Bromodomain and extraterminal domain inhibitor enhances the antitumor effect of imatinib in gastrointestinal stromal tumours

Abstract In gastrointestinal stromal tumours (GISTs), the function of bromodomain‐containing 4 (BRD4) remains underexplored. BRD4 mRNA abundance was quantified in GISTs. In the current study, we investigated the role of BRD4 in GISTs. Our results show a significant enhancement in BRD4 mRNA and a shift from very low‐risk/low‐risk to high‐risk levels as per NCCN specifications. Overexpression of BRD4 correlated with unfavourable genotype, nongastric location, enhanced risk and decreased disease‐free survival, which were predicted independently. Knockout of BRD4 in vitro suppressed KIT expression, which led to inactivation of the KIT/PI3K/AKT/mTOR pathway, impeded migration and cell growth and made the resistant GIST cells sensitive to imatinib. The expression of KIT was repressed by a BRD4 inhibitor JQ1, which also induced myristoylated‐AKT‐suppressible caspases 3 and 9 activities, induced LC3‐II, exhibited dose‐dependent therapeutic synergy with imatinib and attenuated the activation of the PI3K/AKT/mTOR pathway. In comparison with their single therapy, the combination of JQ1/imatinib more efficiently suppressed the growth of xenografts and exhibited a reduction in KIT phosphorylation, a decrease in Ki‐67 and in the levels of phosphorylated PI3K/AKT/mTOR and enhanced TUNEL staining. Thus, we characterized the biological, prognostic and therapeutic implications of overexpressed BRD4 in GIST and observed that JQ1 suppresses KIT transactivation and nullifies the activation of PI3K/AKT/mTOR, providing a potential strategy for treating imatinib‐resistant GIST through dual blockade of KIT and BRD4.

neurofibromatosis type 1 (NF1), BRAF, or succinate dehydrogenase (SDH) genes. 3,4 Both the NIH (Bethesda, MD) and National Comprehensive Cancer Network (NCCN) schemes are deemed effective for risk stratification in GIST, whereas their prognostic utility is not uniformly validated, and the NCCN scheme lacks sufficient evidence-based data for some uncommon settings. 5,6 Another challenge in the current care of GIST patients is the inevitable resistance to imatinib that emerges following initial responses due to acquired secondary mutations. 7,8 Therefore, it is desirable to characterize aberrations in nonkinase biochemical pathways that may crosstalk with or regulate RTK-driven signalling to overcome the current limitations in prognosis and drug resistance, particularly for cases refractory to imatinib. 9,10 In recent decades, large breakthroughs have led to functional evaluation, epigenetic profiling and targeted therapy in human cancer. 11 Cancer cells characteristically exhibit aberrant epigenetic regulation and use the regulatory components of the chromatin to carry out transcriptional programmes initiating oncogenesis. 12,13 There are two particularly important epigenetic readers, bromodomain and extraterminal domain (BET) proteins, that have two characteristic tandem bromodomains (BD1 and BD2) and primarily recognize the acetylated lysine of histones H3 and H4, and occasionally acetylated nonhistone proteins, to modulate gene expression. 11,14,15 A well-studied BET family member, BRD4, usually recruits the P-TEFb transcription elongation factor or other transcriptional factors or chooses modifiers of histones to promote activation of target genes at the transcriptional stage. 16,17 BRD4 is crucially involved in several cellular processes, including the progression of the cell cycle, cell growth control, apoptosis and tumour initiation. 18 BRD4 is often overexpressed and clinically associated with various human cancers via its elevation of the expression and enhancement of the oncogenic functions of major proteins in cancer, such as aldehyde dehydrogenases (ALDH) in ovarian cancer, androgen receptor (AR) and ETS-related gene (ERG) in prostate cancer and c-Myc in leukaemia. 19,20 BRD4 was shown to regulate PTEN and the PI3K/AKT pathway. [21][22][23] The important roles of BRD4 in transcriptional activation and the initiation of tumour formation have led to the consideration of selective inhibition of BRD4 as a promising therapeutic strategy against cancer. 24 The amino-terminal bromodomains of BRD4 have been targeted using several selective small-molecule inhibitors, such as I-BET762 and JQ1, and promising antiproliferative effects have been observed along with promising clinical significance in a range of cancers. 25,26 The first-in class thienodiazepine small-molecule JQ1 binds to BET protein bromodomains with acute complementarity and an affinity in nanomoles, resulting in transient, competitive and robust displacement of BRD4 from acetylated chromatin, thus selectively occluding transactivation of specific gene sets in a context, or dose-dependent manner. 13,27 These findings established a convincing logic for targeting BRD4 in cancer therapy.
In this study, the levels of mRNA and protein expression were significantly increased in higher-risk tumours. Furthermore, protein overexpression was strongly associated with unfavourable KIT/ PDGFR/BRAF mutation types and independently predicted poor disease-free survival (DFS) in primary imatinib-naïve GISTs. In imatinib-resistant GIST cells, stable silencing of endogenous BRD4 decreased cell growth and migration and enabled resensitization to imatinib, concomitant with significant suppression of the activity and/or expression of kinases in the PTEN/PI3K/AKT/mTOR signalling cascade. Apart from inducing apoptosis and autophagy, BRD4 inhibition using JQ1 apparently inhibited total and phosphorylated KIT levels through transcriptional repression and inactivated downstream kinases. Compared with either drug alone, the therapeutic synergy of combined therapy with JQ1 and imatinib was substantiated in vitro and in vivo, with concomitant downregulation of KIT and the phosphorylation levels of proteins in the PTEN/PI3K/AKT/ mTOR pathway. Taken together, we robustly characterized BRD4 as a progression-associated metabolic driver for accurate prognostication in GIST and provided a rationale for dual blockade of KIT and BRD4 in GIST therapies.

| Cell lines
After imatinib therapy, GIST48 cells harbouring a homozygous V560D mutation acquired a heterozygous D820A mutation in KIT exon 17, and GIST430 cells harbouring a heterozygous in-frame deletion in KIT exon 11 developed a heterozygous missense mutation in exon. 28 Authentication of both cell lines was performed by shorttandem-repeat genotyping, and lines were periodically confirmed to be mycoplasma free using PlasmoTest (Invivogen) and maintained in IMDM (Invitrogen) containing FBS (10%), penicillin/streptomycin (100 U/mL) and l-glutamine (4 mM; Invitrogen) at 37°C in CO 2 (5%).

| IHC
For antigen retrieval, TMA sections were microwaved and incubated with a primary antibody against BRD4 (BD transduction), and the protein expression was detected using a ChemMate EnVision Kit from Dako. BRD4 immunoexpression was recorded as the mean percentage of labelled cells in the tumoural cytoplasm, and samples with 50% or more tumour cells whose cytoplasms stained moderately or strongly were classified as overexpressing samples.

| In vitro wound-healing assays
Wound-healing studies were carried out as described previously. 10 Briefly, using the tip of a P-100 pipeteman, slashes were made in nearly confluent cell cultures after the inhibitors were added.
Photographs of the plates were taken using an inverted Leica DMI 3000B microscope (Leica Microsystems). Experiments were performed in triplicate.

| Colony formation assay
Single-cell suspensions (10 mL) were seeded in p100 culture dishes in triplicate. After 24 hours, imatinib, JQ1, singly, a combination, or DMSO, was added to the cell culture and incubated further for 10 days. Likewise, MK-2206, an AKT inhibitor, was used as a single agent or in combination with imatinib. Finally, at the end of the experiment, the cells were fixed and stained for 20 minutes with crystal violet (0.5%) and then washed, dried and photographed.

| RNA extraction and real-time quantitative PCR
Real-time PCR was performed as previously described. 29

| Western blot assay
Western blot assay was performed as previously described. 31,32 Total cell protein was isolated using NP40 lysates in the presence

| GIST xenograft models
To generate xenografts of subcutaneous human tumours, the flanks

| Statistics
To present all quantitative data, the mean ± standard deviation (SD) was used from at least three independent experimental data points.
Prism V (GraphPad Software) was used for statistical analyses.
One-way analysis of variance (ANOVA) or an unpaired two-tailed Student's t test was applied to determine the statistical differences between two groups of data. When P < .05, the data were deemed significant.

| Overexpression of BRD4 mRNA and protein correlated with unfavourable and poor clinicopathologic outcomes
The endogenous BRD4 mRNA levels in the LCM-isolated tumour cells of 40 primary GISTs were quantified, revealing significant upregulation in higher risk categories defined by both the NCCN guidelines ( Figure 1A,B) and the NIH scheme ( Figure 1C). In regard to BRD4 mRNA levels, NCCN guidelines more effectively distinguished expression was associated with poor overall survival (OS) in GIST patients ( Figure 1C). Together, BRD4 overexpression remained an independent adverse prognostic factor, along with increased NIH risk levels and epithelioid histology. Moreover, the wound-healing assay exhibited significantly fewer migrating cells when BRD4 was depleted, indicating that BRD4 has a promigratory effect ( Figure 2C). Accumulating evidence has revealed that the PI3K/AKT/mTOR pathway plays important roles in relaying PTEN signalling in GIST and the neoplastic lipogenesis of other cancers. Therefore, we explored the potential regulatory basis of these kinases linked to the overexpression of BRD4 in GIST cell models, demonstrating that the phosphorylated forms of PI3K, AKT and mTOR were all prominently downregulated by BRD4 knockout ( Figure 2D). In addition, the total and phosphorylation of KIT protein was significantly decreased in BRD4-KO GIST48 and GIST430 cell lines ( Figure 2E). Furthermore, KIT mRNA expression and promoter activity were found suppressed in BRD4-KO cells as measured by qRT-PCR ( Figure 2F).

| The sensitivity of imatinib-resistant GIST cells is restored by inhibition of BRD4
Next, we investigated whether BRD4 inhibition restores imatinib sensitivity in GIST. Depending on the efficiency of the BRD4-KO clones, BRD4 silencing resulted in a variable but significantly increased percentage of nonviable GIST48 and GIST430 cells after treatment with imatinib ( Figure 3A), implying that suppression of BRD4 expression may reduce imatinib resistance.
One of the downstream intracellular signalling cascades of BRD4/PTEN signalling is the AKT/mTOR pathway. Thus, we investigated whether PI3K/AKT pathway inhibition leads to re-sensitization of imatinib-resistant GIST to imatinib. For this, GIST48 cells were treated with MK-2206 (a potent PI3K/AKT inhibitor) as a single agent or in combination with imatinib for 10 days; the long-term viability of the cells was assessed, and the cells stained with crystal violet. We observed that inhibition of PI3K/AKT reversed the sensitivity of GIST48 cells to imatinib, as revealed by the significant cell growth inhibition seen in GIST48 cells when they were exposed to a combination of imatinib and MK-2206 ( Figure 3B,C). Collectively, these data indicate that the AKT/mTOR signalling pathway plays a vital role in imatinib resistance in GIST48 cells.

| JQ1 inhibited viability, exhibited synergy with imatinib and induced apoptosis and autophagy in vitro
A synthetic analogue of cerulenin, JQ1, has been shown to potently inhibit BRD4 and suppress the tumour growth of various human cancers in vitro and in vivo. 24 However, the therapeutic potential and molecular underpinnings of JQ1 remain unexplored in GIST, especially in regard to crosstalk with KIT-elicited signal transduction.
As measured by the MTS assay, JQ1 had a deleterious effect on the survival of both cell lines ( Figure 4A). Another BRD4 inhibitor, I-BET151, also dose-dependently suppressed the cell viability of both GIST cell lines, although its potency was inferior to that of JQ1 ( Figure 4B). Accordingly, we selected JQ1 for further assessment of    triggered by JQ1 ( Figure 5F), implying a potential link between AKT inactivation and JQ1-induced cytotoxicity.

| JQ1 monotherapy versus JQ1/imatinib combined therapy in vivo
Given the synergy between JQ1 and imatinib, a GIST430-derived However, these alterations in histomorphology, kinase expression and activity and proliferative and apoptotic markers were mild to moderate in the monotherapy groups receiving imatinib or JQ1 alone.

| D ISCUSS I ON
GISTs mainly possess activating oncogenic 'driver' mutations in KIT or sometimes PDGFRA. 33,34 Imatinib mesylate possibly inhibits the activity of the KIT kinase and is the first-line drug for unresectable and advanced GIST treatment, attaining at least a limited response in nearly 80% of metastatic patients. 34,35 Nevertheless, most patients develop resistance to imatinib within 2-3 years of treatment initiation, making the clinical management  25 However, the clinical application of JQ1 is limited. 25 Since some cancer cells are insensitive to JQ1 treatment, subsequently leading to treatment failure. 26,40 Therefore, new drugs or models need to be identified to overcome the obstacles associated with JQ1 treatment.
In this study, GIST tissues exhibited a positive correlation between BRD4 mRNA abundance and risk levels. This result indicated that BRD4 overexpression in GIST is, at least in part, transcriptionally regulated. The strong association between BRD4 overexpression and adverse clinicopathologic variables, such as higher risk levels, was also validated in our cohort of GISTs. Moreover, we confirmed that BRD4 overexpression not only correlated with unfavourable genotypes but also conferred an independent negative prognostic impact, with a twofold higher risk of shorter DFS. A previous study reported that JQ1 inhibited the PI3K/AKT pathway via upregulating PTEN. 41 The biological implications of overexpressed BRD4 in GIST were supported by our findings that BRD4-silenced cell lines demonstrated decreased cell growth and migration.
Although overexpressed BRD4 has emerged as a potential therapeutic target in various cancer types, knowledge about the biology of BRD4 inhibition in GIST cell and xenograft models is scant. 19 One such mechanism hypothesized to be involved in imatinib resistance Accumulating evidence indicates that apoptosis and autophagy share common regulatory elements within oncogenic pathways activated by RTKs, including the PI3K/AKT/mTOR pathway. 42,43 To date, few studies have described the autophagy-modulating role of BRD4 inhibition in tumour cells. 44 One such study reported that glioma cells treated with JQ1 exhibited increased autophagic activity. [45][46][47] In imatinib-resistant GIST cells, increased expression of LC3-II, a widely accepted autophagic marker, was inducible by JQ1, and its level was even higher upon combined treatment with JQ1 and imatinib. Although AKT-inactivation-induced apoptosis directs cells to programmed death, the significance of BRD4-inhibitioninduced autophagy in GIST requires further elucidation. Specifically, the relevant issues regarding JQ1-induced autophagy include its potential dependency on AKT, engagement of KIT expression and activation and the resultant cytoprotection to maintain energy homeostasis versus cytotoxicity to promote cellular death under nutrient deprivation. 48 A better understanding of these aspects will help decipher the possible benefits or detriments of autophagic inhibition in GIST cell growth in combined therapy against BRD4 and KIT. In

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
JFM and PDS designed the research study. JFM, PFS and ZMM performed the experiments. JFM analysed the data. JFM and PDS wrote the manuscript. All authors reviewed and approved the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of this study are available from the corresponding author upon request.