KIT^D816V Induces SRC-Mediated Tyrosine Phosphorylation of MITF and Altered Transcription Program in Melanoma

The receptor tyrosine kinase KIT is important for melanocyte and mast cell development. Upon binding of the KIT ligand (also known as stem cell factor, SCF), KIT autophosphorylates its intracellular domain leading to recruitment of signaling proteins to the receptor and activation of downstream signaling pathways. Mutations in certain residues of KIT lead to a constitutively active and ligand-independent receptor and are found in several types of cancer including gastrointestinal stromal tumors, testicular seminomas, systemic mastocytosis, acute myeloid leukemia, and melanoma (1). Previously, it was believed that the loss of KIT expression facilitated melanomagenesis and increased metastasis (2, 3). However, more recently, The Cancer Genome Atlas Network identified KIT mutations and focal amplifications as major elements of the melanoma subtype, triple wild-type (4). Moreover, using imatinib to target melanoma harboring KIT alterations in clinical trials shows promising results (5–7).

The D816V KIT mutation is well characterized in systemic mastocytosis and acute myeloid leukemia. In contrast, although KIT^D816V was identified in melanoma many years ago, it remains uncharacterized (8). The KIT^D816V receptor shows a pattern of intracellular localization compared with wild-type KIT (9, 10). Interestingly, it has been shown that KIT^D816V is able to generate oncogenic signaling without the need to reach the cell membrane (10). Antibody staining of KIT in mucosal melanoma tissue samples carrying the D816V mutation showed intracellular localization (4). Moreover, KIT^D816V aberrantly activates oncogenic signaling pathways partially by virtue of its altered substrate specificity (11). The nuclear transcription factors STAT-1, STAT-3, and STAT-5 are recruited by KIT^D816V to the cytoplasm to amplify oncogenic signaling in leukemic cells. In this case, the activation of the survival factor AKT is crucial for cellular transformation (12, 13).

Wild-type KIT has been shown to signal to the melanocyte master regulator microphthalmia-associated transcription factor (MITF) that is known to control critical melanoma functions including metabolism and resistance to MAP kinase pathway inhibition (14, 15). Wild-type KIT engages the MAPK pathway leading to MITF phosphorylation at S73 and S409, thus affecting the transcription activation potential of MITF (16). However, in vivo studies have shown that these phosphorylation events are not important for coat color and eye development in mice (17), suggesting a more complex role of MITF phosphorylation in vivo. Like KIT, MITF is an essential mediator of melanocyte and mast cell development (18) and its involvement in melanomagenesis is well established (19). Interestingly, in transformed mast cells, KIT^D816V has been shown to upregulate MITF protein levels, resulting in abnormal mast cell proliferation and malignant mast cell disease (20). Collectively, these data suggest that MITF might be an important target of KIT^D816V in melanoma.

Here we show that a triple protein complex formation between KIT^D816V, MITF, and SRC family kinases results in tyrosine phosphorylation of MITF and subsequent effects on MITF-dependent gene expression. Phosphorylation of MITF was shown to be a crucial event mediating the oncogenic effects of KIT^D816V. Thus, we have defined a major oncogenic pathway involving MITF tyrosine phosphorylation in KIT^D816V-positive melanoma cells. More importantly, our findings suggest KIT^D816V-transformed melanoma cells to be sensitive to SRC family kinase inhibitors.

The HEK293T cells (Thermo Scientific Open Biosystems), B16F0 melanoma cells (LGC Standards AB), and A375 melanoma cells (ATCC) and 501 mel cells were cultured according to manufacturers' recommendations. All cell lines were obtained between 2009 and 2013 and routinely tested negative for Mycoplasma using the MycoAlert mycoplasma detection kit (Lonza). Cells were cultured for less than 2 months in our laboratory. Cell line authentication was not routinely performed.

B16F0 and A375 melanoma cells stably transfected (through G418 selection, 900 μg and 600 μg, respectively, after transfection) with empty CMV vector or KIT^D816V were plated (5,000 cells/well) in 96-well plates and assayed with PrestoBlue cell viability reagent (Life Technologies).

The number of live B16F0 or A375 cells were determined using PrestoBlue (see previous section) 48 hours after plating (5000 cells/well in 96-well plates). Cells were treated with SU6656 (Sigma-Aldrich) or dasatinib (Shanghai Yingxuan) at the following concentrations: 10 nmol/L, 50 nmol/L, 70 nmol/L, 150 nmol/L, and 200 nmol/L. As a control, cells were also treated with DMSO in a volume equivalent to the drug used (never exceeding 0.1 % of the total volume). CellTiter-Glo Luminescent Cell Viability Assay (Promega) was used to determine vemurafenib (Selleckchem) dose escalation (72 hours) assay.

The QuikChange mutagenesis kit (Stratagene) was used, according to the manufacturer's recommendations, to generate tyrosine-to-phenylalanine substitution mutants of MITF. The following primers, designed at bioinformatics.org/primerx, were used for this reaction:

• (MITF Y22F) Forward-5′ CCCCACCAAGTTCCACATACAGCAAGC 3′

• (MITF Y22F) Reverse-5′ GCTTGCTGTATGTGGAACTTGGTGGGG 3′

• (MITF Y35F) Forward-5′ GCACCAGGTAAAGCAGTTCCTTTCTACCAC 3′

• (MITF Y35F) Reverse-5′ GTGGTAGAAAGGAACTGCTTTACCTGGTGC 3′

• (MITF Y90F) Forward-5′ CTGTGAAAAAGAGGCATTTTTTAAGTTTGAGGAGCAGAGC 3′

• (MITF Y90F) Reverse-5′ GCTCTGCTCCTCAAACTTAAAAAATGCCTCTTTTTCACAG 3′

Isolated gel (SDS-PAGE) bands were minced, destained, and subjected to in-gel digestion by trypsin (sequencing grade; Promega) as described previously (21). The resulting peptides were analyzed by online reverse-phase C18 nanoscale liquid chromatography (LC) tandem mass spectrometry (MS). The experiments were performed on an EASY-nLC system (Proxeon Biosystems) connected to a LTQ-Orbitrap Velos (Thermo Electron) through a nano-electrospray ion source, as described previously (22). The peptides were separated by a linear gradient of increasing acetonitrile in 0.5 % acetic acid. The mass spectrometer was operated in data-dependent mode to automatically switch between full scan MS and MS/MS acquisition. Survey full-scan MS spectra were acquired in the orbitrap with resolution of 70,000 after accumulation to a target value of 1e6 in the linear ion trap. The ten most intense peptide ions with charge states 2 were sequentially isolated to a target value of 5e4 and fragmented with higher-energy collisional dissociation in the octopole collision cell and analyzed in the orbitrap with a resolution of 7,500. All LC/MS-MS files were processed with the MaxQuant software suite, as described previously (23). Tandem mass spectra were initially matched with a mass tolerance of 7 ppm on precursor masses and 0.02 Da for fragment ions, strict trypsin specificity and allowing for up to three missed tryptic cleavage sites. Cysteine carbamidomethylation (Cys +57.021464 Da) was searched as a fixed modification, whereas N-acetylation of protein (N-term +42.010565 Da), N-pyro-glutamine (Gln −17.026549 Da), oxidized methionine (+15.994915 Da) and phosphorylation of serine, threonine and tyrosine (Ser/Thr/Tyr +79.966331 Da) were searched as variable modifications. Identified peptides were filtered to an estimated false discovery rate < 0.01.

The phospho-specific antibodies against MITF were generated by immunizing rabbits with synthetic peptides (JPT Peptide Technology) with the following sequences; CLENPTK-pY-HIQQAQR, CRHQVKQ-pY-LSTTLA and CEKEAF-pY-KFEEQS corresponding to mouse MITF-M phosphorylated tyrosine 22, 35, and 90, respectively. The resulting antibodies were purified with affinity chromatography. Immunoblotting and immunoprecipitation was performed as described previously (24).

Transfection was performed using JetPEI (PolyPlus) transfection reagent according to manufacturer's recommendation with mouse MITF-M in P3XFLAG-CMV-14 and either KIT wild-type or KIT^D816V mutant. Anti-FLAG antibodies (Sigma-Aldrich) were used for immunoprecipitation and protein G beads (Dynabeads, Life Technologies) were added to pull down bound proteins. In addition, protein G Sepharose beads from GE Healthcare were employed as an extra control in coimmunoprecipitation experiments. Relative protein expression was quantified with ImageJ (Schneider and colleagues 2012) using the optical density signals obtained from Western blotting.

Coimmunoprecipitation of protein in the SFKs was performed in the lysate of transfected HEK293T cells using either anti-FLAG antibodies (Sigma-Aldrich), anti-c-Myc (clone 9E10, Roche), anti-V5 antibody (Sigma-Aldrich) to immunoprecipitate the respective SFK. The anti-FLAG antibodies were then used for immunoblotting to detect MITF-FLAG. Coimmunoprecipitation of endogenous SRC was carried out by adding anti-v-SRC antibodies (Ab-1, Calbiochem) to A375 cell lysate and processed as described above.

We investigated the phosphorylation state of MITF in cells expressing KIT^D816V. To do this, we coexpressed KIT^D816V and FLAG-tagged mouse MITF in HEK293T cells and used high-resolution orbitrap tandem mass spectrometry (MS) to identify phosphorylation sites on the MITF protein after immunoprecipitation using anti-FLAG antibody. This resulted in identification of three tryptic phosphopeptides derived from the N-terminal part of MITF that were tyrosine phosphorylated in the presence of KIT^D816V on Y22, Y35, and Y90, respectively (Supplementary Fig. S1). Multiple sequence alignments of MITF proteins from different species, including vertebrates and invertebrates, revealed that all three tyrosine sites are highly conserved and likely to have functional impact on MITF; the domain containing Y90 is missing from the zebrafish and Drosophila MITF genes (Supplementary Fig. S2).

To investigate whether MITF tyrosine phosphorylation is induced by KIT^D816V kinase activity and not by wild-type KIT, we overexpressed a FLAG-tagged MITF construct together with either wild-type KIT or KIT^D816V in B16F0 mouse melanoma cells. The MITF protein was isolated by immunoprecipitation of FLAG-tagged MITF and Western blots performed with the 4G10 anti-phosphotyrosine antibody. The results showed that KIT^D816V induced tyrosine phosphorylation of MITF (Fig. 1A, top, lane 1) whereas no phosphorylation was observed upon activation of wild-type KIT with KIT ligand (KITLG; Fig. 1A). KIT phosphorylation was also assessed by immunoprecipitating KIT and staining with the 4G10 antibody. This showed that KIT^D816V was tyrosine phosphorylated similar to KITLG stimulated wild-type KIT (Fig. 1A).

To further characterize these tyrosine phosphorylation sites, we generated and affinity purified phosphospecific antibodies against the phosphorylated Y22, Y35, and Y90 tyrosine residues of MITF. The antibodies only recognized the MITF protein when the respective sites were phosphorylated and not when they were mutated to a nonphosphorylatable phenylalanine, thus demonstrating their specificity (Fig. 1B). More importantly, these tyrosine residues were only phosphorylated in the presence of KIT^D816V and not in the presence of wild-type KIT. For example, using the pY22 antibodies, phosphorylation of Y22 of MITF was only detected in the presence of KIT^D816V and only when this residue was tyrosine, never when it was mutated to phenylalanine (Y22F; Fig. 1B). Similarly, the phosphospecific pY35 and pY90 MITF antibodies only recognized MITF in the presence of KIT^D816V when the respective residues were tyrosine and not when they were mutated to phenylalanine (Fig. 1B). These results were observed in B16F0 melanoma cells (Fig. 1B) as well as in HEK293T cells and A375 human malignant melanoma cells (Supplementary Fig. S3).

The SRC family kinases (SFKs) are nonreceptor tyrosine kinases and prominent downstream signaling components of KIT (1). The KIT phosphorylation sites Y568 and Y570 are crucial for SFK binding and activation. To determine whether SFKs are involved in KIT^D816V-induced tyrosine phosphorylation of MITF, B16F0 melanoma cells were transfected with KIT^D816V or KIT^{D816V, Y568/570F} that lacks the SFK-binding sites, together with FLAG-tagged MITF and tyrosine phosphorylation determined using the 4G10 antibodies. Our results show that tyrosine phosphorylation of MITF was triggered by KIT^D816V but not by KIT^{D816V, Y568/570F} (Fig. 1C). Furthermore, treating the cells with the SFK inhibitor SU6656, with or without KITLG, blocked the ability of KIT^D816V to mediate MITF tyrosine phosphorylation (Fig. 1C). These experiments suggest that SFKs are involved in mediating the signals from KIT^D816V to MITF.

Interestingly, when A375 melanoma cells were transfected simultaneously with FLAG-tagged MITF and KIT^D816V expression constructs, the KIT^D816V protein was coimmunoprecipitated with MITF, suggesting direct interactions between these proteins (Fig. 2A). However, the SFK nonbinding KIT^{D816V, Y568/570F} mutant did not coimmunoprecipitate with MITF (Fig. 2A), suggesting that KIT interaction with SFKs is required for the binding of MITF to the KIT^D816V protein. To further verify this, we used the same approach and immunoprecipitated the endogenous SRC protein and showed that MITF coimmunoprecipitated with SRC but only in the presence of KIT^D816V and not KIT^{D816V, Y568/570F} (Fig. 2B). Moreover, the level of MITF coimmunoprecipitation was severely decreased in cells treated with the SFK inhibitor SU6656 (Fig. 2B). This suggests direct binding between KIT^D816V, MITF, and SFKs.

To determine which SFK was involved in this interaction, we transiently transfected HEK293T cells with the different SFKs including BLK, BRK, SRC, FGR, FYN, HCK, LCK, LYN, and RAK, and showed that all these kinases were able to bind to MITF in the KIT^D816V transfected samples (Supplementary Fig. S4). The binding was severely reduced in the presence of SU6656 or KIT^{D816V, Y568/570F} (Supplementary Fig. S4). This suggests that oncogenic KIT^D816V interacts directly with MITF and that any of the above SFKs are needed for this association. The results also suggest that the binding between the SFKs and KIT depends on the kinase activity of SFKs.

Previous reports have shown that wild-type KIT does not influence the nuclear localization of MITF (25). However, as the SFK and KIT^D816V proteins are located in the cytosol and as they mediate MITF phosphorylation, we investigated the subcellular localization of MITF. B16F0 melanoma cells were transfected with either wild-type KIT or KIT^D816V together with FLAG-tagged wild-type MITF or MITF containing the different combinations of the Y22/35/90F mutations. The cells were then stained with PE-conjugated anti-KIT and Alexa Fluor 647–conjugated anti-FLAG antibodies against MITF-FLAG. Consistent with previous reports in melanoma cells (25), our results show that the wild-type melanocyte-specific MITF-M isoform was exclusively localized to the nucleus (Fig. 3A). The presence of wild-type KIT did not affect the localization of MITF (Fig. 3A; Supplementary Fig. S5). However, when coexpressed with KIT^D816V, a large proportion of the wild-type MITF protein was localized to the cytosol as well as in the nucleus (Fig. 3A). Similarly, when the Y22, Y35, and Y90 phosphorylation sites of MITF were mutated singly or in double mutant combinations, the MITF protein was localized to both the nucleus and the cytosol when coexpressed with KIT^D816V but was nuclear in presence of wild-type KIT (Fig. 3A; Supplementary Fig. S5). However, the MITF^Y22/35/90F triple mutant showed exclusive nuclear localization, indicating that all three tyrosine phosphorylation sites are important for the KIT^D816V-mediated cytoplasmic retention of MITF (Fig. 3A; Supplementary Fig. S5).

To further confirm that a proportion of MITF is retained in the cytosol in the presence of KIT^D816V, we fractionated transfected B16F0 melanoma cells into a cytosolic and a nuclear portion. Coexpression with KIT^D816V resulted in elevated levels of MITF protein in both the cytosolic and nuclear fractions, whereas wild-type KIT did not affect the level of MITF and the protein was only found in the nucleus (Fig. 3B). These results confirm our confocal microscopy findings that only KIT^D816V and not wild-type KIT can mediate cytosolic localization of MITF.

To show that endogenous MITF also partially localizes to the cytosol in the presence of KIT^D816V, we stained for the endogenous MITF protein in 501 mel melanoma cells transfected with either wild-type KIT or KIT^D816V. This showed that MITF was located exclusively in the nucleus (Fig. 3C) whereas in the KIT^D816V transfected cells, MITF was localized both in the nucleus and the cytosol (Fig. 3c). Our results indicate that the endogenous MITF protein is partially retained in the cytosol in the presence of KIT^D816V, presumably because of its interaction with KIT and SFKs.

To determine whether KIT^D816V affects MITF-mediated target gene expression, we performed qPCR analysis of MITF target genes (see Supplementary Table S1). siRNA against MITF (siMITF) in B16F0 cells resulted in knockdown of endogenous MITF expression to around 40 % compared with cells treated with negative control siRNA (Fig. 4A, sample 4). The knockdown also reduced expression of a number of genes confirming that these genes are indeed MITF targets (Fig. 4B, sample 4; Supplementary Fig. S6A; Supplementary Table S1). B16F0 cells stably transfected with KIT^D816V negatively affected the expression of genes involved in melanoma/melanocyte differentiation, cell-cycle inhibition, and tumor suppression (gene names indicated in green), whereas expression of genes involved in melanoma proliferation, survival, angiogenesis, and cell-cycle initiation (gene names in red) was increased (Fig. 4B, sample 2; Supplementary Fig. S6B). The addition of siMITF reduced the overall effects of KIT^D816V on the melanoma genes, indicating that the effects observed were mediated by MITF (Fig. 4B, sample 4; Supplementary Fig. S6C).

The addition of human KITLG to cells stably expressing KIT^D816V did not alter the gene expression pattern observed with KIT^D816V without treatment (Supplementary Table S1). In addition, mouse KITLG treatment that activates endogenous KIT resulted in increased expression of melanocyte differentiation genes. In summary, KIT^D816V, via MITF, negatively affects the expression of genes controlling melanocyte differentiation, cell-cycle inhibition, and tumor suppression, and positively regulates the expression of genes involved in proliferation, survival, angiogenesis, and cell-cycle initiation.

To determine whether MITF tyrosine phosphorylation is involved in the KIT^D816V-mediated effects on gene regulation, we repeated the gene expression analysis after transfecting the B16F0 cells with wild-type and mutant MITF constructs. Transfection of the different MITF constructs led to an increase in MITF mRNA levels (Fig. 4C; for all data values see Supplementary Table S2). Transient transfection of wild-type MITF also increased expression of all the target genes as compared with control cells transfected with empty vector (Fig. 4D, sample 4; Supplementary Fig. S6D).

Transfection of the triple MITF^{Y22F/Y35F/Y90F} mutant construct in the absence of KIT^D816V had similar effects on gene expression as wild-type MITF, suggesting that the transcription activation function of the triple mutant is similar to wild-type MITF (Fig. 4D, sample 5; Supplementary Fig. S6E). Transfection of wild-type MITF in cells stably expressing KIT^D816V increased expression of genes involved in survival, proliferation, angiogenesis, and cell-cycle activation (indicated in red), whereas genes controlling differentiation, tumor suppression, and cell-cycle inhibition were repressed (indicated in green in Fig. 4D, sample 2; Supplementary Fig. S6). However, KIT^D816V-expressing cells transfected with the triple MITF^Y22/35/90F mutant resulted in overall enhanced expression of all genes and did not selectively affect a specific class of genes (Fig. 4D, sample 3). Importantly, it did not lead to an increase in expression of genes responsible for melanoma survival or proliferation to the same level as wild-type MITF in the presence of KIT^D816V (Fig. 4D, sample 3; Supplementary Fig. S6F). Indeed, the gene expression profile of the MITF^Y22/35/90F triple mutant in the presence of KIT^D816V was comparable with cells transfected with only wild-type MITF (Fig. 4D, samples 3 and 4; Supplementary Fig. S6G). Thus, the results indicate that KIT^D816V signaling requires intact tyrosine phosphorylation sites of MITF to mediate the differential effects on gene expression.

The addition of human KITLG to cells stably expressing KIT^D816V showed the same gene expression pattern as cells without treatment (Supplementary Table S2). In addition, mouse KITLG treatment that activated endogenous KIT resulted in increased expression of melanocyte differentiation genes.

Hierarchical clustering analysis revealed that the two main clusters of genes tested were differentially affected by the presence of KIT^D816V. The first cluster of genes is involved in tumor suppression, differentiation, and cell-cycle inhibition (gene names indicated in green in Fig. 4B and D) whereas the second cluster drives melanoma proliferation, survival, angiogenesis, and cell-cycle initiation (gene names indicated in red). Moreover, KIT^D816V-transfected samples clustered into separate groups (Fig. 4B), illustrating that oncogenic KIT has different effects on MITF-dependent genes than wild-type KIT. Our results suggest that the effects of KIT^D816V on gene regulation depend on tyrosine phosphorylation of MITF.

The above results suggest that KIT^D816V signaling leads to MITF tyrosine phosphorylation which then has differential effects on the expression of MITF-target genes. To determine whether this affects cell function, we measured the proliferation of B16F0 and A375 melanoma cells stably transfected with KIT^D816V over three days. The proliferation assay showed that KIT^D816V-expressing cells proliferated more than 2-fold faster than nontransfected and CMV empty vector–transfected cells (Fig. 5A and B).Treatment with siMITF over 48–72 hours suppressed proliferation of nontransfected, CMV-transfected, and KIT^D816V-transfected B16F0 and A375 melanoma cells (Fig. 5A and B). This shows that MITF is necessary for proliferation of melanoma cells, including cells expressing KIT^D816V.

Cells stably transfected with either KIT^D816V, wild-type MITF, or MITF^Y22/35/90F alone exhibited a significantly enhanced proliferation rate compared with normal B16F0 and A375 cells (Fig. 5C and D). Cells transfected with both KIT^D816V and wild-type MITF resulted in the greatest proliferative effects. In contrast, cells cotransfected with MITF^Y22/35/90F and KIT^D816V showed similar proliferative ability as cells transfected with either construct alone (Fig. 5C and D). This suggests that the Y22, Y35, and Y90 phosphorylation sites of MITF are essential for the KIT^D816V-mediated effects on melanoma proliferation.

Our results demonstrate that MITF phosphorylation is dependent on the activity of both KIT^D816V and SFKs. To determine whether SFK inhibitors have an impact on the signaling between KIT^D816V and MITF, and consequently on the biological function of melanoma cells, we treated B16F0 and A375 cells with the SFK inhibitors SU6656 and dasatinib. Cells stably transfected with KIT^D816V and treated with DMSO, as a control condition, displayed increased viability compared with normal untransfected cells or cells transfected with an empty vector (Fig. 6A and B). Treating normal and CMV-transfected cells with increasing doses (10–200 nmol/L) of SU6656 or dasatinib did not significantly decrease the number of living cells. However, the viability of KIT^D816V-transformed cells was significantly reduced by treatment with either SU6656 or dasatinib (Fig. 6A and B). To study the effects of the BRAF^V600E mutation that is present in the 501 mel and A375 cell lines, we treated these cells with vemurafenib. Interestingly, the stably KIT^D816V-expressing cells were resistant to vemurafenib treatment (Fig. 6C and D). However, cell viability was significantly reduced in the presence of dasatinib. As expected the wild-type KIT-expressing cells were sensitive to vemurafenib and cotreatment with dasatinib did not yield any synergistic effects.

We have shown that KIT^D816V forms a protein complex with MITF and any of the multiple members of the SRC family of kinases. As a result, MITF is tyrosine phosphorylated on at least three sites leading to differential effects on transcription such that proliferation and survival of melanoma cells is increased.

The presence of KIT^D816V leads to increased expression of MITF target genes including Tbx2, Bcl2, Sox10, Cdk2, Hif1a, p35, and Diaph1 all of which have been implicated in melanoma proliferation, cell-cycle progression, suppression of senescence, survival, and invasion. Interestingly, the presence of KIT^D816V not only increases the expression of these genes, it also reduced expression of melanocytic marker genes such as Tyr, Tyrp1, Dct, and Mlana. It has been reported that overexpression of MLANA is reduced during melanoma progression (26). MLANA is also observed to be downregulated by HIF1α in melanoma cells that have switched to an invasive phenotype (27). Both of these reports fit with our study in that phosphorylated MITF increases HIF1α expression, whereas MLANA expression is reduced. Thus, it is tempting to speculate that the MITF-mediated suppression of Mlana is caused by the increase in Hif1α expression. Our proliferation assays showed that KIT^D816V, via MITF phosphorylation, enhances melanoma proliferation and survival, lend further support to the conclusion that KIT^D816V affects proliferation via MITF phosphorylation. It is not clear how the tyrosine phosphorylation of MITF differentially affects target gene expression. Perhaps the phosphorylated MITF affects interactions with important gene-specific cofactors such that some targets are avoided or preferred.

The melanocyte-specific MITF-M isoform is localized to the nucleus in melanoma cells (28). However, alternative subcellular location of MITF has been reported to be regulated in some cell types. For example, in osteoclast precursors, MITF is cytosolic but translocates to the nucleus in response to treatment with colony-stimulating factor-1 (CSF-1) and receptor activator of nuclear factor-κB ligand (RANKL; refs. 29, 30). These factors lead to signaling to MITF, thus disrupting the association of MITF with the signaling regulatory 14-3-3 complex, which keeps it in the cytosol (30). The 1B1b exon of MITF, which is present in all MITF isoforms except for the MITF-M isoform, has been postulated to regulate cytoplasmic shuttling (31). Recently, the TFEB and TFE3 transcription factors, which are closely related to MITF, have been shown to be important for regulating autophagy and lysosomal function. These proteins are phosphorylated by the mTOR pathway and maintained in the cytoplasm by interactions with the 14-3-3 complex in nutrient-rich conditions. However, upon starvation, mTOR becomes inactive, the TFEB/TFE3 proteins are not phosphorylated and are therefore sent to the nucleus where they activate expression of lysosomal and autophagy genes (32). In addition, the subcellular localization of MITF has also been suggested to be regulated in a similar fashion in pancreatic cancer cells (33). Our results suggest that, in the presence of KIT^D816V, MITF is at least partly cytoplasmic when it is tyrosine phosphorylated. Presumably, the increased cytoplasmic presence of MITF is due to interactions with SFKs and KIT^D816V. Unfortunately, our phospho-tyrosine antibodies against MITF cannot be used to determine whether the cytoplasmic fraction or nuclear fraction of MITF is phosphorylated. However, as the subcellular location of the Y22/35/90F MITF triple mutant was not affected by KIT^D816V, whereas that of wild-type MITF was, it seems likely that subcellular localization is affected by tyrosine phosphorylation. Although it is possible that nuclear MITF is not influenced by KIT^D816V, transcriptional activation of selected MITF targets was significantly reduced by mutating the tyrosine phosphorylation sites of MITF. This suggests that a major effect of the tyrosine phosphorylation is to modulate gene transcription such that only a certain set of genes are transcribed.

SFKs are important signaling components in many types of human tumors, where they regulate proliferation, invasion, angiogenesis, and motility of cancer cells (34). Several investigators have reported increased SRC kinase activity in melanoma cells (35–37). Among the many pathways activated by SFKs are STAT3 and STAT5 that are important for melanoma cell proliferation and cell survival (35, 38, 39). SFKs have been suggested as promising pharmacologic targets in melanoma (35, 36, 40). More specifically, BRAF inhibitor–resistant melanoma tumors, in which increased SFK signaling is driven by the EGF receptor, are especially sensitive to dasatinib treatment. Treatment with dasatinib decreases both tumor growth and proliferation (41, 42). Although dasatinib was originally described to be a dual specificity SRC/ABL inhibitor (43), it was later shown also to inhibit KIT, including KIT^D816V. We have previously demonstrated that the D816V mutation in KIT leads similar kinase specificity of KIT as that of SRC (11). Indeed, we found that KIT^D816V transformed melanoma cells were hypersensitive to SFK inhibitor treatment. Our results thus suggest that melanoma cells in which the KIT^D816V mutation is present are sensitive to SRC inhibitors and dasatinib.

L. Rönnstrand is a consultant/advisory board member for OncoSignature AB. No potential conflicts of interest were disclosed by the other authors.

Conception and design: B. Phung, C.R. Goding, L. Rönnstrand

Development of methodology: B. Phung, K. Bergsteinsdottir

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B. Phung, J.U. Kazi, A. Lundby, J.V. Olsen, L. Rönnstrand

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B. Phung, J.U. Kazi, A. Lundby, J. Sun, G. Jönsson, J.V. Olsen, E. Steingrímsson, L. Rönnstrand

Writing, review, and/or revision of the manuscript: B. Phung, J.U. Kazi, A. Lundby, K. Bergsteinsdottir, C.R. Goding, E. Steingrímsson, L. Rönnstrand

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Phung

Study supervision: B. Phung, E. Steingrímsson

The 50l mel cells were provided as a kind gift from Dr. Ruth Halaban. The authors would like to thank Susanne Bengtsson for expert technical assistance for guidance and thoughtful comments.

This work was supported through grants from the Swedish Cancer Society, the Icelandic Research Fund, the Swedish Research Council, Gunnar Nilsson Cancer Society, Alfred Österlund Foundation, Skåne University Hospital Funds, Skåne University Hospital Cancer Foundation, the Royal Physiographical Society (Lund, Sweden), the Swedish Foundation for International Cooperation in Research and Higher Education (STINT), and the Research Fund of the University of Iceland.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.