TAK1 and IKK2, novel mediators of SCF-induced signaling and potential targets for c-Kit-driven diseases.

NF-κB activation depends on the IKK complex consisting of the catalytically active IKK1 and 2 subunits and the scaffold protein NEMO. Hitherto, IKK2 activation has always been associated with IκBα degradation, NF-κB activation, and cytokine production. In contrast, we found that in SCF-stimulated primary bone marrow-derived mast cells (BMMCs), IKK2 is alternatively activated. Mechanistically, activated TAK1 mediates the association between c-Kit and IKK2 and therefore facilitates the Lyn-dependent IKK2 activation which suffices to mediate mitogenic signaling but, surprisingly, does not result in NF-κB activation. Moreover, the c-Kit-mediated and Lyn-dependent IKK2 activation is targeted by MyD88-dependent pathways leading to enhanced IKK2 activation and therefore to potentiated effector functions. In neoplastic cells, expressing constitutively active c-Kit mutants, activated TAK1 and IKKs do also not induce NF-κB activation but mediate uncontrolled proliferation, resistance to apoptosis and enables IL-33 to mediate c-Kit-dependent signaling. Together, we identified the formation of the c-Kit-Lyn-TAK1 signalosome which mediates IKK2 activation. Unexpectedly, this IKK activation is uncoupled from the NF-κB-machinery but is critical to modulate functional cell responses in primary-, and mediates uncontrolled proliferation and survival of tumor-mast cells. Therefore, targeting TAK1 and IKKs might be a novel approach to treat c-Kit-driven diseases.


INTRODUCTION
The IKK complex consists of IKK1/2 and NEMO. IKK1 and 2 are the catalytically active components whereas NEMO is the regulatory scaffold protein of the IKK complex [1][2][3]. Stimulation of receptors of the TLR/IL-1 (TIR) -or TNFR superfamily results in a poly-ubiquitinylation-dependent phosphorylation of the serine (S) residue motive, S177/S181 of IKK2. This motive is located in the activation loop of IKK2 [4,5] and is pivotal for IKK2 activation [5]. TAK1 is proposed as the upstream IKK2 kinase which mediates S177/S181 phosphorylations and therefore activation of IKK2 [6]. The consequence of IKK2 activation is the phosphorylation of the NF-κB www.impactjournals.com/oncotarget inhibitor IκBα which is ubiqiutinylated and proteasomal degraded. Therefore, NF-κB is released, translocates into the nucleus and induces a cytokine production [7]. Recent publications suggests that receptor tyrosine kinases (RTKs) also activate NF-κB via the IKK complex [8,9].
The IKK complex and NF-κB are important to mediate the effector functions of mature mast cells stimulated with ligands of the TIR superfamily (e.g. IL-33R; [24]) or by crosslinking of the Fcε receptor I [FcεRI; [25]). Recent data suggest that IKKs are also crucial to induce NF-κB-dependent tumorigenesis [26].
We show that stimulation of BMMCs with SCF, the overexpression of wt c-Kit or the expression of constitutively active c-Kit mutants mediates activation of IKK2. Surprisingly, the c-Kit-mediated IKK2 activation does not lead to NF-κB activation but is critical for differentiation, proliferation and survival of primary-and tumor-mast cells. Therefore, the c-Kit-mediated IKK2 activation regulates mast cell effector functions and might be a novel approach to treat c-Kit-driven diseases.

SCF induces IKK2 activation
Aiming to investigate the SCF-induced signaling in BMMCs we unexpectedly found that SCF induces IKK2 activation which was blocked by the c-Kit inhibitor imatinib ( Figure 1A). To delineate the role of IKK2 in the SCF-induced signaling and effector functions, we used IKK2 deficient BMMCs from inducible ikk2 Δ mice [27] and from cre − , Ikk2 F/F control mice, as well as the IKKinhibitor VII [28,29]. We focused on the SCF-induced STAT and PI3K activation, since inactivation of these signaling cascades, but not MAPKs (Supplementary Figures S1A-S1C) compromises mitogenic mast cell responses [17,18,30,31].
The reduced expression of IKK2 in ikk2 Δ BMMCs ( Figure 1B) does not affect surface expression of c-Kit compared to control ikk2 F/F BMMCs (Supplementary Figure S1D). The SCF-induced PKB/Akt activation was unperturbed, whereas the activation of STAT3/5 was reduced in ikk2 Δ compared to ikk2 F/F BMMCs ( Figure 1B). To confirm these results we used the IKK-inhibitor VII [32]. We pre-incubated BMMCs with different IKKinhibitor VII concentrations and stimulated with SCF. We found that the IKK-inhibitor VII does not affect the SCF-induced activation of PKB/Akt but strongly reduced the activation of STAT3/5 and proliferation even at very low concentrations (0,5-0,1 μM) (Supplementary Figures S1E and S1F). Compared to the results obtained with ikk2 Δ BMMCs, these data show that IKK-inhibitor VII treatment or the reduced IKK2 expression affected the same SCF-induced signaling pathways. Therefore, it can be assumed that the IKK-inhibitor VII preferentially blocks IKK activation in mast cells. Given that STATs are critical for mast cell differentiation, we examined whether IKK2 contributes to proliferation and differentiation. As shown in Figures 1C-1G, we found a slightly reduced proliferation ( Figure 1C) and a profound reduction of mast cell numbers in the peritoneal fluid ( Figures 1D and 1E), the ear-( Figure  1F) and back-skin of ikk2 Δ mice ( Figure 1G) compared to control ikk2 F/F mice. These data show that the SCF-induced IKK2 activation mediates differentiation via STAT3/5.

IKKs mediate proliferation and cell survival of tumor mast cells
We next investigated whether IKKs also contribute to the signaling induced by constitutively active c-Kit mutants. Therefore, we used HMC-1.1 (expressing the imatinib-sensitive V560G c-Kit mutant) and HMC-1.2 cells (expressing the imatinib-insensitive D816V/V560G c-Kit mutant) [33]. Indeed, incubation of HMC-1.1 cells (Figures 2A-2C Next, we investigated the transforming potential of the D816V c-Kit mutant and its dependency on IKKs. Confirming others [34], expression of the D816V Kit mutant, but not of wt c-Kit potentiated the basal proliferation of Ba/F3 cells ( Figure 3A). Pre-treatment with the IKKinhibitor VII reduced the basal proliferation of D816V c-Kit-expressing Ba/F3 cells for about 200 fold whereas the proliferation of parental Ba/F3 or wt c-Kit-expressing Ba/F3 cells was reduced up to 50 fold ( Figure 3B). This shows that D816V c-Kit-expressing Ba/F3 cells are up to four times more sensitive to the IKK-inhibitor VII than parental Ba/F3 or wt c-Kit-expressing Ba/F3 cells.
We further examined the induced signaling pathways in these Ba/F3 cell clones. In parental Ba/F3 we could not detect the activation of STATs, PKB/Akt and MAP-kinases ( Figures 3C-3E). In wt c-Kit-expressing Ba/F3 cells, the activation of Erk1/2 but not of PKB/Akt and STATs was slightly increased (Figures 3C-3E). In contrast, in D816V c-Kit-expressing Ba/F3 cells we found a strong activation of Erk1/2, PKB/Akt and STAT3 which were blocked by the IKK-inhibitor VII ( Figures 3C-3E). These data indicate that constitutively active c-Kit mutants mediate IKK-dependent mitogenic signaling. www.impactjournals.com/oncotarget Src-family kinases (SFKs) mediate IKK2 and STAT3/5 activation Now, we focused on the mechanism leading to IKK2 activation in BMMCs. SFKs regulate most of the SCF-induced signaling pathways [35]. Preincubation of the SFK inhibitor SU6656 impaired the activation of c-Kit, IKK2 and STATs but enhanced the activation of PKB/Akt and did not affect the proliferation (Supplementary Figures S2A and S2B). Next, we tested the role Jak2 known as an upstream kinase of STATs [36]. Shivakrupa et al., showed that Lyn is essential for the SCF-induced STAT activation in BMMCs [37]. Additionally we found that the Jak2 inhibitor AG490 did not influence the SCF-induced proliferation (Supplementary Figure S2C). This confirms that SFKs but not Jak2 are required for the SCF-induced STAT activation and the resulting proliferation in BMMCs. Consequently, we suggest that SFKs positively regulate the IKK2-STAT pathway and negatively regulate the PI3K-PKB/Akt pathway. Lyn negatively regulates the PI3K-PKB/Akt pathway and the resulting proliferation in mast cells [13,31]. Thus, we speculated that Lyn also mediates activation of the IKK2-STAT pathway. To test this hypothesis we used lyn −/− BMMCs. Indeed, activation of c-Kit, IKK2 and STATs were reduced with SCF (50 ng/ml). Lysates were analyzed by western blotting. B, C. BMMCs from ikk2 F/F or ikk2 Δ mice were stimulated with SCF (50 ng/ml). Lysates were analyzed by western blotting (B; * longer exposure) or cells were probed with [H 3 ]-thymidine and analyzed by β-counting (C). D, E. Peritoneal fluid from ikk2 F/F or ikk2 Δ mice was collected and analyzed for the presence of mast cells by flow cytometry. Contour graphs are shown, characterizing mast cells as CD117 + /IL-33R + cells (D). Total numbers of CD117 + /IL-33R + mast cells per peritoneum were calculated (this experiment was performed two times; ikk2 F/F mice, n = 3; ikk2 Δ mice, and n = 3) (E). (Continued ) www.impactjournals.com/oncotarget  Given that Ca 2+ is required for IKK2 activation [38], we tested the role of the PLCγ-mediated Ca 2+ mobilization (Supplementary Figure S3A) in the SCF-induced IKK2-STAT signaling. Pre-treatment with the PLC inhibitor U-73122 or the Ca 2+ chelator BAPTA-AM reduced the SCF-induced activation of the c-Kit-IKK2-STAT pathway whereas the proliferation was not affected (Supplementary Figures S3B-S3E).
Next, we examined the role of SFKs in the IKKdependent activation of STATs in HMC cells. Confirming and extending others [39] pre-treatment of HMC-1.1 cells with SU6656 blocked the phosphorylation of STAT5/3 and PKB/Akt ( Figure 4B). These data show that SFKs, as Lyn contribute to activation of the c-Kit-IKK2-STAT pathway in mast cells.

The c-Kit-mediated IKK2 activation does not result in NF-κB activation
IKK2 typically mediates NF-κB activation. Thus, we determined whether SCF or constitutively active c-Kit mutants induce IκBα phosphorylation, degradation and NF-κB activation. Confirming that IKK2 activation depends on Lyn-, PLCγ-and Ca 2+ , the SCF-induced IκBα phosphorylation was blocked in lyn −/− BMMCs ( Figure 4C To test the IKK2-IκBα signaling we stimulated with IL-33 a typical inducer of the MyD88-and IKK2dependent canonical NF-κB signaling [40]. In contrast to  Figure 4F). We next determined whether alternative pathways, independently of IKK2, mediate the SCF- induced IκBα phosphorylation in BMMCs. Therefore, we used ikk2 Δ BMMCs or the IKK-inhibitor VII. In both cases the SCF-induced IκBα phosphorylation was blocked ( Figure 4G, Supplementary Figure S4C). Together these results show that the IKK2-IκBα signaling is intact, and is activated by SCF and IL-33 whereas only IL-33 stimulation induced an IKK activation which is strong enough to induce IκBα degradation.
Finally, we determined the capacity of SCF to induce NF-κB activation. Therefore, we used NF-κB-

Figure 4: The SCF-induced IKK2 activation depends on SFKs but does not induce NF-κB activation. A. Wt-or lyn −/−
BMMCs were stimulated with SCF (50 ng/ml). Lysates were analyzed by western blotting. B. HMC-1.1 cells were treated with SU6656, lyzed and analyzed by western blotting. C. Wt-or lyn −/− BMMCs were stimulated with SCF (50 ng/ml). Lysates were analyzed by western blotting. D. HMC-1.1 cells were treated with SU6656, lyzed and analyzed by western blotting. E. BMMCs were stimulated with SCF or IL-33 (both 50 ng/ml) and lysates were analyzed by western blotting. F. HMC-1.1 cells were treated with the IKK-inhibitor VII and stimulated with IL-33. Lysates were analyzed by western blotting. (Continued ) www.impactjournals.com/oncotarget EGFP-MC/9 cells [41]. In these cells we also found the SCF-induced IKK2 activation but no IκBα degradation ( Figure 4H). Compared to IL-33, SCF did not induce EGFP expression or a strong cytokine production ( Figures  4I and 4J) indicating that SCF either induces a very weak or no NF-κB activation.
Why does SCF not induce IκBα degradation and NF-κB activation? We found that SCF in presence of cycloheximide, a protein-biosynthesis inhibitor, induces IκBα degradation (data not shown). Therefore, in presence of ongoing IκBα re-synthesis, the SCF-induced IκBα degradation is quantitatively not sufficient to induce a net loss of IκBα.
Together, we found that neither stimulation with SCF nor the expression of constitutively active c-Kit mediates IκBα degradation. Consequently, we assume that compared to IL-33, activated c-Kit mediates an alternative mechanism leading to a weak "subthreshold IKK2 activation" [28] which is insufficient to mediate complete NF-κB activation. Huang et al. showed that in addition to phosphorylation of the S177/S181-IKK2 motive, also the SFK-mediated phosphorylation of the Y188/Y199-IKK2 motive is critical for IKK2 activation [42][43][44]. We found that the SCF-induced IKK2 activation is Lyn-dependent. Therefore, we speculated that SCF induces IKK2 activation via a Lyn-dependent phosphorylation of Y188/ Y199-IKK2. Indeed, SCF induced the phosphorylation of Y199 which was reduced in lyn −/− BMMCs ( Figure 5A). In contrast, IL-33 did neither induce the phosphorylation of Y199-IKK2 nor a SFK-dependent IKK2 activation ( Figures 5B and 5C).

SCF induces tyrosine phosphorylation of IKK2 in BMMCs
We identified the PLCγ-Ca 2+ signaling as critical for the SCF-induced IKK2 activation. This indicated that the PLCγ-Ca 2+ signaling activates Lyn and therefore mediates phosphorylation of Y199-IKK2. As shown in Figures 5D and 5E, treatment with U-73122 or BAPTA-AM blocked the activation of Lyn and the resulting phosphorylation of Y199-IKK2. Taken together, these data indicate that the SCF-induced IKK2 activation in BMMCs depends on the PLCγ-Ca 2+ -Lyn pathway leading to Y199-IKK2 phosphorylation and therefore IKK2 activation.

Formation of the novel c-Kit-Lyn-TAK1 complex
Now, we examined the molecular mechanism leading to IKK2 activation. First, we expressed c-Kit in HEK293T cells to test whether endogenously expressed IKK2 associates with c-Kit. Immunoprecipitation of overexpressed c-Kit did not result in co-precipitation of IKK2 ( Figure 6A, lane 2)

Figure 4: (Continued ) The SCF-induced IKK2 activation depends on SFKs but does not induce NF-κB activation.
G. Ikk2 F/F or ikk2 Δ BMMCs were stimulated with SCF (50 ng/ml). Lysates were analyzed by western blotting. H. NF-κB-EGFP-MC/9 cells were stimulated with SCF (50 ng/ml). Lysates were analyzed by western blotting. I, J. NF-κB-EGFP-MC/9 cells were stimulated with SCF or IL-33 (both 50 ng/ml). Cells were analyzed for EGFP production by flow cytometry (I) or supernatants were analyzed by ELISA for IL-6 production (J). excluding association between c-Kit and IKK2. TAK1 is the upstream kinase of IKK2 [6]. Therefore, we speculated that additional TAK1 overexpression leads to co-precipitation of endogenous IKK2. Thus, we co-transfected HEK293T cells with TAK1 and c-Kit, performed a c-Kit-specific immunoprecipitation, and indeed detected IKK2 ( Figure 6A,  lane 4) showing that TAK1 mediates the interaction between c-Kit and IKK2. Confirming these results, we also found an association between c-Kit, TAK1 and IKK2 in c-Kit-specific immunoprecipitations from BMMC lysates ( Figure 6B).
Given that Lyn is critical for IKK2 activation, we assumed that Lyn also joins the c-Kit-TAK1-IKK2 complex. In Lyn precipitates we detected c-Kit and TAK1 ( Figure 6C) but not IKK2 ( Figure 6D) indicating that Lyn interacts with c-Kit and TAK1 but not directly with IKK2.
Since, TAK1 was detectable in c-Kit and Lyn precipitates we assumed that TAK1 is the critical determinant for complex formation between c-Kit, IKK2 and Lyn. To determine the role of TAK1 we transfected HEK293T cells with c-Kit or TAK1 alone or both . Given that Y199-IKK2 is a Lyn target, we tested whether Lyn activation is also increased in Lysates were subjected to a TAK1-specific immunoprecipitation and the precipitates were analyzed by western blotting. Furthermore, we also checked the equal expression of transfected proteins in the total cell lysates (input) by western blotting. www.impactjournals.com/oncotarget HEK293T cells co-overexpressing TAK1 and c-Kit. Indeed, in cells co-overexpressing both molecules, highly activated Lyn, co-precipitates with TAK1 and c-Kit (Supplementary Figure S5A, compare lane 3 and  4). Therefore, the preformed Lyn-TAK1-IKK2 complex is highly activated in presence of c-Kit.
Next, we investigated the role of the TAK1activity in this complex formation. The TAK1 activity and its association with IKK2 strongly depend on Lysine 63 (K63) ubiquitinylation [45]. Overexpression of the inactive K63W-TAK1 mutant [45] abrogated the formation of the c-Kit-Lyn-TAK1-IKK2 complex and therefore blocked the phosphorylation of Y199-IKK2 ( Figure 6E, compare lane 5 and 6). These data indicate that TAK1 ubiquitinylation mediates complex formation between c-Kit, IKK2 and Lyn. This expression model prompted us to investigate the role of TAK1 in BMMCs and HMC cells. Therefore, we used the TAK1 inhibitor 5Z-7-oxozeanol [46]. Pre-incubation of BMMCs with 5Z-7-oxozeanol reduced the activation of IKK2, STAT3 and STAT5 but not of PKB/Akt (Supplementary Figure S5B). In HMC cells 5Z-7-oxozeanol very potently blocked the activation of IKKs, STAT3/5 and also of PKB/Akt (Supplementary Figure S5C).
Here, we provide a mechanistical insight of the c-Kit-mediated IKK2 activation. Thereby, overexpression of c-Kit nearly mimics the effects occurring in SCFstimulated cells or in cells expressing constitutively active c-Kit. From these experiments, we conclude that c-Kit forms a complex with TAK, SFKs and IKK2. The complex formation is thereby stabilized by TAK1 and results in a SFK-dependent IKK2 activation. Consequently, TAK1 is pivotal to induce the c-Kit-mediated activation of IKK2, STAT3/5 in primary-and tumor-mast cells.

The conjoined activation of IKK2 by c-Kit-and MyD88-dependent pathways
Co-stimulation with SCF and IL-33 induces a potentiated IL-6 production [47] and proliferation in BMMCs. Both the potentiated proliferation ( Figure 7A), and cytokine production (Supplementary Figure S6A) depends on MyD88.
Next, we investigated the role of the novel PLCγ-Ca 2+signaling which activates Lyn and therefore IKK2 in BMMCs. Pharmacological inhibition of PLCγ, SFKs, IKKs and TAK1 as well as Lyn and IKK2 deficiencies reduced the potentiated proliferation ( Figures 7B-7F) and the IL-6 production ( Supplementary Figures S6B-S6F). In contrast, inhibition of MAPKs did neither affect the potentiated proliferation ( Supplementary Figures S6G and S6H) nor the potentiated IL-6 production in a adequate, biological relevant magnitude. (Supplementary Figures S6I and S6J).
IKK2 together with Lyn activate STAT3. There fore, we examined the role of STAT3 in BMMCs co-stimulated with SCF and IL-33. Inhibition of STAT3 with S3I-201 [48] eradicated the synergistic effect of co-stimulation on proliferation ( Figure 7G) but did not affect the potentiated IL-6 production (Supplementary Figure S6K).
Finally, we determined the effect of co-stimulation (SCF and IL-33) on the activation of TAK1 and IKK2. Single-stimulation with SCF or IL-33 induced the activation of TAK1 ( Figure 7H) and IKK2 ( Figure 7I). Co-stimulation did not increase the activation of TAK1 ( Figure 7H) but enhanced the activation of IKK2 and delayed the IκBα resynthesis ( Figure 7I). These data shows that the SCF-induced and Lyn-dependent IKK2 is targeted by MyD88-dependent pathways. This leads to enhanced IKK2 activity which results in potentiated, NF-κB-dependent cytokine production, and in STAT-dependent potentiated proliferation.

c-Kit supports IL-33-induced and TAK1-and IKK2-dependent signaling in HMC cells
In contrast to BMMCs, IL-33 induced a c-Kitdependent activation of Erk1/2, JNK1/2 and PKB/Akt in HMC cells and in BMMCs generated from kit D814V tg mice [19,47]. Here, we investigated whether TAK1 and IKK2 contribute to IL-33-induced and c-Kit-dependent signaling in HMC-1.1 cells. As shown in Figures 7J and 7K pre-incubation with the TAK1 inhibitor 5Z-7-oxozeanol or with the IKK-inhibitor VII blocked the IL-33-induced activation of PKB/Akt, JNK1/2 and Erk1/2. This shows that in contrast to BMMCs, constitutively active c-Kit together with TAK1 and IKK2 mediate the IL-33-induced activation of PKB/Akt, JNK1/2 and Erk1/2. Therefore, constitutively active c-Kit controls IL-33-induced and TAK1-and IKKdependent signaling and effector functions.

DISCUSSION
Here, we show that SCF and constitutively active c-Kit mutants activate TAK1 and IKK2. In contrast to TAK1 and IKK2 activation induced by antigen receptors, TIR-and TNFR superfamily family members [49][50][51], the c-Kit-mediated TAK1 and IKK2 activation is weak and results in IκBα phosphorylation but not in IκBα degradation and NF-κB activation. Instead, the c-Kitmediated TAK1 and IKK2 activation induces STATdependent differentiation, proliferation and survival of primary-and tumor-mast cells.
Our findings differ from an earlier publication by Wu et al., who reported that SCF induces IκBα phosphorylation and degradation in BMMCs. In a later paper, the same group reported that SCF induces IκBα phosphorylation but not degradation in BMMCs which is in accordance with our data. The reason for these different findings remains unknown [52,53]. We consistently found IκBα phosphorylation without degradation upon stimulation with SCF. Therefore, the relevant questions are: how does c-Kit mediate TAK1 and IKK2 activation www.impactjournals.com/oncotarget  -counting (A-G). H, I. BMMCs were stimulated with SCF or IL-33 (both 50 ng/ml) or co-stimulated with both. Co-stimulation was performed by pre-incubation of the cells with SCF for 30min (pre-SCF 30min) followed by stimulation with IL-33 (for 10, 30, 60, 90 min). Lysates were analyzed by western blotting.(Continued ) www.impactjournals.com/oncotarget and what is the difference of the IKK2 activation induced by TIR family members such as the IL-33R?
Both, c-Kit and IL-33R induce the activation of TAK1 and IKK2. Given that c-Kit does not induce IκBα degradation and NF-κB activation, this indicates that critical components for IκBα degradation (e.g. activation of MyD88-IRAK4-Traf6 signaling module) are not activated. Compared to IL-33R, c-Kit does not mediate the MyD88-and IRAK4-dependent (data not shown) pathways excluding the canonical pathway for IKK2 activation. Thus, we assumed an alternative mechanism.
Indeed, in BMMCs the SCF-induced IKK2 activation is mediated by the PLCγ-Ca 2+ -Lyn-dependent phosphorylation of IKK2 on Y199.  described that additional to phosphorylation of the S177/ S181-IKK2 motive, the SFK-dependent phosphorylation of the Y188/Y199-IKK2 motive is also critically involved in IKK2 activation [42][43][44]. Therefore, we speculate that the SCF-induced IKK2 activation also depends on the SFK-mediated Y199 phosphorylation. Interestingly, the SCF-induced phosphorylation of the S177/S181-IKK2 motive is transient whereas the phosphorylation of Y199-IKK persists up to 120min. This indicates that in contrast to the Y199-IKK2 phosphorylation, the phosphorylation of S177/S181-IKK2 is strongly regulated by phosphatases [54]. Interestingly, this suggests that the SCF-induced and long-persisting Y199-IKK2 phosphorylation primes IKK2 for additional stimulation with IL-33. Thereby, IKK2 phosphorylated on Y199 is targeted by the IL-33induced signaling leading to increased IKK2 activity and therefore to potentiated mast cell effector functions.
Consistent with BMMCs, we also found a SFKdependent IκBα phosphorylation in cells expressing constitutively active c-Kit. These data indicate that SCF in BMMCs, and constitutively active c-Kit, mediate IKK activation via SFKs.
The inefficiency of SCF or constitutively active c-Kit to induce IκB degradation leads to a model in which fewer IKK2 molecules are activated in response to c-Kit activation than in response to stimuli such as IL-33. Thus, the threshold of IKK activation which is necessary to induce IκBα degradation is not attained and therefore NF-κB still blocked.
Here, we also provide the mechanistical insight for this alternative IKK2 activation. IKK2 forms a complex with TAK1 and Lyn. This TAK1-IKK2-Lyn complex associates with c-Kit, via TAK1. The association of the TAK1-IKK2-Lyn complex with c-Kit activates Lyn which phosphorylates and activates IKK2. Since expression of the K63W-TAK1 mutant abrogated the association of the c-Kit-Lyn-TAK1-IKK2 complex, we hypothesize that TAK1 ubiquitinylation is essential to stabilize this complex and to mediate STAT3 activation. Which E3-Ligase mediates TAK1 ubiquitinylation is unknown. We presume a critical role of the E3-Ligase Cbl which is involved in c-Kit-mediated signaling in mast cells [55].
Importantly, in BMMCs and in HMC cells, the c-Kit-driven IKK activation also influences IL-33-induced signaling pathways, and therefore enables IL-33 to deploy its full biological effector functions. Therefore, the c-Kit-TAK1-IKK2 signaling not only mediates proliferation and survival, but also influences NF-κB mediated effector functions induced by TIR-family members.
These data show that the activated c-Kit-TAK1-IKK2 signaling module dramatically alters and enhances cellular effector functions. Therefore, dysregulation of the c-Kit-TAK1-IKK2 signaling module could induce pathological effector functions in inflamed tissues after infection, allergy and necrosis or supports tumor growth.
Especially the facts that (i) TAK1 or IKK2 inhibition induces cell death of tumor-but not of primary-mast cells [32] and that (ii) TAK1 or IKK2 inhibition overcomes the resistance of tumor cells to c-Kit inhibitors, suggest that targeting of the c-Kit- and were stimulated with IL-33. Lysates were analyzed by western blotting. www.impactjournals.com/oncotarget TAK1-IKK2 signaling module is a novel, specific and effective approach to therapeutically target pathogenic cellular effector functions.

Mice
Mice were maintained at the Animal Research Facility of the Universitätsklinikum Jena and Universität Frankfurt. Animal experiments were approved by the appropriate institutional and governmental committees for animal welfare. We used sex-and age-matched myd88 −/− , lyn −/− , ikk2 Δ [27], and either wild type (wt) littermates or ikk2 F/F -mice (cre − littermates) [27]. To induce deletion of IKK2, 6-8 week old ikk2 Δ or control mice (ikk2 F/F ; cre − mice) were injected once with 250 μL of poly (I:C) (Sigma) i.p. and were analyzed after 21d. [27] Cell culture

Immunoprecipitation
Cell lysates from HEK293T or BMMCs (prepared with lysis buffer supplemented with 0,5% NP40) were incubated (overnight) with a Lyn-, c-Kit-, or TAK1specific antibody (santa cruz) or goat-, mouse-, or rabbit-IgG as non-specific control immunoglobulins (Gentaur). Protein-G sepharose was added and incubated for 4 h. Precipitates were washed with lysis buffer and subsequently treated and boiled in 6 x Laemmli buffer.

Histology
Samples of back skin or ear pinna were fixed in 4% (v/v) buffered formalin and embedded in paraffin ensuring a crosssectional orientation. For detection of mast cells, 5 μm sections were metachromatically stained (Giemsa). Mast cell numbers were counted per ear or back skin section over a length of 1 cm. Mean of 3 sections per mouse was calculated ± SD.

Statistical analysis
All experiments were performed at least three times. Proliferation assays, and ELISAs were performed at least in a 6-fold determination. Cytokine concentration, stimulation indices (SI) and mast cell numbers per peritoneum are indicated as the mean of measurements ± standard error of the mean (SEM). The statistical analysis was performed with IBM SPSS Statistics version 21.0 (IBM, Ehningen, Germany). Statistical significance was assessed by students T-test. Statistical significance was accepted for p < 0,05 (*, p < 0.05; **, p < 0.01; ***, p < 0.001).

CONFLICTS OF INTEREST
The authors declare no competing financial interests.