Inhibitors of Deacetylases Suppress Oncogenic KIT Signaling, Acetylate HSP90, and Induce Apoptosis in Gastrointestinal Stromal Tumors

Activating mutations of KIT or platelet-derived growth factor receptor A (PDGFRA) represent the key oncogenic events in gastrointestinal stromal tumor (GIST; refs. 1, 2). The imatinib mesylate (IM) small molecule inhibitor of KIT and PDGFRA yields long-lasting responses in the majority of patients, with a median survival of almost 5 years (3). However, 20% of patients have primary resistance to IM, and most responding patients eventually develop secondary resistance and progress. Second line therapy with sunitinib yields 7% responses, but patients progress after a median of 6 months with a dismal outcome at time of progression. Therefore, alternative treatment strategies are urgently needed (4).

Secondary mutations within the split kinase domain (exon 13 and 17) of KIT account for acquired IM resistance in most patients (5, 6). Although mutations within the ATP-binding pocket (exon 13, exon 14) are generally sensitive to novel direct KIT inhibitors such as sunitinib and nilotinib, mutations within the activation loop (exon 17) are often crossresistant to these newer generation KIT inhibitors (7, 8). Overall, sensitivity to KIT kinase inhibitors varies considerably depending on the type of secondary mutation. Therefore individual KIT kinase inhibitor drugs are not likely to inhibit the myriad KIT drug resistance mutations that can be shown, even within the same patient (6). To avoid the pharmacologic challenge of genomic heterogeneity, novel therapeutic strategies include inhibition of heat shock protein 90, which has been shown to inhibit oncogenic KIT, irrespective of the underlying secondary mutation (9).

Histone deacetylase inhibitors (HDACI) represent a promising novel class of anticancer agents, several of which are currently undergoing clinical trials in malignant diseases. Acetylation of lysine residues of core histone proteins leads to relaxed chromatin structure and enables transcription (10). HDACI exhibit an apparent selectivity for tumor repressing or cell cycle inhibitory genes (e.g., gelsolin, CDKN1A/p21^WAF1, or CDKN2A/p16; ref. 11). In addition, a plethora of nonhistone proteins that play a key role in oncogenesis and cancer progression are targets of acetylation and deacetylation (e.g., p53, HSP90; ref. 12), and HDACI may exert their activity through these proteins as well (13).

Suberoyl hydroxamic acid (SAHA) is the first HDACI that was approved by the Food and Drug Administration, and is now included in the treatment of recurrent and advanced cutaneous T-cell lymphoma (14). Notably, SAHA has been shown to inhibit IM-resistant BCR/ABL and induce apoptosis in BCR/ABL-positive leukemia cell lines (15).

Against this background, we sought to evaluate the therapeutic relevance of HDACI in GISTs in vitro.

Cell lines. GIST-T1 was established from a human, untreated, metastatic GIST containing a 57-bp deletion in exon 11 (16). GIST882, as previously described, was established from an untreated human GIST with an homozygous missense mutation in KIT exon 13, encoding a K642E mutant KIT oncoprotein (17). GIST48 was established from a GIST that had progressed, after initial clinical response, during IM therapy. GIST48 has a primary, homozygous exon 11 missense mutation (V560D) and a heterozygous secondary exon 17 (kinase activation loop) mutation (D820A). GIST48B is a subline of GIST48, which, despite retaining the activating KIT mutation in all cells, expresses KIT transcript (data not shown) and protein at essentially undetectable levels. GIST62 was also derived from a KIT-positive GIST with a KIT exon 11 in-frame mutation that has subsequently lost KIT expression (9). GIST522 was established from an IM-resistant, progressing GIST and contains a primary heterozygous KIT exon 11 mutation (delEVQWK554-558). However, the KIT oncoprotein has been transcriptionally silenced. The KIT-negative cell lines serve as a negative control for evaluation of KIT inhibitors, but in a GIST cell context.

Leiomyosarcoma cell lines LMS03 and SK-LMS-1 were established from high-grade leiomyosarcomas, LPS141 from a dedifferentiated liposarcoma, and RMS176 from an embryonal rhabdomyosarcoma.

Reagents and antibodies. IM and LBH589 were kindly provided by Novartis Pharma. SAHA was from Alexis, and Trichostatin A (TSA) and valproic acid (VPA) were purchased from Calbiochem (Merck). A rabbit polyclonal antibody to KIT was from DAKO. Polyclonal rabbit antibodies to phospho-KIT Y703 and poly(ADP-ribose) polymerase (PARP) were from Zymed Laboratories. Polyclonal rabbit antibodies to total p42/44 mitogen-activated protein kinase (MAPK), phospho-p44/42 MAPK T202/Y204, phospho-AKT S473, total AKT, phospho-PKC𝛉 (Thr538), cleaved caspase 3, phospho-KIT Y719, acetylated-lysine, and Histone H3 were from Cell Signaling. Polyclonal goat antibody for total PKC𝛉, HSP70 monoclonal mouse antibody, and polyclonal rabbit antibody against HDAC6 were from Santa Cruz Biotechnology. p21 and β-actin antibodies were purchased from Sigma. Acetylated Histone H3 antibody was from Upstate Biotechnology. A HSP90 antibody was from Stressgen.

In vitro assays. Viability studies were carried out using the CellTiter-Glo luminescent assay (Promega), in which the luciferase-catalyzed luciferin/ATP reaction provides an indicator of cell number (18). For these studies, the cell lines were plated at 15 to 30,000 cells per well in a 96-well flat-bottomed plate (Falcon), cultured in serum-containing media for 1 to 2 d, and then incubated for 72 h with HDAC, or KIT inhibitors or DMSO-only solvent control. The CellTiter-Glo assay luminescence was measured with a Genion Luminometer (Tecan) and the data were normalized to the DMSO-only control group. All experimental points were measured in quadruplicate wells for each plate and were replicated in at least two plates.

Apoptosis studies were done by measuring caspase 3 and caspase 7 activation with the Caspase-Glo 3/7 Assay kit (Promega). This assay uses a proluminescent substrate containing the DEVD sequence recognized and activated by caspase 3 and caspase 7 (19, 20) and the luminescence signal is proportional to net caspase 3 and caspase 7 activation (21). The experimental conditions were all as described above for the CellTiter-Glo studies except treatment duration of 24 and 48 h.

Western blotting and immunoprecipitation. Protein lysates were prepared from cell line monolayers according to standard protocols (22). Protein concentrations were determined with the Bio-Rad Protein Assay (Bio-Rad Laboratories). Electrophoresis and immunoblotting were carried out as previously described (23). Changes in protein expression and phosphorylation as visualized by chemiluminescence were captured and quantified using a FUJI LAS3000 system with Science Lab 2001 ImageGauge 4.0 software (Fujifilm Medial Systems).

Immunoprecipitations were performed with Sepharose protein G beads (Zymed Laboratories) from 400 to 600 μg of total protein as described previously (9).

Cell cycle analysis. Cells were plated in six-well plates, grown until 80% confluence, and then treated for 48 h with DMSO, SAHA (2 μmol/L), or IM (500 nmol/L) and a combination of both drugs. Cells were then trypsinized and stained with DNA prep containing propidium iodide (Beckman Dickonson) followed immediately by flow cytometric analysis (BeckmanCoulter FC500 Flow Cytometer). Modfit LT software 3.1 (Verity Software House) was used for data analysis.

Cell viability assessment by Annexin V staining. After drug treatments for 48 h, cells were resuspended in 500 μL of the staining buffer and Annexin V FITC (525 nm) and 7-AAD (675 nm) were added. After incubation at room temperature for 15 min, Annexin V–positive cells were estimated by flow cytometry. Ten thousand events of each sample were acquired on a BeckmanCoulter FC500 Flow Cytometer. Doublet discrimination was done with FL2 versus FL2 peak histogram.

Quantitative real-time reverse transcriptase-PCR. Cells were plated in 12-well plates, grown until 80% confluence, and then treated with drug-containing media. Cells were then collected with RNAprotect Cell reagent (Quiagen) and processed with RNeasy Mini Kit (Quiagen) according to the manufacturer'sf protocol, for RNA isolation. cDNA was transcribed by reverse Transcriptase-PCR using random primers and KIT-cDNA was amplified and measured using TaqMan gene expression assays (Applied Biosystems) and real-time PCR system LightCycler 480 (Roche). Beta Actin cDNA [also measured using Taqman gene expression assay (AB)] served as reference gene for relative quantification.

HDACI exhibit antiproliferative effects in KIT-positive cell lines. We first evaluated the antiproliferative effects of SAHA in GIST cell lines by treating cells for 72 hours with increasing concentrations of SAHA (10 nmol/L–10 μmol/L), as well as with 500 nmol/L of IM alone and in combination with 2.5 μmol/L of SAHA.

SAHA treatment resulted in strong antiproliferative effects in IM-sensitive GIST882 and GIST-T1 (IC₅₀, 3.5 and 1.7 μmol/L, respectively) and IM-resistant GIST48 (IC₅₀, 3.2 μmol/L) but not in KIT-negative GIST48B, GIST62, and GIST522 (IC₅₀, >10 μmol/L; Fig. 1A). Combination of SAHA with IM (500 nmol/L) yielded slightly additive effects (Fig. 1B) in KIT-positive cell lines but not in KIT-negative GIST48B. For combination studies, a clinically relevant dose of IM that substantially inhibits KIT signaling in IM-sensitive but not IM-resistant GIST was used.

Sarcomas other than GIST are less sensitive to HDACI. To study GIST-independent effects of HDACI, we then compared SAHA antiproliferative effects in GIST to those induced in several other soft tissue sarcoma cell lines (Fig. 1C). IC₅₀s ranged from 2.8 μmol/L in the rhabdomyosarcoma cell line RMS176 to 5 μmol/L in the leiomyosarcoma cell line SK-LMS-1 but were not reached in LMS03 and LPS141 (inhibition of 35% and 24% at 10 μmol/L).

HDAC-inhibition induces apoptosis in KIT-positive but not in KIT-negative GIST, which is enhanced by imatinib. To study induction of apoptosis by HDACI, cells were plated in 96-well plates and treated with SAHA at concentrations ranging from 1 to 5 μmol/L singly or in combination with 500 nmol/L of IM for 24 and 48 hours and then analyzed for caspase 3/7 activation. In GIST882, SAHA alone resulted in a maximum of 5.9-fold induction, compared with control, and IM treatment showed a 3.7-fold induction. Notably, combination of 5 μmol/L of SAHA and 500 nmol/L of IM after 48 hours resulted in a 15-fold induction of activated caspase 3/7, compared with vehicle control.

In IM-resistant GIST48, additive effects of combination treatment with 500 nmol/L of IM and 5 μmol/L of SAHA (13-fold) after 24 hours were less pronounced than those seen in GIST882, compared with IM and SAHA alone (5-fold and 11-fold increase, respectively; Fig. 2A). Notably, combinations of imatinib with lower doses of SAHA (1 and 2 μmol/L) resulted in synergistic proapoptotic effects.

We further evaluated apoptosis in cell lysates from GIST cells that were treated with SAHA for 18 hours. SAHA at concentrations of ≥2 μmol/L resulted in dose-dependent cleavage of caspase 3 and the caspase substrate PARP (86-kDa fragment) when measured by Western blot in GIST882, GIST-T1, and GIST48 (Fig. 2B). Combination of 500 nmol/L of IM and SAHA at 2 μmol/L induced caspase 3 cleavage comparable with that induced by 10 μmol/L of SAHA alone. GIST48 caspase cleavage was demonstrable at 2 μmol/L of SAHA, with a 2-fold induction at 5 μmol/L and a synergistic effect when 2 μmol/L of SAHA were combined with IM, with the combination exceeding cleavage levels seen with 10 μmol/L of SAHA alone. In contrast, no induction of apoptosis markers was seen in KIT-negative GIST48B (Fig. 2B). Representative apoptosis studies were also performed with LBH589 in GIST882, which at 500 nmol/L showed a 7.4-fold induction of caspase 3 cleavage and 7.8-fold induction of PARP cleavage. Again, synergistic induction was shown when LBH589 was combined with IM (Supplementary Fig. S1).

We further performed apoptosis assessment by Annexin V/7-AAD staining in GIST882. Forty-eight hours of treatment with IM (500 nmol/L), SAHA (2 μmol/L), and LBH589 (25 nmol/L) resulted in moderate induction of early apoptosis (15%, 7%, and 20%, respectively) when given singly (Supplementary Fig. S2). In contrast, combinations of IM with SAHA or LBH589 at the same dose levels showed a synergistic proapoptotic effect resulting in early apoptosis in 38% (IM plus SAHA) and 49% (IM plus LBH589) of GIST 882 cells.

For cell cycle analyses, GIST882 cells were treated for 48 hours with DMSO, IM (500 nmol/L), SAHA (2 μmol/L), and LBH589 (25 nmol/L), which resulted a low sub-G₁ fraction (1%, 13%, 2%, and 5%), whereas combinations of both SAHA and LBH589 resulted in additive increase of sub-G₁ cells (29% and 36%). With regard to effects of inhibitors on cell cycle distribution, IM showed a strong reduction of the S phase, whereas other inhibitors did not. No other substantial changes were observed (Supplementary Fig. S3).

HDACI inhibit KIT- and KIT-dependent signaling pathways. To elucidate the mechanism of HDACI-induced sensitization to apoptosis, we investigated the effects of HDACIs on KIT and KIT-depending signaling pathways. To avoid confounding effects of cell apoptosis and general disruption of the cellular architecture, an early time point (18 hours) was chosen as used in similar experiments by other groups (13). In GIST882, GIST-T1, and GIST48, SAHA treatment resulted in marked inhibition of KIT-phosphorylation (GIST882 and GIST-T1: IC₅₀, 1.7 μmol/L, near total inhibition at 5 μmol/L; GIST48: IC₅₀, 3 μmol/L, 81% inhibition at 5 μmol/L), which was paralleled by substantial inhibition of phospho-AKT and phospho-MAPK (Fig. 3A), and by decreased total KIT expression (70% and 80% decrease at 5 μmol/L in GIST-T1, GIST882, and GIST48) involving equally the mature (160 kDa) and immature (145 kDa) forms of KIT. LBH589 showed a similar, dose-dependent effect on KIT and KIT-depending signaling at 20-fold lower concentrations (IC₅₀ for inhibition of p-KIT, 105 nmol/L; total KIT, 190 nmol/L; pAKT, 30 nmol/L; pMAPK, 190 nmol/L; Supplementary Fig. S1).

In KIT-negative GIST48B, no detectable levels of phosphorylated KIT and minimal levels of total KIT were found. SAHA treatment caused a minor decrease in phospho-MAPK starting at 2 μmol/L, with partial inhibition at 5 μmol/L (Fig. 3A).

Time course studies on GIST882 cells were performed to determine the onset of KIT inhibition in correlation to histone acetylation. Cells were treated for 30 minutes to 24 hours with 5 μmol/L SAHA (Fig. 3B). Immunoblots for KIT and for KIT-depending signaling pathways revealed near total inhibition of phospho-KIT, total KIT, pAKT, and pMAPK after 9 hours. Caspase cleavage was shown at 12 hours and peaked at 24 hours. Induction of cell cycle protein p21 started at 3 hours of treatment. Notably, acetylation of core histone protein H3 was observed at the 12-hour time point and peaked at 24 hours.

Effects of a HDACI panel on GIST viability. We then evaluated other HDACI, including TSA, VPA, sodium butyrate (NaB), and LBH589, which showed IC₅₀s between high nmol/L (LBH589, TSA) to low mmol/L (NaB, VPA) concentrations, with only minor differences seen between IM-sensitive and IM-resistant cells (Fig. 4A and B). Additive antiproliferative effects were seen with all HDACI and IM in GIST48 (Fig. 4C). Notably, LBH589, a third-generation pan-deacetylase inhibitor, was 22 and 39 times more potent than SAHA against GIST48 and GIST882 respectively (Fig. 4A and B).

Effects of HDACI on histone H3 acetylation. The lowest SAHA concentrations inducing histone H3 acetylation were 1 μmol/L in GIST882, 0.5 μmol/L in GIST-T1, GIST48, and GIST48B (Fig. 2B), and for LBH589 were 10 nmol/L in GIST882 and GIST48 (Supplementary Fig. S1). Notably, minor induction of total H3 was observed. Of note, combination treatment of HDACI (SAHA and LBH589) and IM resulted in synergistic induction of histone acetylation (Supplementary Fig. S1; Fig. 2B). Although LBH589 at 100 nmol/L resulted in 5.5-fold induction of histone acetylation, the combination with IM resulted in 15-fold induction (Supplementary Fig. S1).

GIST histone H3 acetylation and KIT inhibition differ after treatment with various HDACI. The effects of several HDACI were compared on pKIT, pAKT, acetylated, and total histone H3 (Fig. 4D) after 18 hours of treatment, using two concentrations for each HDACI, around the IC₅₀s established previously in the GIST882 proliferation studies. Inhibition of pKIT was by far strongest for cells treated with SAHA, whereas other HDACI showed only minor pKIT inhibition. Notably, KIT-inhibition differed between SAHA 2 μmol/L, TSA 300 nmol/L, VPA 3 mmol/L, and NaB at 1 mmol/L despite comparable levels of histone acetylation. Interestingly, moderate inhibition of pKIT was accompanied by complete inhibition of pAKT following treatments with NaB and VPA (Fig. 4D).

SAHA decreases KIT-RNA in a time and dose-dependent manner. To evaluate the effect of HDAC inhibition on KIT transcript levels, real-time PCR was performed following treatment of GIST882 with SAHA 2 and 5 μmol/L for 4, 8, and 12 hours (Fig. 5A). KIT transcripts decreased in a time and dose-dependent manner in GIST882, with a decrease of 70% using SAHA 2 μmol/L and 84% using SAHA 5 μmol/L, after 12 hours of treatment (Fig. 5A). By contrast, GIST882 treatment with IM at this time showed no reduction compared with DMSO. We then compared SAHA effects at 12 hours with LBH589 and 17-AAG and found a similarly strong decrease of KIT mRNA levels following LBH589, but not after 17-AAG, in GIST 882 and GIST48 (Fig. 5B). The KIT-negative liposarcoma cell line LPS141 showed 10⁻⁴ times lower levels of KIT mRNA levels compared with the untreated GIST cell lines (Fig. 5B).

Given the important role of PKC𝛉 in the regulation of KIT expression (24), we also investigated the effects of HDACI on PKC𝛉 expression and phosphorylation (Supplementary Fig. S4). Notably, SAHA and LBH589 were shown to reduce expression of both total and phosphorylated PKC𝛉 by 20% to 40%.

To evaluate the half-life of KIT following block of translation, we treated GIST882 with 100 μmol/L cycloheximide in a time course experiment and found >80% reduction of total KIT after 3 hours and near-total loss after 6 hours (Fig. 5C).

HDACI inhibit KIT through acetylation of HSP90. To investigate possible nonhistone-related KIT inhibitory mechanisms, we treated GIST882 with SAHA 5 μmol/L for 12 hours, then performed immunoprecipitation for HSP90, and counterstained with antiacetyl lysine and HDAC6. Substantial increase of HSP90 acetylation was shown following SAHA treatment (Fig. 6), whereas HDAC6 dissociated from HSP90 (Fig. 6A). These findings were confirmed by immunoprecipitation of HDAC6, where we showed commensurate dissociation of HSP90 (Fig. 6B). KIT immunoprecipitations revealed increased binding of HSP70 to KIT, whereas HSP90 was found to slightly dissociate when normalized to relative levels of KIT (Fig. 6C).

Identification of KIT as a crucial oncogenic regulator pathway has revolutionized the treatment of GIST (25). The KIT inhibitor imatinib has tripled the median survival of patients with metastatic GIST and many patients live 5 years or longer with the disease. However, most patients inevitably develop resistance. Resistance is mostly conferred by secondary mutations within the split kinase domain of KIT (5, 6, 26). Although resistance mutations within the ATP-binding domain (T670I, V654A) can successfully be targeted by alternative KIT inhibitors such as sunitinib, mutations within the activation loop are an ongoing pharmacologic challenge (28–30). In addition, less than nine different resistance mutations have already been described and different mutations may even occur within the same patient (5, 6, 30). Direct, ATP-competitive inhibitors are therefore unlikely to sufficiently inhibit all possible KIT mutations.

Alternative strategies aim to inhibit the oncogenic signal of KIT regardless of existing secondary mutations. Among those are inhibitions of KIT-dependent signaling pathways (e.g., PI3K, AKT, or mammalian target of rapamycin; refs. 31, 32) or indirect inhibition of KIT using HSP90 inhibitors (9, 33, 34) that are already being tested clinically.

The posttranslational modification of histones, e.g., through acetylation and deacetylation of lysine tails, has been shown to be an important mechanism of transcriptional regulation (35). Interestingly, many genes up-regulated by HATs are important for differentiation, cell cycle control, and apoptosis (36). Aberrant acetylation, either through overexpression of HDAC or HAT dysfunction is commonly found in epithelial and hematologic cancers (37–39). In this context, HDACIs exhibit an apparent selectivity for activating transcription of tumor-suppressing genes (36).

Recently, several groups have highlighted the potential therapeutic relevance of HDACIs in IM-resistant chronic myelogenous leukemia cells. HDACI destabilized the BCR-ABL oncoprotein even in the presence of the highly IM-resistant T315I gatekeeper mutation (13, 40, 41). Given the similarities of the oncogenic tyrosine kinase mechanisms between CML and GIST, we therefore investigated the potential therapeutic value of HDACI in GIST.

The data reported herein are the first to show strong antiproliferative and proapoptotic effects in both IM-sensitive and IM-resistant GIST by inhibitors of histone deacetylases. We show that these effects are conferred by inactivation of the KIT-oncogenic signaling cascade. The inactivation mechanisms include both inhibition of KIT transcript expression and also degradation of KIT oncoproteins via suppression of deacetylation of the KIT chaperone HSP90.

KIT-positive GIST lines were sensitive to HDACI (IC₅₀, 1.7–3.5 μmol/L), whereas KIT-negative GIST lines were unresponsive (Fig. 1). These findings compare well with data from other SAHA-sensitive cancer cell lines, for which IC₅₀s (SAHA) in the low micromolar dose range were described (15, 40, 42). These dose levels are considered to be therapeutically achievable. Antiproliferative effects of HDACI were lower in non-GIST sarcomas, with only 10% to 20% inhibition after treating leiomyosarcoma (LMS03) and liposarcoma (LPS141) cell lines with 10 μmol/L SAHA indicating that GISTs are uniquely susceptible to HDACI.

In KIT-positive GISTs, HDACI treatment induced dose-dependent inhibition of KIT phosphorylation, paralleled by AKT inhibition (Fig. 3A). In GIST882, inhibition of AKT-phosphorylation was even more pronounced with all HDACI used than inhibition of KIT phosphorylation (Supplementary Fig. S1; Figs. 3 and 4), suggesting that HDACI may have inhibitory effects on the PI3K-AKT pathway independent of KIT inhibition. Interestingly, in studies with different HDACI, the effects on histone acetylation did not strictly correlate with KIT inhibition, suggesting that KIT inhibitory effects are partly histone independent and perhaps dependent on certain HDAC subtypes. Although TSA, SAHA, and LBH589 are pan HDACI, VPA and NaB only inhibit class I HDACs.

Our studies suggest that HDACI have several mechanisms contributing to KIT inhibition: Histone acetylation may cause reduced KIT transcription, most likely by affecting expression of KIT transcriptional repressors. HDACI might also regulate oncoprotein function, e.g., by modifying KIT acetylation, and may influence KIT oncogenic signaling through effects on nonhistone HDACI targets.

Direct, ATP-competitive inhibitors of KIT, such as IM, are known to have KIT inhibitory effects within several minutes of treatment (data not shown). Time course studies with HDACI (Fig. 3B) showed onset of KIT inhibition after 3 and 6 hours of treatment, with a maximum effect between 12 and 18 hours. Only subtle histone acetylation was seen at the early time points (3–6 h) and the maximal acetylation level was seen only at the last time point (24 hours). These findings suggest the possibility of HDACI-mediated indirect inhibition of KIT rather than a direct, biochemical inhibition as seen with IM. Notably, signaling studies showed both inhibition of KIT phosphorylation and decreased KIT expression. Although direct KIT inhibitors such as IM and sunitinib do not decrease total KIT expression, we have previously shown that HSP90 inhibitors strongly decrease KIT expression levels (9).

HSP90 is subject to several posttranslational modifications that affect its function. Hyperacetylation of HSP90, especially acetylation of the amino acid K294, decreases the affinity for most clients and certain cochaperones (43, 44). Inhibition of HDAC6, a class II HDAC, has recently been shown to disrupt the chaperone function of HSP90 resulting in degradation of HSP90 clients, such as BCR-ABL (12, 13). This seems to be caused by lower affinity for ATP (40). As the assembly of the functional HSP90 chaperone complex requires ATP, HDACI effects on HSP90 can be explained in part by the inhibition of its complex formation. In line with this model, we here show that HDACI treatment of GIST cells causes dissociation of HDAC6 from HSP90 resulting in hyperacetylation of HSP90 and a consequent loss of KIT (Fig. 6).

Treatment of GIST48B and GIST62 with SAHA resulted in substantial inhibition of pMAPK (GIST48B and GIST62) and pAKT (GIST62) despite the lack of measurable KIT activation (data not shown; Fig. 3A). These findings indicate that apart from HDACI effects on KIT, other signaling pathways might be affected.

Of note, KIT transcript levels were found to be down-regulated in a time- and dose-dependent fashion, with maximal effects seen after 12 hours of SAHA treatment (Fig. 3B). Because HDACI are thought to enable rather than suppress transcription, these effects can be explained by induction of a gene responsible for KIT transcriptional repression. The loss of KIT expression seen after 12 hours of treatment with HDACI could be explained both by transcriptional repression and HSP90 inhibition. However, the extent of KIT inhibition due to a transcriptional block at this time point would mainly depend on the half-life of mature KIT in GIST cells. We therefore treated GIST882 with cycloheximide and showed that the KIT half-life, in GIST, is <3 hours (Fig. 5C). In contrast, the HSP90 chaperone (>24 hours) or the KIT regulator PKC𝛉 (>12 hours) exhibited a more than four times longer half-life following cycloheximide treatment (Fig. 5C). Ou and colleagues (24) recently reported that PKC𝛉 is a major regulator of KIT transcription. Notably, HDACI caused partial inhibition of PKC𝛉, and PKC𝛉 inhibition may therefore contribute to the inhibition of KIT expression in GIST.

Mechanisms of resistance to direct KIT inhibitors largely depend on secondary mutations within the KIT ATP binding pocket or activation loop. Our studies suggest that HDACI-mediated inhibition of KIT oncoproteins is independent of the KIT mutation location, with equal relevance for GISTs with imatinib-sensitive or imatinib-resistant KIT mutations. Hence, HDACI might overcome resistance to both ATP binding pocket and activation loop imatinib-resistance mutations, providing a new strategy to inhibit KIT.

Similar to studies evaluating combinations of IM and HDACI in CML, we found no evidence for antagonistic effects and we identified additive proapoptotic effects in GIST882 and GIST48 (15, 40, 45). Notably we observed synergistic effects of HDACI and IM on histone acetylation in the KIT-positive GIST lines (Fig. 3A) but not in the KIT-negative GIST48B.

We observed induction of p21 by SAHA in all GIST cell lines (Fig. 2B), including KIT-negative GIST48B and GIST62 (data not shown; Fig. 2B), and this might result directly from histone acetylation (46). However, in the KIT-negative GIST48B and GIST62, p21 induction did not cause marked cell cycle arrest, nor induction of apoptosis (data not shown; Supplementary Figs. S3 and S4; Fig. 2B). These findings suggest that inhibition of oncogenic KIT is the most consequential mechanism of HDACI action in GIST.

Taken together, our data show that HDACI have disease-specific effects in GIST by inhibiting the crucial KIT oncogenic pathway. We consistently found additive effects of HDACI and IM in IM-sensitive GIST, regardless of the HDACI used. LBH589 exhibits the highest potency of all HDACI tested, although biological effects of SAHA and LBH589 are similar. Possible mechanisms of action for HDACI in GIST include acetylation of HSP90 with consequent destabilization of KIT, but effects on KIT transcriptional activity may also play a partial role. Given the antiproliferative and proapoptotic effects of HDACI at doses that can be achieved therapeutically, our data provide a strong rationale for the clinical evaluation of HDACI/DACI in GIST.

S. Bauer: Consultant/advisory board, GIST. The other authors disclosed no potential conflicts of interest.

Grant support: Anonymous donor, Max-Eder Fellowship from the Dr. Mildred Scheel Stiftung für Krebsforschung (S. Bauer), and Life-raft Group Research Initiative (S. Bauer).

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.