Purpose: Although tyrosine kinase inhibitors have improved survival in advanced gastrointestinal stromal tumor (GIST), complete response is rare and most patients eventually fail the first-line treatment with imatinib. Sunitinib malate is the only approved second-line therapy for patients with imatinib-resistant or imatinib-intolerant GIST. The clinical benefit of sunitinib is genotype-dependent in regards to both primary and secondary mutations, with GIST patients harboring the KITAY502-3ins exon 9 mutation being the most sensitive.

Experimental Design: As sunitinib resistance is now emerging, our goal was to investigate mechanisms of progression and to test the efficacy of novel tyrosine kinase inhibitor on these resistant mutants in vitro. N-ethyl-N-nitrosourea mutagenesis of Ba/F3 cells expressing the KITAY502-3ins mutant was used to investigate novel patterns of resistant mutations evolving in the presence of sunitinib.

Results: Tumors from patients who developed sunitinib resistance after at least 1 year of radiographic response were analyzed, showing similar findings of a primary KITAY502-3ins mutation and a secondary mutation in the KIT activation loop. Ba/F3 cells expressing these sunitinib-resistant double mutants showed sensitivity to both dasatinib and nilotinib.

Conclusions: Sunitinib resistance in GIST shares similar pathogenetic mechanisms identified in imatinib failure, with acquisition of secondary mutations in the activation domain after an extended initial response to the drug. Moreover, in vitro mutagenesis with or without N-ethyl-N-nitrosourea of Ba/F3 cells expressing KITAY502-3ins showed acquisition of secondary mutations restricted to the second kinase domain of KIT. In contrast, in vitro resistance to imatinib produces a broader spectrum of secondary mutations including mutations in both KIT kinase domains. (Clin Cancer Res 2009;15(22):686270)

Translational Relevance

Therapy with tyrosine kinase inhibitors (TKI) benefits 80 of patients with advanced gastrointestinal stromal tumor (GIST), but most patients eventually develop drug resistance. Sunitinib malate is a broad-spectrum TKI approved as second-line therapy for imatinib-resistant GIST patients. The clinical benefit of sunitinib is genotype-dependent, with GIST patients harboring a KIT exon 9 mutation being the most sensitive. As sunitinib resistance is now emerging, our goal was to investigate mechanisms of progression and to test the efficacy of novel TKI on these resistant mutants in vitro. To establish a comprehensive profile of acquired mutations that confer drug resistance to KIT exon 9 mutants, Ba/F3 cells expressing KIT502-3AYins were subjected to N-ethyl-N-nitrosourea mutagenesis and sunitinib selection. The KIT mutations identified by in vitro screen recapitulated the pattern of second-site mutations observed in TKI-resistant patients. Our findings highlight new mechanisms of resistance to second-generation TKI and provide a preclinical rationale for alternative therapeutic options for patients failing sunitinib therapy.

Although imatinib achieves a partial response or stable disease in most gastrointestinal stromal tumor (GIST) patients, with lasting responses over 5-year period in >20 of patients, complete responses are rare (1). Clinical responses to imatinib in GIST depend on the presence and type of KIT or platelet-derived growth factor receptor- (PDGFRA) gain-of-function mutations. Thus, patients with exon 11 mutations show a partial response rate of 84, whereas patients with tumors harboring a KIT exon 9 or no detectable mutation had a partial response rate of 48 and 0, respectively (2). It is now clear that a majority of patients who initially benefit from imatinib eventually become resistant. The most common mechanism of acquired resistance is through a secondary KIT mutation, usually located in either the NH2-terminal or the COOH-terminal kinase domain, which disrupts imatinib binding by stabilizing the receptor in a more active conformation. The mechanism for the development of secondary KIT mutations remains unclear, but resistant patients with identifiable second-site mutations had been treated with imatinib longer than resistant patients lacking second-site mutations (3). The only Food and Drug Administrationapproved second-line tyrosine kinase inhibitor (TKI) for patients with advanced GIST who have progressed on or are intolerant to imatinib is sunitinib malate (Sutent; Pfizer). The clinical benefit from sunitinib following imatinib failure is influenced by the genomic location of both primary and secondary mutations of the activated kinase. Thus, progression-free survival and overall survival are significantly longer for patients with either KIT exon 9 mutation or KIT/PDGFRA wild-type tumors.

Furthermore, imatinib-resistant secondary mutations within the ATP-binding pocket (KIT exons 13 and 14) appear to be sensitive to sunitinib inhibition (4). However, after an initial response, patients developing sunitinib resistance are being diagnosed in the clinic. It remains unclear if similar mechanisms identified in imatinib failure are also responsible for the development of sunitinib resistance. Because sunitinib activity encompasses a broader spectrum of targeted kinases compared with imatinib, including antivascular endothelial growth factor receptor activity, it is possible that additional mechanisms play a role in the acquisition of resistance. The goal of our study was first to investigate the clinicopathologic and genomic characteristics associated with patients failing sunitinib therapy. Second, using an in vitro model, we tested the efficacy of novel TKI on the sunitinib-resistant mutants. Furthermore, to predict patterns of mutations arising during sunitinib therapy, we used a cell-based screen to identify mutations giving rise to drug resistance, the results of which can be used to generate a genotype-dependent algorithm for drug selection. Using N-ethyl-N-nitrosourea (ENU), a DNA-alkylating agent that is a highly potent mutagen in mice (5), we established a robust, unbiased mutagenesis system.

ENU mutagenesis alters predominantly AT base pairs and produces A/T->T/A transversions, A/T->G/C transitions, and, with much lower frequency, G/C->A/T transitions, G/C->C/G transversions, A/T->C/G transitions, and G/C->T/A transitions, thus producing a broad spectrum of missense mutations, which either may be loss-of-function or gain-of-function mutations. ENU mutagenesis was used to compare incidence and types of BCR-ABL kinase domain mutants emerging in the presence of imatinib, dasatinib, and nilotinib alone and in dual combinations in Ba/F3 cells. We have used this approach to investigate the development of drug-resistant mutations in KIT. As the pattern of imatinib-induced resistant mutations has been described in-depth, we focused on identifying mutations conferring sunitinib resistance and acquired secondary mutations associated with KIT502-3AYins primary mutations. By transforming the Ba/F3 murine pro-B-cell line with KIT502-3AYins, with or without ENU mutagenesis, we were able to reproduce the pattern and relative abundance of sunitinib- and imatinib-resistant secondary mutations identified previously in drug-resistant GIST patients with the exon 9 mutation.

Clinicopathologic features

Patients with a diagnosis of GIST, who developed imatinib resistance and subsequently developed progression on sunitinib after at least 1 year on therapy, were identified from our prospective sarcoma database at the Memorial Sloan-Kettering Cancer Center. Those who underwent surgical resection of their resistant tumors were included in the study. Based on the previous experience with imatinib resistance, we selected a 1-year cutoff of clinical response to sunitinib therapy (defined as failure to progress radiographically) to exclude patients with primary resistance and to allow sufficient time for clonal selection of second-site mutations under drug pressure. Clinical information was obtained from the prospective sarcoma database and review of medical charts, including extent of disease at the start of each TKI treatment, duration of imatinib and sunitinib therapy before development of resistance and/or surgical resection, and the best clinical response obtained on each TKI. This study was approved by the institutional review board.

The histologic slides from the surgical specimens were reviewed and the diagnosis of GIST was confirmed based on the morphology and immunostaining for the KIT antibody. Tissue from both imatinib- and sunitinib-resistant GIST resection was available in most cases for analysis of tumor response, such as degree of necrosis, mitotic activity, and expression of KIT.

KIT/PDGFRA genotyping

Mutation analysis was done as described previously (6). Genomic DNA was isolated from snap-frozen tumor tissue samples stored at 70C using a standard phenol-chloroform organic extraction protocol. All cases were tested for the known sites of KIT (exons 9, 11, 13, 14, and 17) and PDGFRA (exons 12, 14, and 18) mutations. Genomic DNA (1 g) was subjected to PCR using Platinum TaqDNA Polymerase High Fidelity (Life Technologies). Primer sequences and annealing temperatures were as described (6, 7). Direct sequencing of PCR products was done for all exons tested and each ABI sequence was compared with the National Center for Biotechnology Information human KITand PDGFRA gene sequences.

DNA constructs

The retroviral vector plasmid containing wild-type human KIT cDNA (GNNK- isoform), pMSCV-WTKIT-IRES-GFP, was generously provided by Dr. Gary Gilliland (Harvard Medical School). KIT mutations recapitulating the genotype found in sunitinib malate-resistant GIST patients were generated by site-directed mutagenesis PCR using QuikChange II XL Site-Directed Mutagenesis kit (Qiagen). KIT double-mutant isoforms hosting primary exon 9 mutation and an acquired secondary mutations in exon 13 or 17 were generated as follows: KIT502-3AYins/V654A, KIT502-3AYins/D820Y, and KIT503-3AYins/N822K. KIT502-3AYins single mutant was used as a control. All constructs were verified with direct sequencing.

Establishing Ba/F3 KIT mutants stable transformant cell lines

The interleukin-3dependent Ba/F3 murine pro-B cell line was obtained from the German Collection of Microorganisms and Cell Culture. Ba/F3 cells were transfected and selected as described previously (8). Briefly, Ba/F3 cells were cotransfected with retroviral vector plasmids containing KIT cDNA mutant isoforms and linear hygromycin-resistant DNA via electroporation. The electroporated cells were first grown in the presence of hygromycin for 10 days followed by sorting according to green fluorescent protein. Green fluorescent proteinpositive cells were further grown for another 2 weeks in the presence of interleukin-3 and then transformed to interleukin-3independent on interleukin-3 withdrawal. Ba/F3 KIT502-3AYins cells were supplemented with 20 ng/mL KIT ligand; all the double mutants described above did not require KIT ligand supplement. Cells were stained with PE-conjugated anti-CD117 antibody (BD Biosciences) and monitored by flow cytometry. Cell lysates were subjected to Western blotting using anti-KIT antibody (Calbiochem).

Inhibitors and in vitro drug testing

The following TKI were used: imatinib, sunitinib, sorafenib, dasatinib, and nilotinib. Nilotinib and dasatinib were synthesized in-house and kindly provided by Dr. Bayard Clarkson's laboratory at Memorial Sloan-Kettering Cancer Center. Imatinib, sunitinib, and sorafenib were purchased commercially and purified with column chromatography by Dr. Bayard Clarkson's laboratory at Memorial Sloan-Kettering Cancer Center. The kinases inhibitors were dissolved at 10 mmol/L in DMSO as stocks and working dilutions were freshly made before experiments.

Cell proliferation assays

Ba/F3 cells (0.5 105 per well) expressing KIT mutants were incubated with 10, 100, 1,000, 5,000, and 10,000 nmol/L imatinib, sunitinib, nilotinib, dasatinib, and sorafenib in 96-well plates at 37C for 48 h in triplicate. Cells were incubated for 4 h with [3H]thymidine (1 Ci; 0.037 MBq) before harvesting and [3H]thymidine incorporation was determined. Growth inhibition was plotted as the ratio of the average [3H]thymidine incorporation in drug-treated wells relative to no-drug controls. IC50 values for tested inhibitors were calculated by GraphPad Prism software version 5.00.

Apoptosis assays

Cells at a density of 1 106 were cultured in 24-well plates and incubated with imatinib, sunitinib, nilotinib, dasatinib, and sorafenib at concentrations to a range of 10, 100, 1,000, and 5,000 nmol/L for 48 h. Cells were harvested and stained with antiAnnexin V-PE antibody and 7-amino-actinomycin D (BD Biosciences). A minimum of 20,000 events were analyzed by FACScan (Becton Dickinson) within 1 h after staining. Data were analyzed by FlowJo 7.1.3.

Immunoprecipitation and Western blotting

Ba/F3 cells expressing KIT mutant isoforms were starved from serum and treated with the indicated concentrations of each inhibitor for 90 min. Cell lysis, immunoprecipitations, and immunoblotting were done as described previously (8). Cell lysates were incubated with anti-KIT antibody (Assay Designs) and Magna beads (Pierce Biotechnology) overnight at 4C. The precipitated KIT was separated by electrophoresis and transferred to nitrocellulose. Blots were probed with anti-phospho-tyrosine antibodies PY20 and PY99 (Santa Cruz Biotechnology), stripped, and reprobed with rabbit polyclonal anti-KIT antibody (Calbiochem).

ENU mutagenesis

Ba/F3 KIT502-3AYins cells were maintained in complete culture medium, RPMI 1640 containing 10 serum, 1 penicillin/streptomycin, and 20 ng/mL KIT ligand at an exponential growth rate. Cells were incubated with ENU (50 g/mL) at a density of 5 106/mL for 24 h and then washed three times with RPMI 1640, replated in complete medium, and expanded for 1 week under exponential growth conditions.

In vitro mutagenesis screen

ENU-exposed Ba/F3 KIT502-3AYins cells were cultured in 96-well plates (1.0 105 per well) in 150 L complete medium in the presence of various concentrations of sunitinib and imatinib. Sunitinib was supplemented at 0.05, 0.1, 0.25, 0.5, and 1 mol/L, and imatinib at 0.5, 1, 2.5, 5, and 10 mol/L, which corresponded to 1, 2, 5, 10, and 20 times their IC50 values (sunitinib = 54 nmol/L and imatinib = 517 nmol/L), respectively. The number of resistant colonies was counted by visual inspection every 2 to 3 days for at least 4 to 6 weeks. Single colonies were picked and expanded for analysis in 1 mL complete medium in the presence of the corresponding concentration of inhibitors used in the screen. Genomic DNA was extracted and subjected to PCR for possible mutations in the main hotspots (KIT exons 13, 14, and 17; ref. 3). In parallel, Ba/F3 KITAY502-3ins cells without ENU exposure were incubated with the same concentrations for each inhibitor. Resistant colonies were inspected and analyzed with the same protocol as the ENU-exposed cells. Ba/F3 KITWK557-8del cells with or without ENU treatment were grown in complete medium supplemented with 0.05, 0.1, 0.25, 0.5, and 1.0 mol/L imatinib according to 1, 2, 5, 10, and 20 times its IC50, respectively. Resistant colony selection and analysis were conducted as described above as an isoform control.

Fluorescence in situ hybridization

Cells were treated with colcemid (Life Technologies) at 50 ng/mL for 45 min before harvesting. KITAY502-3ins DNA was labeled with digoxingenin-11-dUTP (Roche) by nick translation. The slides and the probes were codenatured at 70C for 5 min and then incubated at 37C overnight in a dark moist chamber. Posthybridization wash was carried at 45C with 50 formamide/2 SSC. Mouse anti-digoxigenin IgG (5 g/mL; Roche) and FITC-conjugated goat anti-mouse IgG (5 g/mL; Invitrogen) were used for detection. The cells were then stained with 4,6-diamidino-2-phenylindole (Molecular Probes). Slides were viewed with a Nikon E600 epifluorescence microscope.

Secondary KIT mutations in the activation loop are associated with sunitinib resistance

We identified three imatinib-resistant patients who showed a clinical response to sunitinib for >1 year (mean, 23 months) before progressing and subsequently referred for surgical management. Tissue from the sunitinib-resistant nodules was available for genotyping. The sunitinib-resistant patients shared similar clinicopathologic and molecular findings: all had their primary GIST located in the small bowel, harbored a primary KIT exon 9 mutation, and showed a partial response to imatinib therapy for a mean of 24 months (range, 18-28) before developing progression. Furthermore, both patients with tissue available before sunitinib therapy lacked second-site imatinib-resistant mutations.

On microscopic evaluation, all sunitinib-resistant tumors showed increased cellularity, brisk mitotic activity, and strong and diffuse KIT immunoreactivity. The genotype analysis of the eight nodules from the three sunitinib-resistant patients revealed three different amino acid substitutions in the KIT activation loop (exon 17): D820Y, D820E, and N822K. In one patient, only one of the six resistant nodules analyzed showed the presence of a second-site KIT mutation (D820E).

Ba/F3 cells expressing KIT exon 9/exon 17 double mutants are sensitive to dasatinib and nilotinib inhibition

To investigate the sensitivity of sunitinib-resistant KIT mutants to other TKIs, we established stable Ba/F3 cell lines expressing primary KIT exon 9 (KIT502-3AYins) and secondary mutations in KIT exon 13 (KITV654A) and exon 17 (KITD820Y, KITN822K). KIT502-3AYins/D820Y and KIT502-3AYins/N822K were identified in sunitinib-resistant GIST patients. The third secondary mutation identified in a sunitinib-resistant patient, exon 17 KITD820E, failed transfection. The KIT502-3AYins/V654A mutant described previously in an imatinib-resistant, sunitinib-sensitive patient (4) was used as a control.

The efficacy of five TKIs with anti-KIT activity, including imatinib, sunitinib, nilotinib, dasatinib, and sorafenib, was tested in Ba/F3 KIT transfectants expressing the sunitinib-resistant double mutants, KIT502-3AYins/D820Y and KIT502-3AYins/N822K, recapitulating mutations identified in patients, and compared with the imatinib- and sunitinib-sensitive KIT ectodomain mutant, KIT502-3AYins, as well as to the imatinib-resistant and sunitinib-sensitive double mutant, KIT502-3AYins/V654A. Each cell line was treated with similar escalating doses of the five inhibitors. Inhibition of KIT kinase activity was monitored by immunoblotting and the biological consequences were evaluated by determining proliferation inhibition and induction of apoptosis. Growth-inhibitory effects were determined by 3H incorporation, whereas induction of apoptosis was evaluated by flow cytometry using Annexin V-PE Apoptosis Detection kit (Pharmingen).

Ba/F3 cells expressing exon 17 secondary mutations, KIT502-3AYins/D820Y, were resistant to either imatinib or sunitinib inhibition. Imatinib did not inhibit KIT phosphorylation at concentrations <5,000 nmol/L, nor did it show effects in biological assays. Sunitinib did not show any cellular effects <1,000 nmol/L, inhibiting cell growth with an IC50 of 1,486 nmol/L and inducing mild apoptosis at 1,000 nmol/L (Fig. 1), although it abrogated KIT kinase activity between 100 and 1,000 nmol/L. However, dasatinib completely abolished KIT phosphorylation at a low dose (10-100 nmol/L). Dasatinib also showed an excellent efficacy inhibiting cell growth, with an IC50 of 40.6 nmol/L and inducing apoptosis between 100 and 1,000 nmol/L. Nilotinib treatment resulted in marked decrease of KIT kinase activity <100 nmol/L and showed potent proliferation inhibition with an IC50 of 248 nmol/L. Induction of apoptosis was achieved with a higher dose of nilotinib (1,000 nmol/L; Fig. 1).

Similar experiments were done with Ba/F3 cells expressing the KIT502-3AYins/N822K mutant resulting in a comparable drug sensitivity profile. Although imatinib and sunitinib partially inhibited KIT kinase activity of this double mutant, cell proliferation or induction of apoptosis was not affected <1,000 nmol/L. In contrast, dasatinib and nilotinib inhibited the kinase activity at concentrations <100 and 1,000 nmol/L, inhibited cell growth with an IC50 values of 66.2 and 309 nmol/L, and induced apoptosis at 100 and 1,000 nmol/L, respectively. In contrast, Ba/F3 KIT502-3AYins cells were highly sensitive to both dasatinib and sunitinib and less sensitive to imatinib and nilotinib inhibition. Dasatinib and sunitinib inhibited cell growth with IC50 of 2.9 and 54 nmol/L, respectively. Dasatinib induced significant apoptosis at 10 nmol/L, whereas sunitinib induced apoptosis between 10 and 100 nmol/L. Furthermore, dasatinib dramatically decreased KIT kinase autophosphorylation at <10 nmol/L, whereas sunitinib showed marked inhibition of KIT kinase autophosphorylation between 10 and 100 nmol/L. Imatinib produced moderate inhibition on KIT kinase activity at 1,000 nmol/L, with an IC50 of 517 nmol/L, whereas nilotinib and sorafenib inhibited kinase activity at 100 nmol/L, with IC50 values of 695 and 861 nmol/L, respectively. These three inhibitors induced significant apoptosis at 1,000 nmol/L (Fig. 2).

As noted previously in the imatinib-resistant patients (9), our in vitro results validate that the secondary mutations in the ATP-binding pocket remain sensitive to sunitinib inhibition. Inhibition of KIT activation of the Ba/F3 KIT502-3AYins/V654A double-mutant was achieved with <10 nmol/L sunitinib, whereas inhibition of cell proliferation with an IC50 of 20.6 nmol/L, concomitant with induction of apoptosis at 100 nmol/L. Dasatinib-treated cells also showed dramatic decrease of KIT kinase activity at 10 nmol/L, whereas the IC50 for proliferation was 334.5 nmol/L and at 1,000 nmol/L induction of apoptosis 70. Imatinib partially inhibited KIT phosphorylation at 1,000 nmol/L but did not arrest the growth or induce apoptosis <5,000 nmol/L compared with the untreated control. Although nilotinib showed overt inhibition on KIT phosphorylation at 1,000 nmol/L and an IC50 of 1,045 nmol/L with regards to inhibition of cell proliferation, it did not induce apoptosis at a comparable dose. Table 1 summarizes these results in comparison with the IC50 of Ba/F3 KIT double mutants carrying a KIT exon 11 primary mutation (published in ref. 8).

Sorafenib showed moderate inhibition on KIT502-3AYins single mutant but only mild efficacy on KIT502-3AYins double mutants

In Ba/F3 cells expressing KIT502-3AYins, sorafenib inhibited the mutant kinase at 100 to 1,000 nmol/L, induced moderate apoptosis at 100 nmol/L, and inhibited cell proliferation with an IC50 of 861 nmol/L. In contrast, it inhibited exon 9 double mutants, KIT502-3AYins/D820Y, KIT502-3AYins/N822K, and KIT502-3AYins/V654A, with IC50 values of 842, 933, and 1,672 nmol/L, respectively, but did not induce significant apoptosis at comparable doses.

In vitro screen for sunitinib resistance revealed second-site mutations in the activation loop of KIT

To investigate the mechanisms of sunitinib resistance and generate a comprehensive map of acquired mutations, we developed a cell-based screen for drug resistance of KIT502-3AYins-associated secondary mutations. Ba/F3 KIT502-3AYins cells were treated with the mutagen ENU. Subsequently, the ENU-treated Ba/F3 KIT502-3AYins cells were incubated with sunitinib or imatinib to identify drug-resistant colonies. The incidence, patterns, and types of mutations acquired by sunitinib selection were compared with the ones induced by imatinib treatment. Furthermore, non-ENU-exposed Ba/F3 KIT502-3AYins cells were selected with the identical doses of sunitinib and imatinib to compare the resistance frequencies and mutation spectrum. To exclude the possibility that resistant colonies develop intrinsically due to genetic instability, Ba/F3 KIT502-3AYins cells were also screened without inhibitors. Cells cultured with KIT ligand, KITL, supplemented in the absence of TKI grew back in a week, without finding secondary mutations in randomly tested cells.

In concordance with the sunitinib-resistant genotypes observed in patients, the cell-based mutagenesis also showed that acquired secondary mutations in Ba/F3 KIT502-3AYins cells were only found in the KIT activation loop. Sequencing of 100 sunitinib-resistant clones resulting from the ENU-treated Ba/F3 KIT502-3AYins cells and 117 clones from the untreated cells showed that all secondary mutations are single amino acid substitutions in KIT exon 17, including D816V, D816F, D816A, D816H, D816Y, and D820G (Fig. 3). Among these mutations, D816V was the most prevalent substitution, accounting for 6.7 of ENU-treated resistant colonies. ENU-mutagenesis did not generate distinct secondary mutations compared with the non-ENU group. However, the spectrum of sunitinib-resistant mutations was more limited compared with the imatinib selected screen. Sequencing of 78 clones obtained after ENU treatment and 158 clones from non-ENU Ba/F3 KIT502-3AYins cells, imatinib-resistant mutations were observed in both KIT kinase domains, including T670I, D816V, and D816F.

The incidence of each type of mutation identified in sunitinib-resistant clones emerging from Ba/F3 KIT502-3AYins cells varied from 1 to 14 according to different doses (Fig. 4A), with overall mutation rates of 3.3, 9.5, and 5.6 at 2, 5, and 10 times IC50. The rate of imatinib-resistant clones under similar conditions was higher (19.4, 8, and 67, respectively). The resistant mutations were limited to D816V at higher doses (10 times IC50) for both inhibitors tested. At >20 times IC50, both drugs sufficiently suppressed the emergence of resistant colonies.

In contrast, the mutation rates of imatinib-resistant clones selected by similar dosages in Ba/F3 KIT557-8WKdel cells were 90 regardless of ENU treatment, displaying a similar mutation spectrum, including D816V, D816Y, N822K, N655K, and T670I (data not shown). The most prevalent mutation remained at D816, D816V with ENU treatment (79) and D816Y in the non-ENU group (86.5).

Frequency of resistant clones increased with ENU exposure and lower concentration of inhibitors

To determine whether the induced drug resistance was dose-dependent, ENU-treated Ba/F3 KIT502-3AYins cells were exposed to escalating concentrations of sunitinib (0.05, 0.1, 0.25, 0.5, and 1 mol/L corresponding to 1, 2, 5, 10, and 20 times IC50, respectively). Non-ENU-treated Ba/F3 KIT502-3AYins cells were treated with similar doses of sunitinib and imatinib. The frequencies of resistant colonies were determined by visual inspection of the number of colonies relative to the total number of cells. The frequency of resistant colonies decreased with increasing sunitinib dose, with a rate of 6.8 106, 6.7 106, 5.4 106, 2.2 106, and 0.35 106 at 1, 2, 5, 10, and 20 times IC50, respectively (Fig. 4B). Imatinib selected colonies of ENU-treated Ba/F3 KIT502-3AYins cells showed similar dose-dependent resistant frequencies of 10 106, 5.8 106, 2.9 106, 0.35 106, and 0.1 106 at similar doses.

Cells exposed to lower doses than the IC50 of sunitinib (10 and 20 nmol/L) grew back without individual colony formation within a week, suggesting that these concentrations did not sufficiently inhibit Ba/F3 KIT502-3AYins cells. Conversely, at >20 times the IC50 of sunitinib, no resistant colonies evolved within 60 days. Similar observations were made with imatinib.

Untreated Ba/F3 KIT502-3AYins cells yielded a lower frequency of resistant colonies compared with ENU-treated cells at similar drug concentrations (P < 0.01, pair of comparison 2). The frequency of resistant colonies observed in non-ENU conditions was 1.4, 1.15, 0.8, 0.4, and 0 colonies per million cells at 1, 2, 5, 10, and 20 times IC50 of sunitinib, respectively (Fig. 4B). The incidence of imatinib-resistant colonies in non-ENU-treated group was 5.1, 3.1, 1.45, 0.05, and 0 colonies per million cells at equipotent dosages. No overt difference in the median growth time of resistant colonies was seen between sunitinib- and imatinib-selected BaF3 KIT502-3AYins cells.

Sunitinib-selected resistant clones were insensitive to sunitinib and imatinib inhibition

To validate that clones derived from the sunitinib-resistant screen were indeed refractory to sunitinib inhibition, cells were cultured in the presence of various concentrations of sunitinib, and IC50 values were determined. In parallel, the imatinib IC50 values was also determined in these cells. Sunitinib-selected clones, with or without secondary mutations, were tested, showing similar insensitivity to sunitinib compared with sunitinib-sensitive parental Ba/F3 KITAY502-3ins cells. The imatinib IC50 values in these clones was >5 mol/L.

KITAY502-3ins sunitinib-resistant clones show biochemical activation of KIT but no alteration in KIT copy number

By Western blot, protein extracts from KITAY502-3ins-resistant clones showed similar levels of total and phosphorylated KIT expression to the parental sunitinib-sensitive Ba/F3-KITAY502-3ins cells. To explore the possibility of KIT copy number changes in sunitinib-resistant clones without detectable secondary mutations, we investigated the genomic integration of KITAY502-3-ins cDNA plasmid in both parental Ba/F3 KITAY502-3ins and sunitinib-resistant clones. Cells were subjected to fluorescence in situ hybridization using digoxigenin-labeled KITAY502-3ins cDNA, showing a single integration site on chromosome 6C2 in both groups. Therefore, no genetic difference related to the KIT locus was noted between the parental Ba/F3 KITAY502-3-ins and the sunitinib-resistant clones.

The primary genetic event responsible for the pathogenesis of GIST is a gain-of-function mutation in the KIT proto-oncogene or, less commonly, in the PDGFRA gene (10, 11). More than two-thirds of all GIST patients carry mutations in KIT exon 11, which encodes the juxtamembrane domain, whereas 10 to 15 of patients show mutations within the extracellular domain of KIT (in exon 9; refs. 6, 12). The vast majority of KIT exon 9 mutations represent an identical tandem duplication of six nucleotides, encoding AY 502-503. GISTs harboring KIT exon 9 mutations define a distinct subset, characterized predominantly by small bowel location and an aggressive clinical behavior (6, 13).

Therapeutic inhibition of KIT and PDGFRA kinase activity by imatinib mesylate (Gleevec; Novartis) has emerged as frontline treatment for patients with metastatic or locally advanced inoperable GIST. Imatinib achieves disease control in 70 to 85 of patients with advanced GIST. The clinical response varies significantly according to different molecular subtypes of GIST. As such, patients with GIST carrying KIT exon 11 mutations show the best response to imatinib with a partial response rate of 84 compared with only 48 in patients with KIT exon 9 mutations (2). KIT exon 9 mutations were the strongest adverse prognostic indicator for response to imatinib, increasing the relative risk of progression by 171 and the relative risk of death by 190 (14). Furthermore, patients whose tumors harbored KIT exon 9 mutations had a significantly superior progression-free survival when treated with the high-dose regimen compared with patients with exon 11 mutations (14). However, in spite of the prolonged responses with escalating imatinib dose, most patients subsequently experienced disease progression.

The only Food and Drug Administrationapproved second-line TKI for patients with advanced GIST, who have progressed on or are intolerant of imatinib, is sunitinib malate. Sunitinib is a small-molecule TKI with potent antiangiogenic and antitumor activities, which has shown efficacy against GIST, acceptable tolerability, and safety in a double-blind placebo-controlled phase III trial (15). Patients with KIT exon 9 mutations GISTs showed the most sustained response to sunitinib inhibition. As sunitinib resistance is now emerging in the clinical practice, new therapeutic strategies are needed to treat these patients.

As learned from the imatinib resistance experience, second-site KIT mutations develop more frequently in association with primary KIT exon 11 mutations, which confer the most sustained clinical response to imatinib. This finding might reflect the long time-span of exposure to the drug and not only due to the location of the primary mutations. Thus, secondary mutations were found in 73 to 86 of imatinib-resistant patients harboring exon 11 primary mutations and only 19 to 33 of patients with exon 9 primary mutation developed secondary mutations (3, 9, 16). Furthermore, the pattern of the second-site mutations in the setting of imatinib resistance was exclusively point mutations, evenly divided between the first and the second kinase domains (3, 17, 18). In keeping with these observations, the three patients who developed sunitinib resistance after at least 1 year of clinical benefit had second-site mutations in the KIT kinase activation domain (N822K, D820Y, and D820E). Expression of these sunitinib-resistant double mutants in an in vitro Ba/F3 model showed sensitivity to both dasatinib and nilotinib, suggesting alternative therapeutic options for these patients.

To establish a comprehensive profile of acquired secondary mutations that confer drug resistance to KIT exon 9 mutants, Ba/F3 cells expressing KIT502-3AYins were subjected to ENU mutagenesis followed by sunitinib selection and emerging resistant clones were identified and characterized. The KIT mutations identified by in vitro mutagenesis recapitulated the type and location of second-site KIT mutations observed in both sunitinib- and imatinib-resistant patients. Thus, all sunitinib-resistant secondary mutations identified in the Ba/F3-KIT502-3AYins cells were similarly point mutations in the KIT kinase activation loop. However, the mutations identified in the in vitro screen were dominated by D816V. The high incidence of this mutation is most likely conferred by the significant advantage in its activation rate. Substitutions at this site, D816H/V, have shown increases in activation rate of 184- and 536-fold, respectively (19). Furthermore, D816 mutations appear to negatively influence the inhibitory conformation of the juxtamembrane domain of KIT (19). Several substitutions at this site have been described in TKI-resistant GIST patients, including D816H/F/A/G/E/V. However, the rarity of D816V mutation in progressing GIST patients might be an indication of its high malignancy, which may induce tumor lethality in vivo. Alternatively, the possibility of this dominant D816V genotype being intrinsic to the Ba/F3 cells system remains a consideration, because Ba/F3 cells are murine pro-B cells and D816V is the common mutation in mastocytosis.

Importantly, KIT exon 13 and 14 mutations were not detected either in the sunitinib resistance screen or in progressing tumors of patients of sunitinib-resistant patients, as sunitinib is known to be efficacious with ATP-pocket second-site mutations (4, 9). In contrast, in vitro mutagenesis of Ba/F3 KIT502-3AYins cells and selection with imatinib identified secondary mutations mapped in both KIT kinase domains (mostly D816V and T670I). These findings are in agreement with results of in vitro mutagenesis of Ba/F3 Bcr-Abl cells, in which the vast majority of imatinib-resistant mutations were substitutions within the kinase domain of the fusion protein (20, 21). However, the spectrum of resistance mutations obtained with Ba/F3 KIT502-3AYins cells was narrower compared with the results obtained with Bcr-Abl. The observed difference most likely is not related to the methodology used, which was very similar in both studies but rather intrinsic to the specific oncogenes tested. The relatively limited variety of mutations observed in the Ba/F3 KIT502-3AYins screen cannot be explained only by the lower rate of mutation detected in the resistant clones, because a significantly higher rate of mutations (90) noted in the Ba/F3-KIT557-8WKdel cells was not accompanied by an enhanced mutation spectrum. To overcome the low yield of resistant colonies, ENU mutagenesis was applied, which indeed increased the frequency of both sunitinib- and imatinib-resistant colonies, indicating that ENU mutagenesis is beneficial in establishing an unbiased and robust cell-based resistance screen system of KIT gene.

The results of ENU mutagenesis followed by sunitinib selection were in concordance with the findings observed with sunitinib-resistant patients, defining D816 and D820 amino acids as the most vulnerable sites for acquired mutations within the KIT kinase activation loop. Similarly, imatinib selection identified T670, D816, and D820 as resistant mutation hotspots. The frequency of secondary mutations with sunitinib or imatinib selection with Ba/F3 KIT502-3AYins cells was quite low (Fig. 4). In contrast, clones carrying a juxtamembrane KIT557-8WKdel mutation had a significantly higher mutation rate. In keeping with these differences, previous biochemical studies revealed that tumors with KIT exon 9 mutation had less AKT activation than tumors with mutations of exon 11, suggesting that these KIT mutants may recruit different downstream substrates (22). The latter hypothesis may also explain the genotype-dependent response of GIST patients to different TKIs.

Recently, the crystal structure of the KIT ectodomain was characterized, showing that KIT dimerization is driven by bivalent binding of KIT ligand and stabilized by lateral D4-D4 and D5-D5 interactions of two neighboring KIT ectodomains (23). Activating oncogenic mutations reported thus far within the ectodomain of KIT map to the D5-D5 interface, which presumably stabilize receptor dimers and thus activate the kinase in the absence of ligand. In contrast, juxtamembrane domain mutations are thought to induce kinase activation by disrupting the autoinhibitory conformation of the juxtamembrane domain. The different mechanisms of activation of constitutive KIT kinase activity seen with different mutant oncoproteins may affect the mutation rates yielded from the mutagenesis screen. KIT exon 11 mutants largely rely on the structural changes conferred by secondary mutations to overcome inhibition of KIT kinase activity and with extracellular domain mutants similarly acquisition of secondary mutations that stabilize the active conformation of the KIT kinase is required. It is therefore not surprising that the spectrum of secondary mutations between the two mutant isoforms did not differ. Although both D816 and D820 do not interact directly with the inhibitors, the mutations at these sites affect the A-loop conformation as well as the overall conformation of the kinase (Fig. 5), preventing drug binding (19).

Unfortunately, comprehensive molecular studies are constrained by the low number of patients who are surgical candidates after failing two TKIs; thus, the tissue available for investigating mechanisms of resistance is limited. Taken together, the data obtained from clinical samples as well as from the in vitro mutagenesis screens suggest that sunitinib resistance shares similar pathogenetic mechanisms seen in imatinib failure, with acquisition of secondary mutations in the activation loop conferring resistance to both drugs. In contrast, the imatinib resistance screens showed a slightly higher mutation rate and a wider mutation spectrum, spanning both kinase domains. In addition, the in vitro resistance screens applied to either ectodomain or juxtamembrane domain KIT mutants suggest that the location of the primary mutation may trigger different mechanisms of drug resistance, with a higher rate of resistant mutations seen with primary KIT exon 11 mutants.

R. DeMatteo, advisory board and honoraria, Novartis. R. Maki, research funding, Pfizer. The other authors reported no conflicts of interest.

We thank Kai Xu (Structural Biology Program, Sloan Kettering Institute), Diann DeSantis for clinical follow-up, and Milagros Soto for editorial support.

1
Blanke
CD
,
Demetri
GD
,
von Mehren
M
, et al
. 
Long-term results from a randomized phase II trial of standard- versus higher-dose imatinib mesylate for patients with unresectable or metastatic gastrointestinal stromal tumors expressing KIT
.
J Clin Oncol
2008
;
26
:
620
5
.
2
Heinrich
MC
,
Corless
CL
,
Demetri
GD
, et al
. 
Kinase mutations and imatinib response in patients with metastatic gastrointestinal stromal tumor
.
J Clin Oncol
2003
;
21
:
4342
9
.
3
Antonescu
CR
,
Besmer
P
,
Guo
T
, et al
. 
Acquired resistance to imatinib in gastrointestinal stromal tumor occurs through secondary gene mutation
.
Clin Cancer Res
2005
;
11
:
4182
90
.
4
Prenen
H
,
Cools
J
,
Mentens
N
, et al
. 
Efficacy of the kinase inhibitor SU11248 against gastrointestinal stromal tumor mutants refractory to imatinib mesylate
.
Clin Cancer Res
2006
;
12
:
2622
7
.
5
Russell
WL
,
Kelly
EM
,
Hunsicker
PR
,
Bangham
JW
,
Maddux
SC
,
Phipps
EL
. 
Specific-locus test shows ethylnitrosourea to be the most potent mutagen in the mouse
.
Proc Natl Acad Sci U S A
1979
;
76
:
5818
9
.
6
Antonescu
CR
,
Sommer
G
,
Sarran
L
, et al
. 
Association of KIT exon 9 mutations with nongastric primary site and aggressive behavior: KIT mutation analysis and clinical correlates of 120 gastrointestinal stromal tumors
.
Clin Cancer Res
2003
;
9
:
3329
37
.
7
Agaram
NP
,
Laquaglia
MP
,
Ustun
B
, et al
. 
Molecular characterization of pediatric gastrointestinal stromal tumors
.
Clin Cancer Res
2008
;
14
:
3204
15
.
8
Guo
T
,
Agaram
NP
,
Wong
GC
, et al
. 
Sorafenib inhibits the imatinib-resistant KITT670I gatekeeper mutation in gastrointestinal stromal tumor
.
Clin Cancer Res
2007
;
13
:
4874
81
.
9
Heinrich
MC
,
Maki
RG
,
Corless
CL
, et al
. 
Primary and secondary kinase genotypes correlate with the biological and clinical activity of sunitinib in imatinib-resistant gastrointestinal stromal tumor
.
J Clin Oncol
2008
;
26
:
5352
9
.
10
Hirota
S
,
Isozaki
K
,
Moriyama
Y
, et al
. 
Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors
.
Science
1998
;
279
:
577
80
.
11
Heinrich
MC
,
Corless
CL
,
Duensing
A
, et al
. 
PDGFRA activating mutations in gastrointestinal stromal tumors
.
Science
2003
;
299
:
708
10
.
12
Rubin
BP
,
Singer
S
,
Tsao
C
, et al
. 
KIT activation is a ubiquitous feature of gastrointestinal stromal tumors
.
Cancer Res
2001
;
61
:
8118
21
.
13
Lasota
J
,
Kopczynski
J
,
Sarlomo-Rikala
M
, et al
. 
KIT 1530ins6 mutation defines a subset of predominantly malignant gastrointestinal stromal tumors of intestinal origin
.
Hum Pathol
2003
;
34
:
1306
12
.
14
Debiec-Rychter
M
,
Sciot
R
,
Le Cesne
A
, et al
. 
KIT mutations and dose selection for imatinib in patients with advanced gastrointestinal stromal tumours
.
Eur J Cancer
2006
;
42
:
1093
103
.
15
Demetri
GD
,
van Oosterom
AT
,
Garrett
CR
, et al
. 
Efficacy and safety of sunitinib in patients with advanced gastrointestinal stromal tumour after failure of imatinib: a randomised controlled trial
.
Lancet
2006
;
368
:
1329
38
.
16
Nishida
T
,
Kanda
T
,
Nishitani
A
, et al
. 
Secondary mutations in the kinase domain of the KIT gene are predominant in imatinib-resistant gastrointestinal stromal tumor
.
Cancer Sci
2008
;
99
:
799
804
.
17
Debiec-Rychter
M
,
Cools
J
,
Dumez
H
, et al
. 
Mechanisms of resistance to imatinib mesylate in gastrointestinal stromal tumors and activity of the PKC412 inhibitor against imatinib-resistant mutants
.
Gastroenterology
2005
;
128
:
270
9
.
18
Wardelmann
E
,
Merkelbach-Bruse
S
,
Pauls
K
, et al
. 
Polyclonal evolution of multiple secondary KIT mutations in gastrointestinal stromal tumors under treatment with imatinib mesylate
.
Clin Cancer Res
2006
;
12
:
1743
9
.
19
Gajiwala
KS
,
Wu
JC
,
Christensen
J
, et al
. 
KIT kinase mutants show unique mechanisms of drug resistance to imatinib and sunitinib in gastrointestinal stromal tumor patients
.
Proc Natl Acad Sci U S A
2009
;
106
:
1542
7
.
20
von Bubnoff
N
,
Veach
DR
,
van der Kuip
H
, et al
. 
A cell-based screen for resistance of Bcr-Abl-positive leukemia identifies the mutation pattern for PD166326, an alternative Abl kinase inhibitor
.
Blood
2005
;
105
:
1652
9
.
21
Bradeen
HA
,
Eide
CA
,
O'Hare
T
, et al
. 
Comparison of imatinib mesylate, dasatinib (BMS-354825), and nilotinib (AMN107) in an N-ethyl-N-nitrosourea (ENU)-based mutagenesis screen: high efficacy of drug combinations
.
Blood
2006
;
108
:
2332
8
.
22
Duensing
A
,
Medeiros
F
,
McConarty
B
, et al
. 
Mechanisms of oncogenic KIT signal transduction in primary gastrointestinal stromal tumors (GISTs)
.
Oncogene
2004
;
23
:
3999
4006
.
23
Yuzawa
S
,
Opatowsky
Y
,
Zhang
Z
,
Mandiyan
V
,
Lax
I
,
Schlessinger
J
. 
Structural basis for activation of the receptor tyrosine kinase KIT by stem cell factor
.
Cell
2007
;
130
:
323
34
.
24
Liegl
B
,
Kepten
I
,
Le
C
, et al
. 
Heterogeneity of kinase inhibitor resistance mechanisms in GIST
.
J Pathol
2008
;
216
:
64
74
.

Competing Interests

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.