Sustained Mutant KIT Activation in the Golgi Complex Is Mediated by PKC-θ in Gastrointestinal Stromal Tumors

The genetic aberrations associated with gastrointestinal stromal tumors (GIST) are among the best characterized of all human tumors. The activation of receptor tyrosine kinases, including KIT or platelet-derived growth factor receptor alpha (PDGFRA), is a primary step in GIST tumorigenesis (1). Mutations of KIT and PDGFRA are mutually exclusive in GISTs; activating mutations of KIT are present in 75% of GISTs and mutations of PDGFRA are present in approximately 35% of GISTs that lack KIT mutations (2–4). KIT mutations occur most frequently in exon 11 (∼70%), within the juxtamembrane domain, and exon 9 (∼9%). Although rare (less than 2%), mutations in exons 13 and 17, within the ATP binding pocket and kinase activating loop, have been reported (5). These mutations typically lead to activation and overexpression of KIT (6).

In colon cancers expressing wild-type KIT (WT-KIT), activation of WT-KIT is achieved by binding to a ligand known as stem cell factor (SCF), and sustained activation of WT-KIT requires PKC-δ, which promotes recycling of activated KITs in endosomes (7). However, the mechanism by which sustained activation and overexpression of mutant KIT (MT-KIT) occurs in other cancers, including GISTs, acute myelogenous leukemia, and mastocytosis, has yet to be elucidated (8). Furthermore, the cellular localization of MT-KIT differs from that of WT-KIT and is not fully understood. Most WT-KITs are localized to the plasma membrane before ligand binding, but some MT-KITs are abnormally localized to the intracellular compartment (9, 10).

Mutant proteins are often censored by the endoplasmic ER quality control (ERQC) system, are less stable than the WT protein, and are rapidly degraded through various protein degradation pathways. The purpose of ERQC is to remove misfolded proteins that may potentially cause damage to normal cellular functions, and a variety of substrates for ERQC have been identified (11–13). For example, the folding-deficient ΔF508-CFTR protein has decreased stability that is rapidly degraded by the proteasome pathway (14, 15). Because receptor proteins are generally transported to the plasma membrane via the ER and Golgi, MT-KIT expression is likely regulated by ERQC.

Several factors are expected to be involved in the regulation of MT-KITs. One such factor is Protein Kinase C-θ (PKC-θ), which is frequently overexpressed in GISTs and is used along with KIT as a diagnostic marker for GISTs (16). PKC-θ has a discrete expression pattern and plays important roles in T-cell activation, skeletal muscle signal transduction, and neuronal differentiation (17, 18). Despite its frequent overexpression in GISTs, the role of PKC-θ in the tumorigenesis of GISTs has not been well characterized.

In this study, we found that the stability of MT-KITs was much lower than WT-KIT, regardless of mutation type, and that these MT-KITs were mainly degraded through the proteasome. We also demonstrated that MT-KITs were abnormally localized to intracellular compartments, including the ER and the Golgi. Moreover, we found that PKC-θ was exclusively and strongly expressed in GISTs and leads to the stabilization of intracellular MT-KITs in the Golgi complex. We also demonstrated a strong correlation between the expression of PKC-θ and MT-KITs in 50 human GIST tissues and correlations between PKC-θ overexpression and clinicopathological parameters, such as high tumor grade, frequent recurrence/metastasis, and poor patient survival in 154 human GIST tissues. Taken together, we propose that sustained overexpression and activation of MT-KIT can be accomplished by PKC-θ-mediated stabilization of MT-KIT.

For in vitro KIT expression studies, Cos-7, HEK293, and HeLa cell lines were used. GIST882 (K642E; ref. 19), GIST430 (exon 11 in-frame deletion; ref. 20), GIST430 (exon 11 in-frame deletion and V654A; ref. 20), and GIST48 (V560D; ref. 20) cell lines expressing MT-KIT (kind gift from Dr. Sebastian Bauer, University Hospital Essen, Essen, Heidelberg, Germany) were cultured with Iscove's Modified Dulbecco's Medium supplemented with 15% FCS (Life Technologies) and 1% penicillin/streptomycin. GIST cell lines were established by Prof. Jonathan Fletcher (Boston, MA). The colon cancer cell lines (DLD-1, LS174T, and Colo320DM) expressing WT-KIT were cultured with RPMI1640 medium supplemented with 10% FBS (Life Technologies) and 1% penicillin/streptomycin. For screening the expression of PKC isoforms, colon cancer cell lines (HCT116 and RKO), leukemia cell lines (HMC-1 and Kasumi-1), small cell lung cancer cell lines (H69, H128, and H209), and control cell lines (HepG2 and NIH3T3) were used.

DLD-1, HCT116, RKO, Cos-7, HEK293, Colo320DM, Ls174T, H69, H128, H209, HepG2, NIH3T3, and HeLa were purchased from the Korean Cell Line Bank (Cancer Research Institute, Seoul, Korea). Kasumi-1 and HMC-1 were purchased from American Type Culture Collection. Within 6 months of this study, cell line authentication was performed through sequencing analysis of KIT, microscopic examination, qRT-PCR, and Western blotting analysis to monitor morphology, growth patterns, mutation status, and expression of KIT proteins. In addition, cell lines in use were regularly screened to monitor mycoplasma infections on a bimonthly basis.

Both fresh-frozen and formalin-fixed paraffin-embedded GIST tissue samples from 50 patients and another 104 formalin-fixed paraffin-embedded GIST samples were used in this study. All specimens were obtained via surgical resection, and some of the clinicopathologic findings of the 104 formalin-fixed paraffin-embedded GIST samples had been reported previously (21). The specimens were obtained from the archives of the Department of Pathology, Yonsei University, Seoul, Korea, and from the Liver Cancer Specimen Bank of the National Research Resource Bank Program of the Korea Science and Engineering Foundation of the Ministry of Science and Technology. GIST tissues were subjected to immunohistochemical analysis for KIT and PKC-θ. Authorization for the use of these tissues for research purposes was obtained from the Institutional Review Board of the Yonsei University of College of Medicine.

To construct KIT expression vectors with a 3xFLAG tag at the C-terminal end with or without tags, the KIT coding region was amplified by PCR using cDNA from DLD-1 cells and was cloned into a PCMV vector. To construct the PKC-θ expression vector, the PKC-θ coding region was amplified by PCR using cDNA from GIST882 cells and was cloned into a PCMV vector. Primers used for the construction of PKC-θ and mutant KIT expression constructs are listed in Supplementary Table S1. The generation of the other PKC constructs and the WT KIT expression vector was described previously (22). To downregulate PKC-θ, siRNA against PKC-θ was used and the sequences are as follows: 5′-AAACCACCGUGGAGCUCUACU(UU)-3′.

Reverse transcription was performed using the M-MLV reverse transcriptase (Life Technologies). The qRT-PCR was performed using SYBR Premix Ex Taq II (TaKaRa) and the ABI PRISM 7500 Sequence Detector (Applied Biosystems). The amount of KIT mRNA was normalized to that of GAPDH mRNA. Primer sequences are listed in Supplementary Table S1.

HeLa cells transfected with KIT expression vectors were biotin labeled 48 hours after transfection. Referring to the manufacturer's instructions, biotinylation, cell lysis, isolation of labeled proteins, and protein elution was subsequently performed using the Pierce Cell Surface Protein Isolation Kit (Thermo Scientific).

Whole-cell lysates were prepared using passive lysis buffer (Promega) with a protease inhibitor cocktail (Roche). The membranes were incubated with primary antibodies against GAPDH (Trevigen), FLAG (Sigma-Aldrich), KIT (Dako), HA (Santa Cruz Biotechnology), phospho-KIT (Invitrogen), PKC-θ, phospho-PKC-θ (Cell Signaling Technology) for 1 hour at room temperature. Western blot images were analyzed with a LAS 4000 mini camera (Fujifilm). Immune complexes of WT and MT-KIT proteins were collected by gently rocking 1 mg of total protein on an orbital shaker with prewashed anti-FLAG M2-agarose affinity gel (Sigma-Aldrich) at 4°C. The immune complexes bound to the affinity gel were washed and then boiled with a 100 mmol/L Tris–HCl–1% SDS solution to elute the complexes. The relative density of each band was quantified by ImageJ (NIH) software.

Cells grown on slides were rinsed with phosphate-buffered saline (PBS), fixed with 4% paraformaldehyde for 15 minutes, and permeabilized in 0.2% Triton X-100 in PBS. The slides were incubated with primary antibody for 1 hour and incubated for 50 minutes with the appropriate fluorescent-labeled secondary antibody (Invitrogen). All images were obtained using a LSM700 confocal microscope (Carl Zeiss).

For immunohistochemical analysis (IHC), 4-μm thick sections were obtained from formalin-fixed and paraffin-embedded GIST tissue specimens of 154 patients. IHC was performed using the Ventana Discovery XT autoimmunostainer (Ventana) with antibodies against KIT (1:200, Abcam), CD34 (1:100, DAKO), PKC-δ (1:100, BD Biosciences), and PKC-θ (1:150, BD Biosciences), based on the manufacturer's standard protocol. IHC results were classified into three grades based on the intensity of staining and proportion (negative, focal positive, and strong positive). The IHC results were reclassified into two categories for statistical analysis. Negative and focal positive cases were classified as negative expression, and cases with strong positive staining were classified as positive expression.

Because the overexpression of MT-KITs is common in GISTs with KIT mutations, we aimed to elucidate the mechanism of MT-KIT overexpression. In order to compare the protein expression of WT and mutant KITs, we generated one WT-KIT and four MT-KIT expression vectors. Each mutant KIT vector contained a representative mutation that has been previously reported in a specific domain of the KIT protein in GISTs. After the generation of the WT-KIT expression vector, MT-KIT-V560D (Juxtamembrane domain), MT-KIT-K642E (Tyrosine kinase 1 domain), MT-KIT-V654A (Tyrosine kinase 1 domain), and MT-KIT-D820A (Tyrosine kinase 2 domain) constructs were generated by point mutagenesis (Fig. 1A). To compare the relative stabilities of WT and MT-KIT proteins, the five vectors were transfected into HeLa cells, and mRNA and protein expression levels were measured 48 hours after transfection by qRT-PCR and Western blotting. We found that the KIT mRNA expression levels from the five vectors were similar, while the protein expression levels were remarkably different, at 27% (V560D), 33% (K642E), 60% (V654A), and 31% (D820A) of the protein expression level of the WT-KIT vector (Fig. 1B). We next evaluated protein stability by blocking protein synthesis with cycloheximide (CHX) at 40 hours after transfection and harvesting protein at 4 and 8 hours after CHX treatment. Western blotting analysis showed that WT-KIT proteins were stably detected even 8 hours after CHX treatment, while the expression levels of the four MT-KIT proteins were rapidly decreased by CHX (Fig. 1C). These findings indicated that overexpression of MT-KITs in GISTs is not due to an increased innate stability of MT-KIT proteins; rather, an unknown complementary mechanism is required to block MT-KIT degradation.

We previously demonstrated that WT-KIT proteins are expressed on the cell surface and internalized into the cytoplasm after SCF binding in colon cancer cell lines (7). To compare the localization of WT-KIT and MT-KITs, we examined the localization of MT-KIT proteins in the transfected HeLa cells by immunofluorescent microscopy. Expression of WT-KIT proteins was clearly detected only on the cell surface. Conversely, MT-KIT-V560D and MT-KIT-K642E were mostly expressed in intracellular regions, while MT-KIT-V654A and MT-KIT-D820A were localized both to the plasma membrane and intracellular regions (Fig. 2A). It was also noted that MT-KIT-V560D, MT-KIT-V654A, and MT-KIT-D820A accumulated in a specific intracellular region, while MT-KIT-K642E showed diffused localization. These findings were further demonstrated by a biotinylation assay. After biotinylation of cell surface proteins from HeLa cells transfected with each vector, KIT expression on the cell surface and in cell lysates were analyzed. WT-KIT, MT-KIT-V654A, and MT-KIT-D820A were detected as biotinylated forms (surface forms), while no biotinylated forms of MT-KIT-V560D and MT-KIT-K642E were detected (Fig. 2B). We also evaluated the localization of endogenous WT-KIT and MT-KITs in colon cancer and GIST cell lines. We found clear cell surface expression of endogenous WT-KIT proteins in colon cancer cell lines (DLD-1, Colo320DM, and LS174T cells) and mostly intracellular expression of the endogenous MT-KIT proteins in GIST cell lines (GIST882, GIST430, GIST430(V654A) and GIST48 cells) (Supplementary Fig. S1).

Next, we sought to determine which protein degradation pathway was responsible for the low expression of MT-KITs. We performed an inhibitor assay targeting three major pathways: proteasome-, lysosome-, and autophagy-mediated protein degradation pathways (23, 24). HeLa cells were transfected with each vector and treated with CHX for 4 or 8 hours before harvest. At the same time, the protein degradation pathways were inhibited with a proteasome inhibitor (MG132), lysosome inhibitor (Bafilomycin A1), or autophagy inhibitor (3-Methlyadenine). Western blotting analysis showed that WT-KIT protein levels were slightly decreased following CHX treatment, and while this decrease was completely inhibited by MG132, Bafilomycin A1 and 3-Methlyadenine had no effect. On the other hand, KIT-V560D, KIT-K642E, KIT-V654A, and KIT-D820A were rapidly degraded after CHX treatment, and the degradation was efficiently blocked by MG132 treatment. Bafilomycin A1 and 3-Methlyadenine had little effect, except that Bafilomycin A1 slightly rescued the degradation of KIT-V560D (Fig. 3A).

Because the enhanced degradation of mutant proteins is directly linked to the abnormal localization in cells, we further characterized the specific intracellular localization of MT-KITs. HeLa cells were transfected with the WT-KIT and MT-KIT expression vectors and stained with KIT/Calnexin (ER marker) or KIT/GM130 (cis-Golgi marker). Subsequent immunofluorescent microscopic analysis showed that V560D and K642E broadly colocalized with the ER, while V654A and D820A only partially colocalized with the ER (Fig. 3B). Furthermore, V560D, V654A, and D820A were almost completely colocalized with the Golgi complex, while K642E showed only slight colocalization with the Golgi complex. WT-KITs did not colocalize with the ER or the Golgi complex (Fig. 3B and C). To confirm these findings, a deglycosylation assay was performed. Cell lysates from HeLa cells transfected with the KIT expression vectors were incubated with Peptide -N-Glycosidase F (PNGase F) or Endoglycosidase H (Endo H) and subsequently analyzed by Western blotting. We detected a deglycosylated single band with lower molecular weight after PNGase F treatment in both WT-KIT and MT-KIT samples, which indicates that the upper band of all the MT-KITs was the mature form (complex glycosylated form). On the other hand, Endo H treatment only affected the lower MT-KIT band, indicating that the lower band represents the immature form (core-glycosylated form; Fig. 3D). These results were also supported by the result that Brefeldin A (ER to Golgi transport blocker) treatment induced accumulation of the lower band for both WT-KIT and the MT-KITs (Supplementary Fig. S2). Thus, the upper MT-KIT band represents Golgi-retained or plasma membrane bound forms and the lower band represents ER-retained forms (not fully glycosylated). Moreover, these abnormally localized MT-KITs are subject to rapid ER-associated degradation (ERAD) to relieve unfolded protein-mediated ER stress (25).

We previously demonstrated that PKC-δ contributes to SCF-induced activated WT-KIT protein recycling by blocking lysosome-mediated degradation and promoting recycling to the plasma membrane (7). Despite the generalized overexpression of KIT proteins in GISTs with KIT mutations, we herein demonstrated that MT-KITs were more unstable than WT-KIT. Therefore, we suspected that a PKC isoform strongly expressed in GISTs may also mediate MT-KIT overexpression. To validate this hypothesis, KIT expression vectors were cotransfected into HeLa cells with PKC expression vectors (conventional, unconventional, and atypical) and the expression of KIT was analyzed by Western blotting. Compared with the control that was cotransfected with EGFP, the expression level of WT-KIT was slightly increased by cotransfection with PKC-θ or PKC-δ (1.2-fold or 2-fold, respectively). The expression of KIT-V560D was almost doubled after cotransfection with PKC-θ, and the expression of KIT-K642E was variably increased by PKC-δ (2-fold), PKC-ϵ (1.6-fold), PKC-η (1.5-fold), PKC-θ (2.8-fold), and PKC-I (1.3-fold). In addition, the expression of KIT-V654A was also variably increased by PKC-δ (1.5-fold), PKC-ϵ (1.8-fold), PKC-η (2.5-fold), PKC-θ (3.2-fold), PKC-I (2.7-fold), and PKC-ζ (3-fold). For KIT-D820A, cotransfection with PKC-δ (1.2-fold), PKC-η (1.4-fold), PKC-θ (1.9-fold), and PKC-I (1.3-fold) leads to an increase in expression (Fig. 4A–E). Thus, the expression of WT-KIT and the four MT-KITs was dependent on multiple PKC isoforms, particularly PKC-θ.

Next, we screened various KIT relevant or irrelevant cancer cell lines, including colon cancer cell lines (WT-KIT), GIST cell lines (MT-KITs), leukemia cell lines (MT-KITs), small cell lung cancer (SCLC) cell lines (WT-KITs), and control cell lines (no expression of KIT), for PKC isoform expression. KIT expression was low in the SCLC cell lines and was analyzed separately (Supplementary Fig. S3). Western blotting analysis showed that various combinations of PKC isoform expression were detected in the cell lines. However, PKC-θ was strongly almost exclusively expressed in GIST cell lines, indicating that PKC-θ might play a critical role in the overexpression and localization of MT-KITs in GISTs (Fig. 4F).

Next, we sought to determine whether PKC-θ binds to MT-KITs to determine how PKC-θ alters MT-KIT expression. To avoid the use of a PKC activator (such as PMA), which has too many undefined roles, we first demonstrated that PKC-θ was autophosphorylated in three different transfected cell lines, including HeLa, Cos-7, and HEK293 cells (Supplementary Fig. S4). We then performed an immunoprecipitation following the cotransfection of KIT and PKC-θ expression vectors in HeLa cells. The immunoprecipitation assay showed that PKC-θ efficiently binds to KIT proteins independent of mutation status (Fig. 5A). This interaction was confirmed using immunofluorescent microscopy, which showed that PKC-θ colocalized with both WT-KIT and the MT-KITs in the intracellular compartment (Fig. 5B). An almost complete colocalization between endogenous MT-KITs and PKC-θ was also observed in all four GIST cell lines (Supplementary Fig. S5).

To identify the specific intracellular region where KITs interact with PKC-θ, we stained for markers of the ER or the Golgi complex in GIST cell lines. We found that endogenous MT-KITs barely colocalized with the ER, but almost completely colocalized with the Golgi complex (Fig. 5C and D). These results were not consistent with our findings using HeLa cells transfected with MT-KITs. The distinct localization of endogenous and synthetic MT-KITs supported the notion that PKC-θ regulates the localization of MT-KITs in GISTs. Because PKC-θ is completely absent in HeLa cells, we determined the localization of WT-KIT and MT-KITs in HeLa cells transfected with both KIT and PKC-θ expression vectors. WT-KIT partially accumulated in the Golgi complex after PKC-θ expression. KIT-V560D and KIT-K642E broadly colocalized with the ER when PKC-θ was not expressed, slightly colocalized with the ER in presence of PKC-θ, and more specifically colocalized with the Golgi complex. In addition, we found that KIT-K654A and KIT-D820A only colocalized with the Golgi complex after PKC-θ expression (Fig. 5E and F). We further determined the endogenous expressions and localizations of MT-KITs after PKC-θ knockdown using siPKC-θ. Immunofluorescent microscopic analysis showed that downregulation of PKC-θ induced complete disappearance of the characteristic and focused localizations of MT-KITs in the Golgi complex (Supplementary Fig. S6A and S6B). In accordance with these results, Western blotting analysis showed that MT- KIT expressions were dramatically decreased 72 hours after siPKC-θ treatment in GIST cell lines (Supplementary Fig. S6C). To address whether MT-KIT expression is transcriptionally regulated by PKC-θ, we measured the KIT mRNA expression level with or without PKC-θ knockdown in four GIST cell lines via qRT-PCR, and found that the KIT mRNA expression was only slightly changed by PKC-θ knockdown, which shows that MT-KIT expression is regulated at the protein level (Supplementary Fig. S6D).

To further elucidate whether PKC-θ kinase activity is required for its function in MT-KIT protein stability and cellular localization, we generated a mutant PKC-θ expression vector with kinase-activating mutation (A148E) and mutant PKC-θ expression vector with kinase-inactivating mutation (K409R; Supplementary Fig. S7A; ref. 26). Then, each MT-KIT expression vector was transfected into HeLa cells with PKC-θ, PKC-θ-A148E, or PKC-θ-K409R, respectively. On Western blotting analysis, we found that MT-KIT expression was more increased by PKC-θ-A148E than by normal PKC-θ, regardless of KIT mutation status (Supplementary Fig. S7B and S7C). It should also be noted that the immature form of MT-KITs was more clearly detected by coexpression of PKC-θ-A148E, which suggests that PKC-θ kinase activity is required to inhibit ER-mediated degradation of MT-KITs. On the other hand, MT-KIT expression was only slightly increased by coexpression of PKC-θ-K409R with weak kinase activity (Supplementary Fig. S7B and S7C). In addition, we examined the cellular localization of MT-KIT proteins in four GIST cell lines (GIST430, GIST430(V654A), GIST882, GIST48) after treatment of Rottlerin, which is an inhibitor of PKC-θ kinase activity (27). Immunofluorescent microscopic analysis showed that Rottlerin treatment effectively disturbed the characteristic localization of MT-KITs in the Golgi complex (Supplementary Fig. S7D). These results indicate that PKC-θ mediates intracellular stabilization of MT-KITs, which eventually leads to the retention of MT-KITs in the Golgi complex, and PKC-θ kinase activity is crucial for its function in MT-KIT protein stability and cellular localization.

If PKC-θ plays an important role in MT-KIT-mediated GIST tumorigenesis, constant activation of MT-KIT would be required. Because intracellular MT-KIT proteins are expected to be degraded in the ER-mediated proteasome pathway, we hypothesized that PKC-θ-mediated stabilization of activated MT-KIT proteins might contribute to GIST tumorigenesis. Therefore, we used phospho-KIT–specific antibodies to evaluate KIT phosphorylation at three sites (Y703, Y721, and Y936), which have been well characterized as binding sites for downstream signaling molecules involved in the PI3-K and MAPK pathways, following the cotransfection of KIT and PKC-θ expression vectors in HeLa cells. Phosphorylation of WT-KIT was not detected at any of the three tyrosine sites when expressed alone, but phosphorylated WT-KIT was detected following PKC-θ coexpression, indicating that Golgi-retained WT-KITs are phosphorylated and activated. All MT-KITs were detected as phosphorylated forms when expressed alone, and a significant increase in phosphorylated MT-KITs were detected following PKC-θ coexpression (Supplementary Fig. S8).

Although the overexpression of MT-KITs and PKC-θ has previously been reported, the correlation between these two proteins has not been well characterized. Based on our findings, we performed an expressional correlation study between MT-KITs and PKC-θ in 50 GIST cases. We examined the expression of MT-KIT, PKC-θ, and PKC-δ by Western blotting. The expression of PKC-δ was used as a control. We found that 41 out of 50 GISTs had a KIT mutation, and most cases had MT-KIT protein overexpression. Overexpressed MT-KITs were mostly phosphorylated. The expression of PKC-θ and PKC-δ was detected in a subset of GISTs. Furthermore, most PKC-θ proteins were highly phosphorylated in cases with high PKC-θ expression (Fig. 6A). Statistical analysis showed a strong and statistically significant positive correlation between MT-KIT and PKC-θ expression (Pearson correlation coefficient = 0.682, P < 0.000001), and no significant correlation between KIT and PKC-δ expression (Pearson correlation coefficient = 0.257, P = 0.071; Fig. 6B and C). The clinicopathological characteristics of the 50 GISTs evaluated are summarized in Supplementary Table S2.

To analyze the clinical implications of PKC-θ overexpression in GIST patients, IHC for both PKC-θ and PKC-δ was performed using 50 GISTs. PKC-δ was used as a control. For PKC-θ expression, 30 cases were categorized as positive for PKC-θ, and 20 cases were categorized as negative. Analysis of PKC-θ or PKC-δ expression with various clinicopathological parameters showed that PKC-θ expression significantly correlated with high tumor grade (P = 0.005), frequent recurrence/metastasis (P = 0.008), and poor patient survival (P = 0.033; Supplementary Table S3). No significant correlations between PKC-δ and clinicopathological parameters were found. Importantly, Kaplan–Meier survival curves showed that both disease-free survival (P = 0.014) and overall survival (P = 0.063) of GIST patients without PKC-θ expression were longer than those of patients with PKC-θ overexpression (Supplementary Fig. S9A and S9B). No significant differences in disease-free survival (P = 0.218) and overall survival (P = 0.841) were observed between GIST patients with and without PKC-δ expression (Supplementary Fig. S9A and S9B).

To further validate the clinicopathologic values of PKC-θ, we used the original 50 specimens (nine GISTs without the KIT mutation and 41 GISTs with the KIT mutation) and the additional 104 GIST tissues (24 GISTs without the KIT mutation and 80 GISTs with the KIT mutation; Supplementary Table S2). The correlation between PKC-θ expression and clinicopathologic parameters was then analyzed in a cohort of GISTs lacking the KIT mutation and a cohort of GISTs with the KIT mutation. This analysis showed that PKC-θ expression significantly correlated with mitotic index (P < 0.001 in GISTs with the KIT mutation and P = 0.02 in GISTs lacking the KIT mutation), high tumor grade (P < 0.001 in GISTs with the KIT mutation and P = 0.015 in GISTs lacking the KIT mutation), frequent recurrence/metastasis (P = 0.001 in GISTs with the KIT mutation and P = 0.017 in GISTs lacking the KIT mutation), and poor patient survival (P = 0.01 in GISTs with the KIT mutation and P = 0.052 in GISTs lacking the KIT mutation; Supplementary Table S4). Correlation between PKC-θ expression and survival was also determined via Kaplan–Meier analysis, which allowed us to confirm that PKC-θ expression was significantly correlated with disease-free survival (P < 0.001 in GISTs with the KIT mutation and P = 0.015 in GISTs lacking the KIT mutation) and overall survival (P = 0.001 in GISTs with the KIT mutation and P = 0.031 in GISTs lacking the KIT mutation; Supplementary Fig. S10A and S10B). Multivariate analysis was then performed using all 154 GISTs to determine whether PKC-θ can be used as an independent prognostic factor. We found that PKC-θ expression was strongly correlated with recurrence/metastasis (P = 0.008) and survival (P = 0.01), which indicates that PKC-θ can be used as an independent prognostic factor for GISTs (Supplementary Table S5).

Recent cancer genomic studies demonstrate that approximately 33 to 66 mutations are accumulated in cancer cells (28). Identifying the roles and interactions of these diverse gene mutations in cancer development is a difficult task, as is distinguishing between the driver gene and passenger gene mutations. However, GISTs are characterized by relatively few gene mutations; KIT, PDGFRA, and BRAF mutations are common and are mutually exclusive (29). KIT is mutated in more than 70% of GISTs, and the mechanism of its pathogenesis has been well characterized (30, 31). Therefore, GIST is an ideal model for demonstrating how the sequence of single gene mutation contributes to the development of a specific type of tumor (32, 33).

Most studies of mutant KIT demonstrate that the overexpression of MT-KITs is a generalized phenomenon, especially in GISTs. However, no studies have been performed to determine the stability of MT-KIT proteins or to elucidate a mechanism of MT-KIT overexpression. Because it is well known that mutations directly affect the stability of the proteins, it is important to elucidate whether the overexpression of MT-KIT results from the increased stability of mutant proteins or through the assistance of cofactors that are specifically expressed in GISTs (34, 35).

KIT is a receptor tyrosine kinase that is typically distributed on the plasma membrane and activated by ligand binding. The activated proteins are moved into the cytoplasm through endocytosis, activate downstream molecules, and are eventually degraded in the cytoplasm or recycled. The recycled proteins are thought to be redistributed on the plasma membrane (36, 37). The localization and processing of the receptor protein is directly linked to the protein function. For example, mutation or cofactor-induced aberrant trafficking and localization of EGFR are related to the development of cancers (38, 39). In addition, various quality control mechanisms exist to remove these mutant proteins (39–43). Therefore, determining the localization and a mechanism for the stabilization of MT-KITs according to the mutation type is crucial. In case of WT-KIT overexpression in several cancers, including colon cancer, the activation mechanism after SCF binding is well known. Activated KITs in turn activate the PI3-K and MAPK pathways, which promote cell proliferation and antiapoptosis. We previously demonstrated that a sustained WT-KIT activation loop was established by PKC-δ-mediated recycling, which promoted colon cancer development and progression (7).

Although the intracellular localization of MT-KIT has been reported, the specific localization according to mutation type, the mechanism for MT-KIT overexpression, and the processes by which MT-KIT proteins are trafficked are unknown. Here, we evaluated the stability of MT-KITs by generating expression vectors for four different MT-KITs, and found that MT-KIT is much less stable than WT-KIT. Thus, we hypothesized that the overexpression of MT-KIT observed in GISTs might be regulated by an unknown factor, potentially a PKC isoform. In particular, PKC-θ is widely expressed in GISTs, and is used as an immunohistochemical diagnostic marker, and is coexpressed with MT-KIT in several GIST cell lines (44). In our study, PKC-θ was strongly and exclusively expressed in GIST cell lines, and bound to and colocalized with MT-KIT. This interaction was directly linked to the stabilization of intracellular MT-KIT and resulted in its sustained activation. Because MT-KITs are rapidly degraded by the proteasome pathway and colocalized with the ER, we believe that PKC-θ supports MT-KITs in bypassing the ERQC system.

It should be noted that the abnormal localization of MT-KITs in GISTs is directly linked to their instability and function. MT-KITs expressed in HeLa cells by transfection of MT-KIT vectors showed characteristic intracellular localizations, which suggests that the abnormal localization of MT-KITs is required for their oncogenic functions. In the four GIST cell lines used in this study, the endogenous MT-KITs were colocalized with both PKC-θ and the Golgi. In the HeLa cells transfected with KIT expression vectors, the colocalization of synthetic MT-KITs with the Golgi was clearly detected when coexpressed with PKC-θ. Moreover, downregulation of PKC-θ by siRNA treatment induced dramatic reduction of MT-KIT expression and disturbed the characteristic colocalization of MT-KITs with the Golgi in GIST cells. These findings indicate that PKC-θ is involved in the trafficking of MT-KIT proteins and induces their stabilization in the Golgi. Although the retention of mutant proteins in the Golgi complex does not occur as commonly as retention in the ER, studies have shown the significance of retention of mutant protein in the Golgi complex, especially in terms of regulating the trafficking speed of mutant proteins and delivering signals (45–47). A recent review has emphasized that signaling at the Golgi mediated by key molecules such as Ras/MAPK and PI(4)P effectors is involved in various biological functions (48). Moreover, it has also been reported that ceramide-induced apoptosis involves translocation and activation of PKC-δ at the Golgi complex, which directly implicates PKC proteins in Golgi-mediated signaling (49, 50). Indeed, we demonstrated that PKC-θ induced an increase in Golgi-retained MT-KITs that were highly phosphorylated, indicating that Golgi-retained MT-KITs might be largely responsible for the tumorigenesis of GISTs.

Combining our results from transfection assays, GIST cells, and tissues, we herein propose that PKC-θ mediates Golgi stabilization of MT-KITs that contribute to GIST progression, which provides a rationale for anti-PKC-θ therapy in GISTs with MT-KIT overexpression (Supplementary Fig. S11).

No potential conflicts of interest were disclosed.

Conception and design: W.K. Kim, H. Kim

Development of methodology: W.K. Kim, J. Kim

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W.K. Kim, S. Yun, C.K. Park, S. Bauer

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W.K. Kim, C.K. Park, S. Bauer, J. Kim, M.G. Lee, H. Kim

Writing, review, and/or revision of the manuscript: W.K. Kim, M.G. Lee, H. Kim

Study supervision: H. Kim

This research was supported by the Mid-career Researcher Program (2015R1A2A2A01004835) through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology.

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.