Avapritinib was obtained from ChemieTek (Indianapolis, IN). Fetal bovine serum (FBS), penicillin/streptomycin, trypsin 0.25%, and Dulbecco’s modified Eagle’s Medium (DMEM), were bought from Hyclone (Pittsburgh, PA). Antibodies against ABCB1, ABCG2, and GAPDH were bought from Thermo Fisher Scientific Inc. (Rockford, IL). Paclitaxel, doxorubicin, mitoxantrone, SN-38, cisplatin, and verapamil were purchased from Sigma-Aldrich (St. Louis, MO). [³H]-paclitaxel (15 Ci/mmol) and [³H]-mitoxantrone (2.5 Ci/mmol) were purchased from Moravek Biochemicals, Inc. (Brea, CA). Alexa Fluor 488 anti-mouse IgG antibody and 4′,6-diamidino-2-phenylindole (DAPI) were purchased from Thermo Fisher Scientific Inc. (Rockford, IL). Formaldehyde was purchased from J.T. Baker Chemical Co. (Phillipsburg, NJ). Bovine Serum Albumin (BSA) and 10X phosphate buffered solution (PBS) were obtained from VWR chemicals, LLC (Solon, OH). The chemicals utilized for the ATPase assay were as described in our previous study [27].

The SW620 cell line which is obtained from a human colorectal adenocarcinoma source, its doxorubicin-resistant SW620/AD300 cell line, the HEK293/pcDNA3.1 cell line which is obtained from a human embryonic kidney source, and HEK293/ABCB1 (transfected with human ABCB1 cDNA [28]) were used in this study to test for ABCB1 reversal activity. The NCI-H460 cell line which is obtained from a non-small cell lung cancer source, its mitoxantrone-resistant NCI-H460/MX20, HEK293/pcDNA3.1, and HEK293/ABCG2, transfected with wild-type (WT) ABCG2 (HEK/ABCG2-R482) or mutants of ABCG2 (HEK/ABCG2-R482G and HEK/ABCG2-R482T) were used for ABCG2 reversal study. The transfected cells were cultured in a growth medium containing 2 mg/mL G418. All cell lines were cultured in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin, at 37 °C and 5% CO_2.

In order to assess drug sensitivity, modified MTT colorimetric assay was used as previously described [4,15]. The IC₅₀ values, that is, the drug concentration required to limit the the growth of cells by 50% were calculated from the cytotoxicity curves using a modified Bliss method [29]. Resistance fold (Rf) was estimated using the following formula:

$\frac{IC_{50} \quad of the in \quad - \quad sensitive \quad (resistant) \quad cells \quad in \quad the \quad presence \quad or \quad absence \quad of \quad a \quad drug \quad (inhibitor)}{IC_{50} \quad of \quad sensitive \quad (parental) \quad cells \quad without \quad a \quad drug \quad (inhibitor)}$

Avapritinib was tested at concentrations of 1 µM and 5 µM. Verapamil (5 µM) and fumitremorgin C (FTC, 5 µM) were employed as as a positive control inhibitors of ABCB1 and ABCG2, respectively.

Cells were trypsinized and treated with avapritinib, or verapamil or FTC at the indicated concentrations for 2 h. After 2 h, the cells were further treated with either 0.1 μM [³H]-paclitaxel or 0.1 μM [³H]-mitoxantrone for additional 2 h. Following this, the cells were washed twice in ice-cold PBS and lysed with 10 mM lysis buffer (pH 7.4, containing 1% Triton X-100 and 0.2% SDS). Cell lysates were aliquoted in vials containing 5 mL scintillation fluid and radiation was estimated using a Packard TRI-CARB 1900CA liquid scintillation analyzer from Packard Instrument Company, Inc. (Downers Grove, IL).

To assess the efflux of [³H]-paclitaxel or [³H]-mitoxantrone from parental and MDR cells, cells were treated with 0.1 µM [³H]-paclitaxel or [³H]-mitoxantrone as mentioned in the accumulation experiment above. The cells were washed twice and cultured in fresh media with or without avapritinib or positive control inhibitors. 1 ml of cell suspension was aliquoted at 0, 30, 60, and 120 min. The aliquots were washed twice with ice-cold PBS and analyzed by the liquid scintillation analyzer as previously described [30].

The cells were harvested and washed with PBS. Cells were lysed using lysis buffer as previously described [31,32]. The lysates were centrifuged at 12000 rpm for 20 min and the supernatant was collected and stored at 80 °C for Western blot analysis. Protein concentration was determined using the bicinchoninic acid (BCA™) protein assay (Thermo Scientific, Rockford, IL).

Equal amounts of total cell lysates (40 μg protein) were resolved by sodium dodecyl sulfate polyacrylamide gel (4-12% gradient gel) electrophoresis (SDS-PAGE). The gel was electroblotted onto polyvinylidene fluoride (PVDF) membranes at 90V for 2 h. The membranes were blocked with 5% skim milk for 2 h followed by incubation with an ABCB1 specific (dilution of 1 : 500), an ABCG2 specific (dilution of 1 : 500), or a GAPDH specific (dilution 1 : 1000) primary antibody, overnight at 4 °C. The following day, membranes were treated with secondary antibody which was HRP (horseradish peroxidase)-conjugated. The protein-antibody complex was detected using the enhanced chemiluminescence detection system.

The two pairs of cell lines were seeded as 10,000 cells per well in 24-well plates and cultured at 37 ℃ for 24 h. The cells were then incubation with 5 µM avapritinib at the indicated time points. The cells were then washed twice with cold PBS and fixed in 4% formaldehyde for 15 min. Subsequently, the cells were permeabilized with 0.25% Triton X-100 followed by incubation with BSA (6% with PBS) for 1 h. Then the cells were treated with antibodies specific for ABCB1 (dilution 1 : 1000) or ABCG2 (dilution 1 : 1000) overnight at 4 ℃. The next day, cells were treated with secondary antibody conjugated with Alexa Fluor 488 for 2 h in the dark. DAPI was used to stain the nuclei for 15 min. Cell images were taken by a Nikon TE-2000S fluorescence microscope (Nikon Instruments Inc., Melville, NY, USA).

The ATP cleaving ability of ABCB1 and ABCG2 in the presence avapritinib was assessed using PREDEASY ATPase Kits. In brief, the crude membranes of High-five insect cells were incubated with avapritinib (0 to 40 μM) and the ATPase activity was measured as previously described [6,33].

Molecular simulations were performed using Maestro v11.1 (Schrödinger, LLC, Cambridge, MA). Ligands and proteins were prepared using default protocols as previous described [34]. The human ABCB1 homology model was established on the basis of the refined mouse ABCB1 protein (PDB ID: 4M1M) which was kindly provided by S. Aller [35]. The docking grid (length: 25 Å) for human ABCB1 was refined by setting the centroid with the amino acid residues that are suggested to have interaction with substrates [35]. The cryo-EM structure of human ABCG2 (PDB ID: 6ETI) is available, and the docking grid was generated at the drug binding cavity by selecting residues of importance, including Ala397, Val401, Leu405, Leu539, Ile543 and Phe547 (A397, V401, L405, L539, I543 and F547) [36]. Glide v 7.4 XP (extra precision) docking default protocol (Schrödinger, LLC, New York, NY) was adopted, then induced-fit docking (IFD) was conducted using the best scored binding pose of ligand and protein from the Glide XP docking to get an optimal binding simulation. The top scoring docked pose of ligand-protein complex was subjected to graphical analysis. The docking scores were calculated and expressed as kcal/mol.

All the studies were performed minimum three independent times. Student’s t-test was used to determine the statistical significance between two groups using Microsoft Excel 2017. The a priori P value was P < 0.05.

In order to assess the cytotoxic effects of avapritinib, the MTT assay was performed. Our results showed that about 80-85% of both sensitive and resistant cells survived treatment with 5 μM of avapritinib (Fig. 1), indicating that avapritinib is relatively non-toxic up to a concentration of 5 μM. Thus, avapritinib was used at concentrations of 1 and 5 μM to test its chemosensitizing and reversal effects in combination with chemotherapeutic drugs.

Figure 1.

Cytotoxicity of avapritinib in parental and drug-resistant cells. Percent cell survival was measured after treatment with avapritinib for 72 h in parental and MDR cells: SW620 and SW620/AD300 cells (A), HEK293/pcDNA3.1 and HEK293/ABCB1 (B), NCI-H460 and NCI-H460/MX20 cells (C), and HEK293/pcDNA3.1 and HEK/ABCG2-R482, HEK/ABCG2-R482G, HEK/ABCG2-R482T (D). Points with error bars represent the mean ± SD for independent determinations in triplicate. The above figures are representative of three independent experiments.

Cell survival assay was conducted to determine the effect of avapritinib on drug sensitivity of ABCB1 substrates. As shown in Table 1, the substrate anti-cancer drugs of ABCB1 exerted a significantly higher IC₅₀ in SW620/AD300 (ABCB1-overexpressing) cells than SW620 (parental) cells. Avapritinib (1 and 5 μM) restored the sensitivity of ABCB1 substrates as evident from a decrease in the IC₅₀ values of paclitaxel and doxorubicin, in SW620/AD300 cells . Similarly, reduction in the IC₅₀ values were observed in HEK293/ABCB1 cells upon treatment with 1 and 5 μM avapritinib (Table 1). For ABCB1 substrates, paclitaxel and doxorubicin were used and for non-ABCB1 substrate, cisplatin was used. Verapamil (5 μM) was used as a positive control inhibitor of ABCB1.

Studies have reported that ABCG2 exists in wild type and mutant forms which differ in their substrate and transport inhibitor specificity [8,37]. To this end, we tested if avapritinib could overcome both wild-type and mutant ABCG2. Our results indicated that avapritinib at 1 and 5 μM decreased the IC₅₀ values of mitoxantrone and SN38 in HEK/ABCG2-R482, HEK/ABCG2-R482G and HEK/ABCG2-R482T cells (Table 2). Additionally, avapritinib sensitized the resistant NCI-H460/MX20 cells (Table 3). However, avapritinib did not exerted any change in the IC₅₀ values of mitoxantrone and SN-38 in the parental cells that do no overexpress ABCG2. FTC, a known ABCG2 blocker, was used as a positive control inhibitor at 5 μM.

To ascertain that avapritinib sensitizes ABCB1 or ABCG2-overexpressing cells by increasing the accumulation of their respective substrates, we conducted the intracellular accumulation assay. The incubation of SW620/AD300 cells with avapritinib (1 or 5 μM) enhances the intracellular accumulation of [³H]-paclitaxel as compared to parental cells (Fig. 2A). Verapamil (5 μM) exerted similar effects. Likewise, avapritinib (1 or 5 μM) displayed a significantly higher accumulation of [³H]-mitoxantrone NCI-H460/MX20 cells than in parental cells (Fig. 2B). FTC (5 μM) showed similar effect.

Figure 2.

Effect of avapritinib on the accumulation of drug transport substrates of ABCB1 and ABCG2. The effect of avapritinib on the accumulation of [³H]-paclitaxel in SW620 and SW620/AD300 cells (A) and effect of avapritinib on accumulation of [³H]-mitoxantrone in NCI-H460 and NCI-H460/MX20 cells (B). Columns are the means of triplicate determinations (n = 3), error bars represent the SD, and *P < 0.05 versus the control group. Verapamil (5 μM) or FTC (5 μM) were used as positive controls for ABCB1- or ABCG2-overexpressing cells, respectively.

In order to determine whether the increase in accumulation of ABCB1 and ABCG2 substrates is due to the blockade of efflux function, we conducted the efflux assay. Our results showed that in the absence of avapritinib, a significantly lower intracellular concentration of [³H]-paclitaxel accumulated in ABCB1 overexpressing cells than in parental cells, due to the efflux of [³H]-paclitaxel by ABCB1 (Figs. 3A and 3B). Interestingly, avapritinib (1 or 5 µM) significantly decreased the efflux of [³H]-paclitaxel from ABCB1 overexpressing cells. Moreover, avapritinib, at 1 or 5 µM, exhibited a similar reduction in efflux of [³H]-mitoxantrone from ABCG2 overexpressing cells (Figs. 3C and 3D). The positive control inhibitors, verapamil or FTC, showed similar effect.

Figure 3.

Effect of avapritinib on the efflux of the drug transport substrates of ABCB1 and ABCG2. A time course (0, 30, 60, 120 min) versus percentage of intracellular [³H]-paclitaxel remaining (%) was plotted to show the effect of avapritinib in SW620 (A) and SW620/AD300 cells (B). A time course (0, 30, 60, 120 min) versus percentage of intracellular [³H]-mitoxantrone remaining (%) was plotted to show the effect of avapritinib in NCI-H460 (C) and NCI-H460/MX20 cells (D). Lines are the means of triplicate determinations (n = 3). Error bars represent the SD and *P < 0.05 versus the control group. Verapamil (5 μM) or FTC 5 (μM) were used as positive controls for ABCB1- or ABCG2-overexpressing cells, respectively.

It is important to understand if the reversal effect of avapritinib is by the inhibition of the efflux function or a downregulation of ABCB1 or ABCG2 expression. To this end, we assessed the effect of avapritinib on the expression levels of ABCB1 or ABCG2 protein. Our results showed that avapritinib did not change the expression levels of ABCB1 or ABCG2 (Figs. 4A and 4B) upon avapritinib (5 µM) treatment for 72h.

Figure 4.

Effect of avapritinib on the expression and subcellular localization of ABCB1 and ABCG2. The effect of avapritinib on the expression of ABCB1 and ABCG2 was tested after the cells were treated with 5 μM avapritinib for 0, 24, 48, 72, and 96 h (A, B). Equal amounts of total cell lysates were used for each sample and Western blot analysis was performed. The differences were not statistically significant (P > 0.05). The effect of avapritinib at 5 μM on the subcellular localization of ABCB1 or ABCG2 in ABCB1-overexpressing SW620/AD300 (C) or ABCG2-overexpressing NCI-H460/MX20 cells (D). Scale bar, 10 μm. DAPI (blue) was used to counterstain the nuclei.

Although our Western blot results confirmed that avapritinib does not affect the expression of ABCB1 or ABCG2, it could be argued that avapritinib translocases the ABC transporter protein from the cell membrane to the cytoplasm of cells. We performed staining of the resistant cells and confirmed that avapritinib (5 µM) did not alter the native subcellular localization of ABCB1 or ABCG2 (Figs. 4C and 4D).

In order to study the effect of avapritinib on the ATP cleaving activity of ABCB1 or ABCG2, we measured the ability to hydrolyze ATP in the presence of avapritinib (0-40 μM). Avapritinib stimulated the vanadate-sensitive ATPase activity of ABCB1 or ABCG2 with a fold-stimulation of 2.3 or 1.7-fold, respectively, as compared to the basal activity (Fig. 5). Avapritinib at 2.7 μM or 1.5 μM exhibited a 50% of maximal stimulation or ABCB1 or ABCG2, respectively.

Figure 5.

Effect of avapritinib on orthovanadate (Vi)-sensitive ABCB1 and ABCG2 ATPase activity. Crude membranes (10 μg protein/reaction) from High-five cells expressing ABCB1 or ABCG2 were incubated with increasing concentrations of avapritinib (0-40 μM). The concentration of avapritinib was plotted for ABCB1 (A) or ABCG2 (B). The inset shows stimulation of ATP hydrolysis at concentrations of 0-10 μM avapritinib. The mean values are plotted, and error bars depict SD obtained from the average of at list three independent experiments.

The top-scored avapritinib docking pose within homology-modeled human ABCB1 (Fig. 6A) as predicted by IFD simulation had a Glide gscore of -9.564 kcal/mol. The core structure of avapritinib was mainly stabilized into a hydrophobic cavity of the transmembrane domain (TMD) formed by residues Met68, Met69, Phe72, Tyr118, Phe303, Ile306, Tyr307, Tyr310, Phe336, Phe343, Phe728, Phe732, Tyr953, Phe957, Leu975, Phe978, and Phe983 (Fig. 6B). The ethanamine group of avapritinib was involved in hydrogen-bonding interaction with Gln725, Tyr310 and Phe728, as well as in π-cation interactions with Tyr310 and Phe728, respectively. Another hydrogen bond was observed between the 4’-nitrogen atom on the triazine ring of avapritinib and the hydroxyl group at the side chain of Ser929. Moreover, π-π stacking interaction was observed between the phenyl ring of avapritinib and the side chain of Phe303. Similarly, the predicted ABCG2-avapritinib complex showed that avapritinib was predominantly settled at a hydrophobic drug-binding pocket at the TMD of ABCG2 (Fig. 6C), with a Glide gscore of -12.089 kcal/mol. Like in its interaction with ABCB1, the ethanamine group, the triazine ring, and the phenyl ring structures of avapritinib all played critical roles in forming interactions with the nearby residues in ABCG2 (Fig. 6D). The ethanamine group of avapritinib formed hydrogen-bonding interactions with Phe432 (chain A) and Thr435 (chain A). The other predicted hydrogen bond was between the 2’-nitrogen atom on the triazine ring of avapritinib and Asn436 (chain B). The Phe432 residue at chain A of ABCG2 was also predicted to be involved in π-π stacking interaction with the phenyl ring of avapritinib.

Figure 6.

Best-scored binding poses of avapritinib to human ABCB1 homology model (A) and human ABCG2 model (C), predicted by IFD simulation. Homology modeled human ABCB1 and human ABCG2 are represented by surface colored by electrostatic potential. The avapritinib molecule is shown by CPK representation with the atoms colored as carbon-violet, hydrogen-white, oxygen-red, and nitrogen-blue. Only polar hydrogens of the ligand are shown. (B) The two-dimensional ligand-transporter interaction diagram of avapritinib and human ABCB1. The amino acids within 3Å are shown as colored bubbles, blue indicates polar residues, and green indicates hydrophobic residues. Grey circles indicate solvent exposure. Hydrogen bonds are shown by the purple arrow, and π-π stacking aromatic interactions are shown by green lines, and π-cation interactions are shown by red lines. (D) Two-dimensional ligand-transporter interaction diagram of avapritinib and human ABCG2, with the same color scheme as in panel (C).