Mechanisms of STAT Protein Activation by Oncogenic KIT Mutants in Neoplastic Mast Cells*

Mutations in the c-kit gene occur in the vast majority of mastocytosis. In adult patients as well as in the cell line derived from mast cell neoplasms, the mutations occur almost exclusively at amino acid 816 within the kinase domain of KIT. Among the downstream effectors of KIT signaling, STAT3 and STAT5 have been shown to be critical for cell proliferation elicited by the KIT-Asp816 mutant protein. However, little is known about the mechanisms of activation of STAT proteins. In this study, we identify and clarify the contribution of various STAT kinases in two widely used neoplastic mast cell lines, P815 and HMC-1. We show that STAT1, -3, and -5 proteins are activated downstream of the KIT-Asp816 mutant. All three STAT proteins are located in the nucleus and are phosphorylated on serine residues. KIT-Asp816 mutant can directly phosphorylate STATs on the activation-specific tyrosine residues in vitro. However, within cells, SRC family kinases and JAKs diversely contribute to tyrosine phosphorylation of STAT proteins downstream of the KIT mutant. Using a panel of inhibitors, we provide evidence for the implication or exclusion of serine/threonine kinases as responsible for serine phosphorylation of STAT1, -3, and -5 in the two cell lines. Finally, we show that only STAT5 is transcriptionally active in these cells. This suggests that the contribution of STAT1 and STAT3 downstream of KIT mutant is independent of their transcription factor function.

contribute to human neoplasms related to these cells such as mastocytosis, acute myeloid leukemia, germ cell tumors, gastrointestinal stromal tumors, and melanoma (1). The most common dysfunctions of the SCF/KIT pathway are the mutations in the c-kit gene that affect residues or domains involved in the inactive to active conformation transition, thus leading to the constitutive activation of the receptor (2). Among them, mutations in the juxtamembrane domain are found in the majority of gastrointestinal stromal tumors (3) and in some melanomas (4), whereas substitution of aspartic acid 816 (Asp 814 in the mouse) is found in about 80% of mastocytosis patients (5,6) and in some cases of acute myeloid leukemias (7,8), melanomas, and testicular germ cell tumors (9 -11).
KIT signals through the recruitment of proteins on its intracellular docking sites thereby inducing the formation of a large multiprotein signaling complex. Known signaling pathways activated by wild-type (WT) KIT include PI3K/AKT, RAS-ERK, the SRC family kinases (SFK), phospholipase C␥, and JAK/STAT pathways (1,12). Signal transducers and activators of transcription (STAT) proteins are latent cytoplasmic transcription factors that transduce the effects of a broad range of hormones, cytokines, and growth factors on target gene expression. There are seven mammalian STAT proteins (STAT1-4, -5A, -5B, and -6) that all share the same arrangement of functional motifs. STAT proteins become activated upon tyrosine phosphorylation and are subsequently translocated into the nucleus where they act on target gene promoters (13).
Constitutive activation of STAT proteins has been demonstrated in various leukemias (14,15). In normal signaling, STAT activation is rapid and transient. In contrast, aberrant permanent STAT activation has been associated with malignant progression in both solid tumors and blood malignancies (14,15). This activation is associated with the persistent activity of oncogenic protein-tyrosine kinase and has been shown to be directly linked to cellular transformation. For example, STAT3 activation directly participates in the transformation by the SRC oncoprotein (16), and STAT5 activation is essential for the transformation by BCR-Abl oncogenic fusion protein (17). STAT3 and STAT5 have been shown to be essential for cell proliferation in the context of KIT-Asp 816 gain-of-function mutant, in cell lines, or in activated mastocytes from patients with systemic mastocytosis (18 -20). However, the pathways linking the KIT receptor to STAT phosphorylation and acti-vation remain unknown. This study was undertaken to delineate the mechanisms of activation of STAT proteins downstream of KIT gain-of-function mutants in the commonly used mast cell leukemia cell lines P815 and HMC-1.

EXPERIMENTAL PROCEDURES
Cells-P815 and FMA3 are mouse mastocytoma cell lines carrying endogenous activating KIT-D814Y mutation and juxtamembrane deletion, respectively. The human mast cell leukemia cell line HMC-1 carrying the two point mutations V560G and D816V (HMC1.2) was kindly provided by Dr. J. H. Butterfield (21). TF1 KIT-D816V and MO7e KIT-D816V cells were derived by retroviral infections of parental TF-1 and MO7e cells, two human cytokine-dependent hematopoietic cell lines. Cell populations were selected by cell sorting and growth in the absence of GM-CSF. TF1 KIT-D816V and MO7e KIT-D816V cell lines were verified by sequencing of c-kit cDNAs. They are both KIT-dependent cell lines as controlled by siRNA and by using inhibitors of KIT. All cells were grown in RPMI 1640 medium supplemented with 10% heatinactivated fetal bovine serum (FBS). All reagents were from Invitrogen. Primary cultures of bone marrow-derived mast cells (BMMC) from wild-type mice were prepared and maintained in Opti-MEM medium with 10% FBS, 2 mM glutamine, 50 M ␤-mercaptoethanol, and 1% conditioned medium from baby hamster kidney cell cultures that express murine IL3. SCF stimulations were done with 250 ng/ml murine SCF (PeproTech). BMMC derived from transgenic mice expressing KIT-D816V were grown in BMMC media without exogenous IL-3 (22). COS-7 cells were grown in DMEM with 10% FBS, 1 mM sodium pyruvate.
For peptide pulldown experiments, 15-mer peptides corresponding to amino acids 563-577 of human KIT INGNNYVYIDPTQLP (Eurogentec) including phosphorylated or unphosphorylated Tyr 568 /Tyr 570 were coupled to Nhydroxysuccinimide-activated Sepharose beads (Amersham Biosciences) and incubated with cell lysates for 2 h at 4°C.
Lysates, immunoprecipitates, or affinity complexes were resolved by SDS-PAGE and transferred onto PVDF membrane (Immobilon-P; Millipore) followed by Western blotting with specific antibodies. Signals were revealed using West Pico chemiluminescent substrate (Pierce).
Vectors, siRNA, and Transfections-The vectors used for COS-7 transfection are pcDNA3-KIT-WT and pcDNA3-KIT-D816V. The following vectors were used for the production of STAT fusion proteins used in in vitro kinase assays: pGEX-STAT1 (a gift from Drs. Ali and Sayeski (23)); pGEX-STAT3 (a gift from Drs. Cao and Cheh Peng (24)); and pGEX-STAT5A (amino acids 494 -793) (a gift from Dr. Ikuta (25)). pGEX-STAT5A-Y694F was obtained by sitedirected mutagenesis using the QuikChange kit (Stratagene) according to the manufacturer's instructions. The following vectors were used for STAT luciferase reporter assay: pGAS-Luc (Stratagene); p4ϫTSV-Luc containing the core region of the LIF-response element of the rat a2M gene (a gift from F. Gouilleux and G. M. Hocke (26)); and p␤CAS-Luc (a gift from W. Leonard (27)) to monitor STAT1, -3, and -5 activities, respectively. pRL-TK expressing Renilla luciferase was used as control in the transfection experiments (Promega). Luciferase activities in the figures are expressed as arbitrary units relative to control Renilla luciferase. COS-7 cells were transfected with FuGENE 6 transfection reagent (Roche Diagnostics) according to the manufacturer's protocol. HMC-1 and P815 cells were transfected by electroporation with an Xcell pulser (Bio-Rad). 8 ϫ 10 6 cells in 500 l of RPMI 1640 medium were chocked in 4-mm electroporation cuvettes at 250 V and 950 microfarads for HMC-1 cells and 280 V and 950 microfarads for P815 cells. jak3, src, and lyn siRNAs were from Dharmacon (ON-TARGET plus SMART-pool). c-kit siRNA was described previously (32). For siRNA knockdown, 0.4 nmol of siRNA was used for each electroporation.
In Vitro Kinase Assay-Purified KIT-D816V and SRC kinases are a gift from L. Gros (AB Science). They were obtained from a baculovirus expression system and purified as described previously (28). Vectors expressing GST-STAT fusion proteins were transformed in the Rosetta TM Escherichia coli strain (Merck). Fusion proteins were purified and eluted by GST-glutathione affinity chromatography (glutathione-Sepharose 4 Fast Flow, GE Healthcare).
In vitro kinase assays were performed with the indicated quantity of enzymes and substrates in the presence of an excess of ATP (250 M). Reactions were carried out in kinase buffer (50 mM Hepes (pH 7.8), 10 mM MgCl 2 , 2 mM MnCl2, Brij 0,01%) at room temperature and stopped after 1 h by adding 50 mM EDTA. GST-STAT substrates were then collected in wells coated with anti-GST antibodies (GST-Tag Antibody Plate, Novagen), and binding was allowed for 1 h followed by three washes in PBS, 0.05% Tween. Substrate phosphorylation was then revealed by detecting the binding of the primary anti-phosphotyrosine-Stat1/3/5 antibodies (Cell Signaling Technology, used at a 1:200 dilution) with a secondary antirabbit HRP-conjugated antibody (used at a 1:10,000 dilution) and incubation with tetramethylbenzidine (substrate reagent, R&D Systems, Minneapolis, MN). The final reaction product was quantified by spectrophotometry at 370 nm.
STAT Luciferase Reporter Assay-COS-7 cells were transfected with a plasmid containing WT KIT or KIT-D816V, a STAT Firefly luciferase reporter plasmid, and a control Renilla luciferase plasmid in a 1:1:0.2 ratio using FuGENE 6 according to the protocol described above. The total amount of plasmid DNA used was 3.3 g to transfect 3 ϫ 10 5 cells. Each transfection was done in triplicate. Luciferase activity was determined 24 h after transfection.
HMC-1 and P815 were transfected by electroporation as described above. The STAT reporter plasmid and control plasmid ratio used was 10:1, and the total amount of transfected DNA was 5.5 g. Immediately after electroporation, each cuvette was split in two, and cells were grown for 6 h with 1 M dasatinib or with an equivalent volume of DMSO as a control.
The luciferase activity was determined using a Dual-Luciferase TM reporter assay kit (Promega). Each measure was done in triplicate. In all figures, specific STAT activity was represented as the ratio of Firefly luciferase over Renilla luciferase activity for 10 g of cell lysates proteins.

Regulated Transient Tyrosine Phosphorylation of STATs by Wild-type KIT Is Constitutively Downstream of KIT Gain-of-Function Mutants-
We have previously reported a robust tyrosine phosphorylation of STAT1, -3, and -5 downstream of the juxtamembrane gain-of-function mutants of KIT (29). With regard to the KIT kinase domain mutant, the most frequent mutant of KIT in mast cells (i.e. KIT mutated at position Asp 816 ), several groups have reported phosphorylation of either STAT1, -3, or -5 family members using various transfected cell models (18, 19, 29 -32). To determine whether the tyrosine phosphorylation of STAT1, -3, and -5A/B was a feature of neoplastic mast cells, we analyzed their phosphorylation status in commonly used mast cell lines. HMC-1 is a human mast cell leukemia cell line carrying two activating point mutations at codons 560 and 816 in the c-kit gene. P815 and FMA3 are murine mastocytoma cell lines carrying the murine D814Y mutation homologous to human D816V and an inframe deletion in the juxtamembrane inhibitory domain of c-kit, respectively. Constitutive tyrosine phosphorylation of STAT1, -3, and -5 was observed in all three cell lines (Fig. 1A).
To establish whether KIT is upstream of STAT phosphorylation in the neoplastic mast cell lines P815 and HMC-1, we first used an inhibitor of KIT kinase activity. KIT-Asp 816 mutant is resistant to imatinib but is dasatinib-sensitive (33,34). Dasatinib treatment resulted in a complete inhibition of the phosphorylation of all three STATs (Fig. 1B), suggesting that KIT signaling is responsible for STAT proteins activation in P815 and HMC-1. This result was then confirmed by reducing KIT expression using RNA interference in HMC-1 cells (Fig. 1C).
To further demonstrate that KIT kinase mutant induced the phosphorylation of STAT proteins, KIT-D816V was transfected in two human hematopoietic cell lines MO7e and TF1 in which STAT1, -3, and -5 tyrosine phosphorylation was minimal in the absence of cytokine stimulation. As shown in Fig. 1D, phosphorylation of STAT proteins is minimal in the parental cells, whereas KIT-transfected cells show robust phosphorylation of all three STAT proteins. STAT phosphorylation was strictly dependent on KIT in these cells as demonstrated by selective reduction of KIT expression by RNA interference (Fig. 1E). In conclusion, KIT kinase mutants found in neoplastic mast cells all lead to the downstream constitutive tyrosine phosphorylation of STAT1, -3, and -5.
Finally, we asked whether phosphorylation of STAT proteins was a pathway activated by wild-type KIT signaling in primary mast cells. To evaluate the activation of STAT1, -3, and -5 upon SCF stimulation, murine primary BMMC were stimulated with SCF, and lysates were analyzed by Western blotting with phospho-specific antibodies. As seen in supplemental Fig. S1, STAT1, -3, and -5 were phosphorylated on their activation-specific Tyr 701 , Tyr 704 , and Tyr 694 , respectively, following KIT stimulation. The kinetics of phosphorylation was rapid and transient, peaking at 5 min and downregulated within ϳ15 min. We also derived BMMC obtained from KIT-Asp 816 transgenic mice (22). In these primary cells, the persistent activation of STAT proteins was also observed, although AKT and ERK pathways were down-regulated as in the context of wild-type KIT stimulation, showing the distinctive feature of STAT proteins activation (Fig. 1D, last lane).
Thus, under KIT gain-of-function mutants, the STAT pathway was not down-modulated as upon WT KIT stimulation, resulting in permanent tyrosine phosphorylation of STAT1, -3, and -5.
Contribution of SFKs and JAKs to STAT Tyrosine Phosphorylation-To identify the STAT kinases downstream of KIT in neoplastic mast cells, we used a candidate approach. The kinases responsible for the tyrosine phosphorylation of STATs include the cytosolic kinases of the Janus (JAK) or SRC family (SFK). Phosphorylation STAT proteins can also occur through direct phosphorylation by an upstream receptor tyrosine kinase as described for FGFR3 (35), PDGFR (36,37), or FLT3 (38).
First, we asked whether JAKs were involved in STATs activation. Members of JAK family kinases have been previously shown to be phosphorylated in the HMC-1 cell line ( Fig. 2A) (39). In our experiments, JAK3 is the main JAK activated in P815 and HMC-1, and its phosphorylation is dependent on KIT catalytic activity (data not shown). The treatment of P815 cells with JAK inhibitor I, a potent inhibitor of JAK1, -2, and -3 and Tyk2, resulted in inhibition of JAK3 autophosphorylation ( Fig. 2A, upper panels) but had no influence on STAT1, -3, and -5 tyrosine phosphorylation ( Fig. 2A, lower panels). The same result was obtained following JAK3 expression knockdown using RNA interference in P815 cells (supplemental Fig. 2). These results suggest that STATs phosphorylation is independent of JAKs activity in P815. In HMC-1 cells, a reduction of phosphorylation of STAT1 and STAT5 was observed, showing a contribution of JAKs to activation of STAT proteins (Fig. 2B). Unlike in P815, JAK inhibition in HMC-1 resulted in decreased SFK Tyr 416 phosphorylation.
Second, to determine the implication of SFK, we conducted the same experiments using the selective inhibitor SU6656. As shown on Fig. 2C, the treatment of both P815 and HMC-1 cells resulted in the total disappearance of phospho-STAT1 and a partial reduction of phospho-STAT3. Phospho-STAT5 was not affected by SFK inhibition in both cells. These results suggest an implication of SFK in STAT1 and STAT3 phosphorylation but not in the activation of STAT5.
To further demonstrate the implication of SFK, we silenced two members of this family, SRC and LYN, which are activated in the HMC-1 cell line. Reduction of LYN expression using specific siRNA did not diminish the phosphorylation of  FEBRUARY 25, 2011 • VOLUME 286 • NUMBER 8 STAT proteins (Fig. 1D, right panels). On the other hand, reduction of SRC expression abolished STAT1 tyrosine phosphorylation (Fig. 1D, left panels). This result strongly supports the implication of SRC as the STAT kinase for STAT1 in HMC-1.

Activation of STATs by KIT Mutant
KIT Phosphorylates STAT1, -3, and -5 in Vitro-We next sought to determine whether KIT-D816V could directly phosphorylate STAT proteins as demonstrated for some other receptor tyrosine kinases. We used an in vitro kinase assay coupled with an ELISA detection system to assess STAT1, -3, and -5 phosphorylation. In vitro kinase assays were performed using purified KIT-D816V and GST-STAT fusion proteins as substrates. Kinase assays were performed in excess of ATP (250 M) and substrates (400 nM) and increasing concentrations of enzyme. Substrate phosphorylation was revealed using specific STAT anti-phosphotyrosine antibodies. As seen in Fig. 3A (upper panel), the tyrosine phosphorylation of all three STATs increased with KIT-D816V protein concentration. As a control, the tyrosine mutant GST-STAT5-Y694F did not show phosphorylation in this assay or in control Western blots (Fig. 3B). Thus, KIT-D816V phosphorylated all three STATs in vitro at the conserved tyrosine residue.
Intrinsic Kinase Properties of KIT-D816V and SRC May Account for STAT Substrate Selectivity-Having shown that KIT-D816V and SFK differentially contribute to tyrosine phosphorylation of STATs in cells, we asked whether this could be explained by the intrinsic catalytic properties of the kinases toward STAT substrates. To address this question, we performed kinase assays to determine SRC and KIT-D816V kinetic parameters toward the different STAT substrates.
Enzyme titration assays were performed in excess of substrates to settle the appropriate amount of recombinant en-zyme to use in the assays. As seen on Fig. 3A (lower panel), GST-STAT phosphorylation displayed a linear increase followed by saturation as the concentration of SRC was raised. For KIT-D816V, saturation was reached for STAT3 but remained linear for STAT1 and STAT5. For each enzyme, the quantity of KIT and SRC enzymes was chosen in the linear part of the curves, i.e. 1 g for both enzymes. The phosphorylation states of the different GST-STAT proteins by SRC and KIT-D816V were then compared using the set enzyme quantity, in 1 h with increasing concentration of STAT substrates. As seen on Fig. 3C, phosphorylation of each STAT fusion proteins increased with substrate concentration, and the reactions followed Michaelis-Menten kinetics. For STAT3 phosphorylation, KIT and SRC kinases showed overlapping kinetics indicating similar enzyme kinetics parameters (V max and K m ). Regarding STAT1 phosphorylation, SRC kinase showed higher V max and lower K m values than KIT, which indicated that SRC is more efficient than KIT for STAT1 tyrosine phosphorylation. Finally, STAT5 phosphorylation kinetics show very different enzymatic parameters with lower K m values for KIT but higher V max values for SRC; therefore, the comparison of SRC and KIT efficiencies to phosphorylate STAT5 in vitro was not conclusive. In conclusion, the enzymatic parameters may account for the contribution of SRC in STAT1 phosphorylation and for the cooperative effect of SRC and KIT on STAT3 phosphorylation observed in cells.
Endogenous Interactions of KIT-D816V with STAT1, -3, and -5-Having shown that KIT-Asp 816 mutant can phosphorylate STAT proteins in vitro, we asked whether STAT1, -3, and -5 were part of the KIT-D816V receptor signaling complex. KIT-D816V was immunoprecipitated from HMC-1 cells, and co-precipitated proteins were probed for STATs by Western blotting using specific antibodies. As seen on Fig. 4A  (left panel), KIT-D816V co-immunoprecipitated with STAT1, -3, and -5 (also seen on supplemental Fig. 4, A and B). Furthermore, these interactions were dependent on KIT-D816V catalytic activity as they were lost or severely diminished when the kinase activity of the receptor was abolished with dasatinib treatment. Interestingly, SFKs were also part of this complex (Fig. 4A, right panel).
The di-tyrosine motif Tyr 589 -Tyr 591 within the juxtamembrane domain of FLT3-ITD has been involved in STAT5 acti-vation (40). Furthermore, the homologous motif in KIT is thought to recruit SFK (41). We assumed that this docking site on KIT, consisting of tyrosines 568 and 570, could also be implicated in STAT binding. We performed affinity pulldown assays with synthetic peptides containing the di-tyrosine motif either in a nonphosphorylated form or in one of the three possible phosphorylated states, i.e. phosphotyrosine 568 alone (Tyr(P) 568 ), phosphotyrosine 570 alone (Tyr(P) 570 ), or both 568 -570 tyrosines phosphorylated (Tyr(P) 568 -Tyr(P) 570 ). As seen Fig. 4B, STAT1, -3, and -5 bound to the phosphorylated peptides. The interaction was phosphorylation dependent as shown by the nonphosphorylated peptide (Tyr 568 , Tyr 560 ). As a control, STAT proteins did not bind a control tyrosine phosphorylated peptide derived from the sequence of CD28 (supplemental Fig. 5). These results suggest that STAT proteins are part of the KIT receptor signaling complex and that the recruitment of STAT proteins occur in the juxtamembrane di-tyrosine motif Tyr 568 , Tyr 570 .
STAT5 Is Transcriptionally Active in the Neoplastic Mast Cells, whereas STAT1 and STAT3 Are Not-STAT tyrosine phosphorylation is required for their translocation to the nucleus and thus for their activation as transcription factors. We used a luciferase-based reporter assay to test STAT transcriptional activity downstream of KIT-D816V. Cells were transfected with reporter plasmids expressing Firefly luciferase under the control of promoters containing STAT1-, -3-, or -5-responsive elements, together with a control plasmid encoding Renilla luciferase driven by the ubiquitous thymidine kinase promoter. To validate the assay, the c-kit mutant cDNA and the reporter constructs were first transfected in COS-7 cells. Transient transfection of KIT-D816V in COS cells leads to robust STAT protein phosphorylation (data not shown). As seen in Fig. 5A, KIT-D816V ectopic expression in COS-7 cells led to the activation of all three STAT reporter constructs. This experiment indicated that KIT-D816V can activate STAT1, -3, and -5 transcriptional activity.
We then addressed whether STAT proteins were transcriptionally active in the mastocyte cell lines. The same experiment as above was performed in P815 (Fig. 5B) and HMC-1 (Fig. 5C) cells. As shown in Fig. 5, B and C, STAT5 reporter was active in both cell types, and this activity was dependent on KIT-D816V kinase activity as revealed by dasatinib treatment. By contrast, we could not detect any transcriptional activity of either STAT1 or STAT3 reporters in those cells. In conclusion, KIT-D816V can activate STAT1, -3, and -5, but only STAT5 is transcriptionally active in P815 and HMC-1 mast cells.  ). B, peptide pulldown assays. HMC-1 SCL was used for affinity pulldown assay with 2 nmol of either nonphosphorylated peptide (Y568Y570) or mono-phosphorylated peptides (pY568 and pY570) or bi-phosphorylated peptide (pY568pY570). The binding of STAT1, -3, and -5 was revealed with specific antibodies. Another SH2containing protein, VAV, was analyzed as a control. SCL was used as positive control. FIGURE 5. KIT-D816V triggers STAT5 transcriptional activity. A, STATs reporter assays in COS cells. COS-7 cells were transfected with three different STAT Firefly luciferase (F-luc) reporter vectors and a control Renilla luciferase (R-luc) vector. STAT1 activation was monitored using pGAS-luc, a commercial vector containing a STAT1 homodimer binding GAS-like sequence. STAT3 activation was monitored with p4ϫTSVluc containing the core region of the LIF-response element of the rat a2M gene, and p␤Casein-luc plasmids were used to monitor STAT5 activity. The activity of the reporter plasmids was compared between COS cells transfected either with WT KIT or KIT-D816V expression vectors. The luciferase activity was measured 48 h after transfection as described under "Experimental Procedures." COS cells expressing unstimulated WT KIT were used as control, and their activity was normalized to 1. Data represent an average of five independent experiments (each done in triplicate) with indicated standard deviation. B and C, STAT reporter assays in mastocytoma cell lines. P815 cells (B) and HMC-1 cells (C) were electroporated with each STAT activity reporter plasmid and Renilla plasmid control. Ϫ histograms represent control cells transfected with the Renilla plasmid only. Cells were grown for 6 h in the presence or absence of dasatinib before lysis. Left panel shows a representative experiment out of three independent experiments done in triplicate. The data shown in the right panel represent an average of three independent experiments (each done in triplicate) with indicated standard deviation for STAT5 activity. Statistically significant differences relative to untreated cells are indicated (Mann-Whitney, ***, p Ͻ 0.0001).

Serine Phosphorylation and Nuclear Localization of STATs in Neoplastic Mast
Cell Lines-Additional serine phosphorylation in the C-terminal transactivation domain of STAT proteins is necessary for full transcriptional activity (42). Because STAT1 and STAT3 were not active, we asked whether they were phosphorylated on the specific serine residues. As shown in Fig. 6A, STAT1, -3, and -5 are phosphorylated on their conserved serines 727, 727, and 726/731, respectively, both in P815 and HMC-1 cells. Furthermore, the inhibition of KIT-D816V activity led to the disappearance of phosphoserine STAT3 and STAT5 in both cells showing that these phosphorylations are KIT-dependent. By contrast, STAT1 serine phosphorylation is basal and independent of KIT catalytic activity.
To further delineate the pathways responsible for STATs serine phosphorylation, we applied a systematic approach using chemical inhibitors. The MAPK family kinases were targeted either with U0126 to inhibit MEK1/2 and as a consequence the ERK pathway or with SB203580 to inhibit the p38 pathway. We also used KN-93 to inhibit CaMKII, bisindolylmaleimide, H-89, and protein kinase G inhibitor to target PKC, PKA, and PKG, respectively, and SB216763 to inhibit GSK3. All these inhibitors were used at the minimal dose required to inhibit the respective kinases (43). As shown on Fig.  6B, treatment of both P815 and HMC-1 with U0126 resulted in the inhibition of STAT3 serine phosphorylation, with no effect on STAT1 and STAT5. A reduction of STAT5 serine phosphorylation was observed with treatment of bisindolyl-FIGURE 6. KIT-D816V-mediated STATs serine phosphorylation and nuclear localization. A, KIT-D816V-mediated serine phosphorylation of STAT3 and STAT5. P815 and HMC-1 cells were starved overnight and subsequently treated or not with dasatinib. SCLs were subjected to immunoprecipitation (IP) with anti-STAT antibodies. Immunocomplexes were resolved by SDS-PAGE and probed with anti-phosphoserine-specific antibody, stripped, and reprobed with anti-STAT as indicated. B, ERK pathway is implicated in STAT3 serine phosphorylation. STAT serine phosphorylation and KIT tyrosine phosphorylation were analyzed as described in A after treatment of cells with MEK inhibitor U0126. C, PKC pathway is implicated in STAT5 serine phosphorylation. STATs serine phosphorylation was analyzed as described in A and B after treatment of cells with PKC inhibitor bisindolylmaleimide. Similar results were obtained in three independent experiments for each Western blot presented. D, cytoplasmic and nuclear localization of STATs. HMC-1 cells were starved overnight and subsequently treated or not with dasatinib. Cytoplasmic fraction (CYTO) was obtained after lysis in hypotonic buffer. Nuclear fraction (NUC) was obtained from hypertonic shock of pelleted nuclei. Lysates were loaded as cell number equivalents (i.e. cytoplasmic and nuclear fractions obtained from 5 ϫ 10 5 cells). STAT1, -3, and -5 location was analyzed by Western blotting both with phosphotyrosine-specific antibodies and antibodies directed against total proteins. FEBRUARY 25, 2011 • VOLUME 286 • NUMBER 8 maleimide (Fig. 6C). Thus, ERK and PKC pathways are likely to be involved in STAT3 and STAT5 serine phosphorylation, respectively, downstream of KIT-D816V. The other inhibitors had no effect on STAT serine phosphorylation (supplemental Fig. 3).

Activation of STATs by KIT Mutant
As all three STATs were phosphorylated on tyrosine and serine residues, but only STAT5 was transcriptionally active, we finally checked their nuclear localization in HMC-1 cells. We performed cellular fractionation, and as shown in Fig. 6D, all STATs were found both in the cytoplasm and nucleus. Furthermore, we observed an enrichment of the tyrosinephosphorylated forms in the nuclear fraction compared with the cytoplasmic one, with phosphotyrosine STAT5 only detectable in the nuclear fraction. Thus, neither serine phosphorylation nor nuclear localization could account for the observed absence of transcriptional activity of STAT1 and STAT3.

DISCUSSION
We show here that KIT is responsible for the permanent phosphorylation of all three STAT proteins STAT1, -3, and -5 in P815 and HMC-1 cell lines with diverse contributions of SFK and JAKs depending on the STAT protein and on the cell line. We also report the involvement of ERK MAPKs and PKC for serine phosphorylation of STAT3 and STAT5, respectively, downstream of KIT in the two cell lines. Importantly, only STAT5 showed transcriptional activity both in P815 and HMC-1 cells. Overall, similar results and conclusions were reached for both murine P815 and human HMC-1 cells. However, a crucial difference was pointed out regarding the implication of JAKs restricted to HMC-1 cells.
We had previously reported the constitutive tyrosine phosphorylation of STAT1, -3, and -5 by KIT juxtamembrane mutants, i.e. regulatory type mutants mostly found in gastrointestinal stromal tumors (29). In neoplastic mast cells, the most common KIT alteration is the kinase domain substitution Asp 816 by a valine residue. This mutant is also found in germ cell tumors (9 -11) and in acute myeloid leukemias (7,8). Previous studies have reported the activation of STAT1 or STAT3 by KIT-D816V in transfected cell lines (18,30,44); however, very few have addressed the activation of STAT in neoplastic cells. A noticeable exception is a recent description of STAT5 activation and STAT5 requirement for cell proliferation in HMC-1 cells (19). Our study is the most exhaustive description of STAT1, -3, and -5 protein phosphorylation and activation in the commonly used neoplastic mast cells. Overall, it seems that KIT juxtamembrane and kinase domain mutants activate all three STAT proteins. The wild-type KIT receptor can transiently activate STAT proteins. Hence, these pathways are not unique to mutants. However, they are down-regulated in the wild-type stimulation, although they are permanently activated in the mutant receptor context. Therefore, we believe that STAT activation downstream of mutant KIT can be called aberrant as it is in other pathological situations. Whether the permanent activation is due to an alternative activation route or to the loss of a negative feedback loop remains to be determined.
Aberrant STAT protein activation has been found in numerous leukemias and other cancers, and STAT activation is thought to contribute to neoplasia (14). As a consequence, STAT proteins are attractive therapeutic targets. Several strategies have been proposed to target various aspects of STAT protein activation or STAT function. These include targeting STAT recruitment to receptors or oncogenic proteins, inhibition of STAT kinases, activation of STAT inhibitors, and interference with STAT protein dimerization, nuclear transport, or DNA binding (45,46). Therefore, it is extremely important to understand the pathways of STAT activation and the STAT function implicated in each disease to choose the appropriate STAT targeting strategy. Although some of the questions raised in this study have been partially addressed before, the work presented here is the most extensive undertaken so far on the mechanism of STAT protein activation in neoplastic mast cell lines.
In the conventional JAK-STAT signaling pathway, the triggering of the cytokine receptor causes STAT tyrosine phosphorylation by the receptor-associated JAK tyrosine kinase (47). However, receptors with intrinsic tyrosine kinase activity can also directly phosphorylate STATs, as described for STAT1 phosphorylation by PDGFR (36) or STAT5 phosphorylation by PDGFR (37) and insulin receptor (48). This activation can also be indirect via the recruitment of cytosolic protein-tyrosine kinases. Among the latter, members of the SFK are well described STAT tyrosine phosphorylation actors in a physiological (49) or pathological context (50). Our data combining observations done in cells together with in vitro kinase assays suggest differential roles for JAK, SFK, and KIT for STAT tyrosine phosphorylation. In murine mastocytoma P815, SFK are the main STAT1 tyrosine kinase, although SFK and KIT cooperate to phosphorylate STAT3, and KIT is the main candidate as STAT5 kinase. In human mast cell leukemia cell line HMC-1, we reached the same conclusion for the contribution of SFK, but in these cells the activations of SFK are in part dependent on JAKs. As mentioned above, the contribution of JAKs upstream of SFK is the main difference we found between P815 and HMC-1.
Our kinase assays suggest that KIT is a STAT1, -3, and -5 kinase in vitro. This was previously shown by Deberry et al. (51) for STAT1. We have shown for the first time a physical interaction of KIT with all three STAT endogenous proteins in HMC-1 cells, which suggest that STAT proteins could also be KIT substrates within cells as well. However, the suspected site of interaction, a di-tyrosine motif in KIT juxtamembrane region, is a docking site involved in the recruitment of many signaling proteins, including docking molecules such as SHP2 (52) as well as SFK (41). Therefore, this site is involved in the formation of a multiprotein signaling receptor complex that includes other potential STAT kinases. Interestingly, the homologous site in the closely related receptors FLT3 and PDGFR was implicated in STAT5 activation by the oncogenic mutants FLT3-ITD and TEL-PDGFR␤, respectively (40,53). Unlike in the context of KIT mutant receptors, in the case of FLT3-ITD, only STAT5 is activated, but as for KIT the phosphorylation of STAT5 is thought to be independent of SFK and JAKs. Therefore, there are both similarities and differ-ences in the activation of STAT proteins by these related mutant receptors.
In our experiments, JAK3 but not JAK2 is activated downstream of KIT in the two mast cell lines (Fig. 2). It is of interest to note that JAK1 and JAK3 have been implicated in the phosphorylation of STAT proteins in the closely related HMC-1.1 cell line that was derived from the original HMC-1 (39). Although not obtained with the same cells, these results support a model implicating JAKs in the phosphorylation of STAT proteins in HMC-1.
There are nine members in the SRC family kinase, many of which are expressed in mast cells. The inhibitors used here do not discriminate among family members. LYN is one candidate as it is activated downstream of KIT in both cell lines. We knocked down LYN expression by RNA interference in HMC-1 and found no differences in STATs phosphorylation status, excluding LYN as the major STAT kinase. We found, however, that SRC is essential for STAT1 tyrosine phosphorylation. It remains to be determined which SFK member could act as a STAT3 kinase in the neoplastic mast cells.
In addition to tyrosine phosphorylation, serine phosphorylation is required for the full transcription activation of STAT proteins (42). The conserved serine is located in the transactivation domain of STAT proteins and embedded in a PMSP motif, a potential consensus site for MAPKs. In P815 and HMC-1 cells, all three STAT proteins were phosphorylated at the conserved serine residue. STAT1 serine phosphorylation was constitutive in these cells independently of KIT activation, whereas STAT3 and STAT5 serine phosphorylation was dependent on KIT kinase activity. In line with several other studies, which concluded that MAPKs are bona fide STAT3 serine kinases (54), our data suggest that the MEK-ERK1/2 pathway is involved in STAT3 serine phosphorylation both in P815 and HMC-1. For STAT5, our screen using chemical inhibitors pointed to kinases of the PKC family in both cell lines. PKCs have already been proposed as STAT kinases for STAT1 in type I interferon signaling and STAT3 in Insulin signaling context (55,56).
In the original paradigm of the STAT signaling pathway, STAT proteins were described as direct signaling effectors connecting membrane receptors to transcription. Indeed, the activation of STAT proteins results in the expression of genes that control critical cellular functions, including cell proliferation, survival, differentiation, and development (13). However, recent studies on STAT protein point to more complex functions for some STAT proteins. For instance, a function of STAT3 in the mitochondria was recently discovered, in which STAT3 promoted oxidative phosphorylation and increased transformation by oncogenic Ras independently of its transcription activity (57,58). Therefore, STAT3 could contribute to KIT-Asp 816 oncogenicity independently of its role as transcription factor.