide mapping as described previously.³³  However, the obtained maps showed a phosphoserine content as high as 95% that interfered with the identification of any tyrosine as autophosphorylation site (data not shown). Therefore, we chose to selectively immunoprecipitate the tryptic peptides of interest, aa's 572-595 and aa's 596-602, from a complete tryptic digest of labeled Flt3 using peptide-specific antibodies. Each isolated tryptic fragment was subjected to 2-dimensional phosphoamino acid analysis and automated Edman degradation, and the radioactivity released in each cycle was quantified. Phosphoamino acid analysis of the tryptic peptide consisting of aa's 572-595 of Flt3 revealed the presence of phosphoserine and phosphotyrosine (Figure 1Ai). Radiosequencing of the peptide showed a peak of radioactivity at positions 1 and 3. In order to analyze potential phosphotyrosine residues lying in the C-terminal of amino acid 581, the material was further digested with AspN, which cleaves the N-terminal of aspartic acid residues. Sequence analysis of AspN-cleaved material pooled with non-cleaved material displayed peaks at positions 1, 3, 4, and 6, providing the in vivo evidence for autophosphorylation of Y572, Y589, and Y591 in addition to S574 as phosphorylation sites in human Flt3. In vivo labeling of the respective Y→F mutants of Flt3 and subsequent disappearance of the corresponding radioactivity peak upon Edman degradation confirmed the identity of pY572, pY589, and pY591 as in vivo autophosphorylation sites (Figure 1A). The tryptic peptide covering aa's 596-602 of WT-Flt3 had a phosphotyrosine at position 4 that was not detected in Y599F-Flt3, demonstrating in vivo phosphorylation of Y599 (Figure 1B).

To investigate the role of individual phosphotyrosines, we focused on pY589 and pY599. pY589 is homologous to pY568 in c-Kit and pY579 in PDGFR-β where they have been reported to bind and activate SFKs and to be pivotal for ligand-stimulated proliferation, differentiation, and migration.^35,38-40  To elucidate the impact of phosphorylation at Y589 and Y599 on Flt3-mediated signaling, we mutated each tyrosine residue individually to phenylalanine and stably introduced WT and mutant Flt3 isoforms into promyeloid 32D cells and examined FL-induced responses. Equal surface expression of each isoform in the obtained pools of transfectants was shown by flow cytometry (Figure 2A). Intact kinase activity upon ligand stimulation was confirmed by in vitro kinase assays of immunoprecipitated active Flt3 with Flt3 as endogenous substrate (Figure 2Aii) and a comparable phosphotyrosine content in immunoprecipitated Flt3 from living FL-stimulated cells (Figure 2Aiii).

When comparing ligand-dependent receptor activation over time, Y589F-Flt3-32D and Y599F-Flt3-32D cells showed sustained receptor phosphorylation compared with WT-Flt3-expressing cells (Figure 2B). Y589F-Flt3-32D cells also showed a prolonged ligand-mediated p42/44-Erk activation (Figure 2D). Moreover, Y589F-Flt3-32D cells displayed consistently higher numbers of living cells in response to FL treatment than WT-Flt3-32D cells (Figure 2C), employing the MTT assay as readout for cell proliferation and/or survival upon IL-3 withdrawal. Y599F-Flt3-32D cells, however, were impaired in their response to FL treatment in terms of IL-3-independent proliferation (Figure 2C-D) and Erk activation. Parental 32D cells expressing no Flt3 did not show any rescue by FL from IL-3 withdrawal (data not shown).

Intrigued by the distinct phenotype of the Y589F and Y599F mutants of Flt3, we wanted to elucidate binding partners of pY589 and pY599 that account for the observed phenotype in the Y→F mutants. For this we chose an affinity chromatography approach employing immobilized synthetic phosphopeptides covering aa's 583-pY589-597 (pY589) or aa's 591-pY599-605 (pY599) of Flt3, respectively. As negative control for unspecific, phosphorylation-independent binding we used immobilized peptides without phosphorylation on the tyrosine residues of interest. As additional control for specificity, a phosphopeptide corresponding to the PI3-kinase binding site in human c-Kit (pY721) was used (data not shown). Lysates of [³⁵S]-labeled 32D cells were incubated with the corresponding peptides, and bound proteins were visualized by the PhosphorImager after separation by SDS-PAGE. Proteins of the approximate sizes of 45, 52, and 55 kDa were found to interact in a phosphospecific manner with pY589 (Figure 3A, proteins a-c), whereas pY599 showed specificity for proteins of approximately 45, 52, 55, and 70 kDa (proteins d-g). For subsequent identification of the potential interaction partners of pY589 and pY599, we employed candidate immunoblotting. As pY589 is homologous to the Src-binding site Y568 in c-Kit, we were prompted to investigate whether Src also binds to pY589 of Flt3. From the SFK members mainly expressed in 32D cells (Lyn, Fgr, and Hck), all were found to interact with pY589 (Figure 3Bi). Surprisingly, pY599 was also able to recruit Lyn, Fgr, and Hck despite the absence of the pYEEI consensus motif for the SH2 domain of Src.⁴¹  However, Src interaction with both sites within the entire Flt3 protein was confirmed by pulldown experiments employing a fusion protein of GST and the SH2 domain of Src. Phosphorylation-dependent binding of Src-SH2 to Flt3 was weakened if either Y589 or Y599 of Flt3 were mutated to a phenylalanine residue (Figure 3C). Coimmunoprecipitation of endogenous Lyn, Fgr, or Hck, respectively, with the WT-Flt3 upon ligand stimulation verified the observed interaction in living cells. In contrast, association was greatly reduced in 32D cells expressing either the Y589F mutant or the Y599F mutant of Flt3 (Figure 3D). Interestingly, Fgr, and to some extent Hck, showed preference for binding to Y589. In order to study the functional consequences of this interaction, 32D cells expressing either the wild-type, Y589F, or Y599F Flt3 were stimulated with FL and lysed. Lyn, being the most prominent representative of SFKs in 32D cells and the SFK evenly recruited to Y589 and Y599, was immunoprecipitated and subjected to an in vitro kinase assay with acid-denatured enolase as exogenous substrate. In cells expressing wild-type Flt3, Lyn kinase activity was stimulated approximately 3-fold (Figure 3E). In contrast, Lyn immunoprecipitated from 32D cells expressing either Y589F or Y599F mutant of Flt3 showed a dramatic decrease in the degree of Lyn kinase activation. Consistently, phosphorylation of Cbl, a well known target of SFK,⁴²  was impaired in FL-stimulated Y589F-Flt3 and Y599F-Flt3 cells (Figure 3F).

Additionally, an observed interaction between purified GST-Src (SH2) and immobilized pY589 or pY599 (data not shown) suggested that SFKs are recruited directly to both pY589 and pY599.

Due to the divergent proliferative behavior of the Y589F and Y599F mutants, SFKs seemed unlikely candidates evenly accounting for both observed phenotypes unless we assume site-selective action of SFKs. For a variety of different receptor systems, the protein tyrosine phosphatase SHP2 has been reported to be crucial for Erk activation and thus to be involved in proliferation and survival.²⁸  This led us to probe for SHP2 among the phosphopeptide pY599-binding proteins and to reveal phosphospecific interaction of SHP2 with pY599 (Figure 4A). The band of about 70 kDa pulled down by pY599 from ³⁵S-labeled cell lysate (Figure 3A, protein d) is therefore suggested to be SHP2. Interaction of SHP2 with pY599 of the entire Flt3 protein was confirmed by a pulldown experiment using a fusion protein of GST with the N- and C-terminal SH2 domains of SHP2. WT-Flt3 and Y589F-Flt3 interact with SHP2 in a ligand-dependent manner, whereas Y599F-Flt3 cannot bind to SHP2-SH2 (Figure 4B). Interaction between SHP2 and Flt3 in living cells was confirmed by coimmunoprecipitation of SHP2 with Flt3 from FL-stimulated 32D cells expressing WT-Flt3 (Figure 4C). In contrast, no interaction was seen between SHP2 and Flt3 in 32D cells expressing the Y599F mutant of Flt3, whereas it was intact in the Y589F mutant. Furthermore, Figure 4D shows lack of SHP2 phosphorylation in Y599F-Flt3 cells that occurs in WT-Flt3 cells. This again stresses the specific importance of pY599 for SHP2 recruitment and action. Y589 and Src-inhibited WT-Flt3-32D cells were able to phosphorylate SHP2 to a comparable level upon FL stimulation as seen in WT-Flt3-32D cells (data not shown). Thus, a contribution of SFK to SHP2 phosphorylation is unlikely. The sequence surrounding Y599 is different from the consensus sequence for the SH2 domain of SHP2 (pYIXV⁴¹  vs 599 pYDLK in Flt3). However, the alternative consensus sequence for binding of SHP2, as proposed by Rönnstrand et al⁴³  (pYAD/EI/L), is similar to the sequence surrounding Y599, although shifted one position. There are other examples in the literature where the peptide sequence in a protein binding to an SH2 domain has been shifted one position compared with the predicted consensus.⁴⁴  We propose that binding of SHP2 to pY599 is direct, as we observed an interaction of purified SHP2-GST fusion protein with immobilized pY599 that could be competed by an excess of soluble pY599 peptide (Figure 4E).

Overall, our data imply that both Y589 and Y599 are autophosphorylation sites that constitute sites of receptor interaction for several signal transduction molecules including SFKs and SHP2.

As demonstrated earlier, Y599 plays an important role for FL-mediated Erk activation and growth in Flt3-32D cells. We next aimed to elucidate the individual impact of SFKs and SHP2 on the pY599 phenotype. Employing the Src-selective inhibitors PP1 or PP2, we could not observe any major negative effect on FL-triggered Erk activation in WT-Flt3-32D cells (Figure 5A). In contrast, PP1 and PP2 even slightly prolonged Erk activation in FL-stimulated WT-Flt3-32D cells as seen in Y589F-Flt3 cells. Furthermore, treatment of cells with the MEK inhibitor U0126 decreased FL-mediated cell growth, demonstrating an importance of activation of the Erk pathway for FL-induced cell growth (data not shown). These data suggest that SFKs do not play a dominant role for FL-triggered and Erk-mediated growth-promoting signals in Flt3-expressing 32D cells.

Parallel Trypan Blue exclusion counting of the different cell lines under the culture conditions described confirmed the obtained MTT data (data not shown).

Silencing of SHP2 by siRNA led to a reduction of SHP2 levels by approximately 50% to 60% in WT-Flt-32D and Y599F-Flt3-32D cells (Figure 5Bi). Examining ligand-mediated activation of Erk in WT-Flt3-32D, WT-Flt3-(SHP2 knockdown [KD])-32D, Y599F-Flt3-32D, and Y599F-Flt3-KD-32D cells, we observed that reduction of SHP2 levels led to lower levels of pErk, implying the involvement of SHP2 in FL-mediated Erk activation. WT-Flt3-KD and Y599F-Flt3 cells furthermore responded similarly, which corroborates our findings concerning the interaction between SHP2 and pY599 (Figure 5Bii-iii). Looking at FL-triggered proliferation/survival, WT-Flt3-KD and Y599F-Flt3 cells again displayed a comparable behavior, and down-regulation of SHP2 was able to diminish mitogenicity in both WT and Y599F-Flt3-32D cells (Figure 5C).

Therefore, we conclude that lack of SHP2 binding, rather than Src binding, accounts for the observed Y599F-Flt3 phenotype in 32D cells and that SHP2 is a major player in FL-induced Erk activation.

We show here for the first time that Y572, Y589, Y591, and Y599 in the juxtamembrane region of Flt3 are phosphorylated in vivo, which could provide valuable insights in how Flt3 mediates its very early signaling steps. The more AML-related ITD mutations of Flt3 often contain at least duplication of one of the juxtamembranous tyrosines. We further demonstrate that Y589 and Y599 are binding sites for downstream signaling molecules including SFKs and the protein tyrosine phosphatase SHP2.

Tyrosine 589 appeared to be involved mainly in the negative regulation of Flt3 signaling. Like its homologue in c-Kit, it was found to interact with SFKs, which constitute a family of nonreceptor tyrosine kinases and share a common modular structure consisting of the catalytic SH1, an SH2, an SH3, and a C-terminal regulatory domain that renders SFKs ideal for transducing signals from autophosphorylated receptors to downstream targets.⁴⁵  Depending on which molecules are recruited to or activated by SFKs, they can have positive effects via Erk activation or a negative impact via Cbl and ubiquitination/receptor degradation on receptor signaling.^46,47  Mutation of Y589 in Flt3 resulted in sustained ligand-mediated Erk activation and elevated proliferative/survival response consistent with the findings in the Src binding mutant of the CSF-1 receptor.⁴⁸  A recent report stresses the importance of Lyn for positive FL-mediated signaling in Ba/F3 cells expressing WT-Flt3 but denies its involvement in receptor degradation.²⁴  This apparently contradicts our observations of a stabilization of the Flt3 signal. However, the use of different cell systems and experimental approaches may explain the discrepancies. Moreover, one can think of specific functions of individual Src family members. The latter effects are easily missed by general Src inhibition but may become apparent by specifically mutating single recruitment sites as in Y589F-Flt3. Still, one cannot exclude contributions of other signaling molecules besides SFK to the observed Y589F-Flt3 phenotype in 32D cells. The observed sustained receptor phosphorylation could be due to defective Cbl phosphorylation (Figure 3F) and ubiquitination in the Y589F mutant or deficient protein tyrosine phosphatase recruitment.

Mutation of Y599 to phenylalanine impairs FL-triggered Erk activation, which may imply a role of SFKs in positive Flt3 signaling as suggested in Robinson et al.²⁴  However, inhibition of SFKs by PP1 in Flt3-32D cells did not have any negative impact on FL-induced Erk activation in our hands. Instead, we revealed binding of SHP2 to wild-type Flt3 and strongly reduced SHP2 binding and phosphorylation in Y599F-Flt3-32D cells. Silencing of SHP2 by siRNA impaired Erk activation in ligand-stimulated 32D-Flt3 cells to a similar extent as seen in Y599F-Flt3 cells. We therefore suggest that SHP2 and its binding to pY599 are major prerequisites for FL-triggered positive downstream signaling. SHP2 is a nontransmembrane protein tyrosine phosphatase containing 2 SH2 domains and constitutes a key positive component of RTK signaling. Interestingly, SHP2 was among the first molecules to be identified as a target of Flt3.¹⁹  Growth factor-mediated activation of SHP2 has in many but not all cases been associated with phosphorylation of SHP2 on tyrosine residues.⁴⁹  Association of the 2 SH2 domains of SHP2 with phosphorylated tyrosine residues is an important mechanism in its activation. Early studies using synthetic phosphopeptides demonstrated that doubly phosphorylated peptides were significantly more effective in activating SHP2 compared with singly phosphorylated peptides.⁵⁰  By requiring both SH2 domains to associate with phosphotyrosine residues for activation, a very high degree of selectivity is achieved.⁵¹  In many proteins that associate with and activate SHP2, 2 phosphotyrosine residues are present and required for full activation.^43,52  Therefore, it is likely that an additional phosphorylated tyrosine residue(s) participates in recruitment of SHP2 to Flt3, which could explain the additional reduction in growth and Erk phosphorylation by SHP2 knockdown seen in the Y599F-Flt3 mutant (Figure 5B-C). Furthermore, it has been shown that SHP2 can indirectly interact with activated Flt3 via Grb2 or Gab2.²²  Dominant-negative mutants of SHP2 inhibit RTK-stimulated activation of the Erk pathway, immediate-early gene activation, cell spreading, and/or cell proliferation in mammalian tissue-culture cells,²⁸  whereas activated mutants of human SHP2 were found to lead to leukemic disorders.^29-31,53  Various modes of Erk activation by SHP2 have been proposed, ranging from SHP2 being an adapter protein for Grb2⁵⁴  to SHP2 dephosphorylating and thus inactivating negative regulators of Erk signaling such as Sprouty.⁵⁵  Moreover, SHP2 has been reported to dephosphorylate Cbp/Pag or paxillin and thereby abolish the binding site for the C-terminal Src kinase (Csk), a negative regulator of SFKs.^56,57  Thus, SHP2 is also capable of positively affecting Src kinase activity. For the EGF and PDGF receptors, it has been reported that SHP2 dephosphorylates the receptor binding site for Ras-GAP and thus preserves the activated status of Ras.^58,59  The following are interesting questions that require further investigation. Which of the mechanisms of SHP2 action applies primarily to Flt3 signaling? Do SFKs, after all, play a role in positive Flt3 signaling as downstream effectors of SHP2? What is the role of SHP2 and Y599 in FL-mediated Akt activation,⁶⁰  and what is the impact of SHP2 on ITD signaling?

Overall, our work presents novel insights into Flt3-mediated signal transduction. We identified the in vivo autophosphorylation sites of the juxtamembrane region of Flt3 and revealed SFKs and SHP2 as binding partners of pY589 and/or pY599, respectively, as well as their impact on FL-mediated signaling in Flt3-32D cells. Future work will now focus on elucidation of additional and possibly novel interaction partners of the found phosphorylation sites by employing an unbiased proteomics approach. The potential role of SFKs in regulating ubiquitination and internalization/degradation of FLt3 will be investigated in future studies as well as potential functional selectivities within members of the SFK family. With this gained knowledge it will be of interest to see whether ITDs differing in the nature of the duplicated tyrosines also confer distinct signaling behavior. If so, these tyrosines might serve as a diagnostic marker and point toward a successful combinatorial therapy consisting of an RTK inhibitor and an inhibitor for the specifically affected signal transduction pathway.⁶¹ 

We thank Drs G. Gilliland and T. Pawson for providing Flt3-cDNA and GST fusion proteins, respectively, and I. Dahlquist and P. Hansvik for expert assistance with the Edman-based radiosequencing and peptide syntheses.