Bafetinib (INNO-406) reverses multidrug resistance by inhibiting the efflux function of ABCB1 and ABCG2 transporters

ATP-Binding Cassette transporters are involved in the efflux of xenobiotic compounds and are responsible for decreasing drug accumulation in multidrug resistant (MDR) cells. Discovered by structure-based virtual screening algorithms, bafetinib, a Bcr-Abl/Lyn tyrosine kinase inhibitor, was found to have inhibitory effects on both ABCB1- and ABCG2-mediated MDR in this in-vitro investigation. Bafetinib significantly sensitized ABCB1 and ABCG2 overexpressing MDR cells to their anticancer substrates and increased the intracellular accumulation of anticancer drugs, particularly doxorubicin and [3H]-paclitaxel in ABCB1 overexpressing cells; mitoxantrone and [3H]-mitoxantrone in ABCG2 overexpressing cells, respectively. Bafetinib stimulated ABCB1 ATPase activities while inhibited ABCG2 ATPase activities. There were no significant changes in the expression level or the subcellular distribution of ABCB1 and ABCG2 in the cells exposed to 3 μM of bafetinib. Overall, our study indicated that bafetinib reversed ABCB1- and ABCG2-mediated MDR by blocking the drug efflux function of these transporters. These findings might be useful in developing combination therapy for MDR cancer treatment.

(QSAR) based in silico studies on ABCB1 and ABCG2 in recent years have successfully discovered common features for ABC transporter binding [15][16][17] . On the other hand, a large hydrophobic drug binding pocket of ABCB1 inside the TMD was previous illustrated through co-crystallization of mice ABCB1 and its bound inhibitor 18,19 . Several druggable sites on ABCG2 were also reported through mutational experiments 20,21 . A recent docking model reported by Klepsch and collaborators was able to predict ABCB1 inhibitor with accuracy of 76% 22 . Therefore, there is a potential of applying computational models to prescreen potential inhibitors among large inhibitors library. Though the crystal structure of human ABCB1 and ABCG2 is still lacking, with the help of the two previously established homology models available to us, we performed a structure-based prescreening of 2571 inhibitors from Selleck Chemicals on ABCB1 and ABCG2. Eight 'hits' from virtual screening were evaluated in vitro. From the presumptive screening results, bafetinib (INNO-406) was identified as a potential inhibitor simultaneously on both ABCB1 and ABCG2.
Bafetinib (INNO-406) is a novel tyrosine kinase inhibitor designed based on chemical structure of imatinib for treating Bcr-Abl+ leukemia 23,24 . It is a potent dual function Bcr-Abl kinase and Src family kinase Lyn inhibitor 23 . Previously it is reported that bafetinib is a substrate of ABCB1 by stating bafetinib is less effective in ABCB1-overexpressing K562/D1-9 cells 25 . However, there is no published data indicating the effect of bafetinib on solid tumor cell lines that do not have Bcr-Abl. Herein in this study, through our in silico screening, we reported for the first time that bafetinib would be used to augment the effect of chemotherapeutical agents in ABCB1-and ABCG2-overexpressed tumor cells. The purpose of this study was to demonstrate the MDR reversal effects of bafetinib and elucidate its potential mechanism.

Results
Bafetinib as a potential ABCB1 and ABCG2 inhibitor through virtual screening. Based on our previous knowledge that cell signaling inhibitors, such as tyrosine kinase inhibitors, could be possible ABC transporter inhibitors, we performed a virtual screening using inhibitor library of Selleck Chemicals provided by ZINC database. As shown in Fig. 1A, the 2571 ligands were prepared and docked into both human ABCB1 and ABCG2 homology models previously developed by our group. Seventy-nine matched ligands, whose SP docking results were in top range in both ABCB1 and ABCG2 model, were selected and re-docked into ABCB1 model using Glide XP method. The top 10 ligands from the XP run were dynasore, NVP-BHG712, flunarizine, fexofenadine, sirtinol, eprosartan, imatinib, baicalin, bafetinib and PRT062607. Among these, NVP-BHG712 and imatinib were excluded since their effects on ABCB1 and ABCG2 had been previously reported elsewhere. Flunarizine, baicalin and bafetinib were previously reported as ABCB1 substrates. Other inhibitors were purchased and examined for cytotoxicity. All inhibitors have IC 20 values larger than 3 μ M (Data not shown) therefore their reversal effects were tested at 3 μ M by MTT assay (Fig. 1B,C). Only bafetinib (Fig. 1D) showed significant reversal effects on both ABCB1-and ABCG2-mediated MDR, and was then used for further study. Bafetinib exhibited a Glide score of −11.8 kcal/mol on human ABCB1 model. As shown in Fig. 1E, bafetinib was predicted to bind into the large drug-binding cavity inside TMD of human ABCB1 through hydrophobic interaction with nearby residues. Also, two π -π stacking interactions were predicted between the two pyrimidine rings and phenyl ring of Phe 303. The NH linker formed a hydrogen bond with carbonyl group of Gln990 (-NH···OC-Gln990, 1.8 Å). The trifluoromethyl group formed a halogen bond with phenolic group of Tyr310 (-CF 3 ···HO-Tyr310, 2.4 Å). Per-residue interactions simulation predicted the importance of Tyr310, Phe336, Phe983, Met986, Ala987 and Gln990 as shown in Fig. 1E.
Similarly, bafetinib was found to be stabilized into TMD of human ABCG2 mainly through hydrophobic interactions with a Glide XP score of −9.6 kcal/mol. Moreover, a cation-π interaction was found between the pyrrolidine nitrogen atom and imidazole ring of His630. The N 1 atom of pyrimidine ring formed a hydrogen bond with phenolic group of Tyr570 (-N 1 ···HO-Tyr570, 2.1 Å). The NH linker formed another hydrogen bond with hydroxyl oxygen of Ser486 (-NH···OH-Ser486, 1.9 Å). Per-residue interaction simulation predicted the importance of Ser486, Phe489 and Leu581 as shown in Fig. 1F.
Bafetinib sensitized ABCB1-overexpressing cells to ABCB1 substrate chemotherapeutic drugs. A non-toxic concentration of bafetinib (3 μ M) was selected to evaluate its reversal effects (Supplementary Information). Doxorubicin-selected SW620/Ad300 cells had a much higher IC 50 value to doxorubicin as well as to another ABCB1-substrate paclitaxel than that in parental SW620 cells. Significant reduction in IC 50 values were observed followed by treatment of 1 μ M or 3 μ M bafetinib in SW620/Ad300 cells as shown in Table 1. Meanwhile, treatment of bafetinib did not significantly change the IC 50 values of doxorubicin and paclitaxel in parental SW620 cells. Similarly, significant decrease of IC 50 values of doxorubicin and paclitaxel were also observed in bafetinib-treated group of transfected HEK/ABCB1 cells as compared to non-treatment group. Cisplatin, which is not a substrate of ABCB1, was used as a negative control. Verapamil at 3 μ M was used as a positive control to evaluate the effects of bafetinib. Reversal effects of bafetinib at 3 μ M were comparable to that of 3 μ M of verapamil.
Bafetinib sensitized both wild-type and mutant ABCG2-overexpressing cells to ABCG2 substrate chemotherapeutic drugs. It has been previously reported that mutations at Arg482 in ABCG2 might alter the substrate and inhibitor specificity of ABCG2. Therefore, we examined both wild-type (R482) and mutant (R482G and R482T) ABCG2 in the present study. As shown in Table 2, bafetinib significantly decreased the IC 50 values of mitoxantrone and SN-38 in HEK/ABCG2-R482, HEK/ABCG2-R482G and HEK/ ABCG2-R482T cells in a concentration dependent pattern. In addition, the reversal effect was also analyzed in parental NCI-H460 and ABCG2-overexpressing NCI-H460/MX20 cells. A similar reversal effect was noticed in NCI-H460/MX20 cells as shown in Table 3. Effects of bafetinib at 3 μ M were found to be more potent than that of 3 μ M FTC, which is used as a positive control. Decreases in IC 50 values were also observed in NCI-H460 cells after treatment of 3 μ M bafetinib and FTC.
Bafetinib increased intracellular accumulation of doxorubicin or mitoxantrone in ABCB1-or ABCG2-overexpressing cells. In order to investigate the potential mechanism by which bafetinib sensitized ABCB1-or ABCG2-overexpressing cells, we determined the intracellular accumulation level of doxorubicin or mitoxantrone based on their own fluorescence. As shown in Fig. 2A,B, doxorubicin-associated fluorescence occurred mainly in the nuclei of SW620 and HEK293/pcDNA3.1 cells. The fluorescences of doxorubicin in resistant SW620/Ad300 and HEK/ABCB1 cells were weaker than that in parental cells while no significant nuclei accumulation was observed, several fluorescence spots were observed on the cell membranes. Pretreatment of the resistant cells with bafetinib increased the fluorescence intensity in resistant cells. The intracellular localization of doxorubicin in bafetinib-pretreated resistant cells was the same as that in the parental cells. Bafetinib had no significant effects on the intracellular accumulation or distribution of doxorubicin in the parental cells. Similarly as shown in Fig. 2C,D, the mitoxantrone-associated fluorescence was distributed in the cytoplasm and the nuclei of parental NCI-H460 and HEK293/pcDNA3.1 cells. In contrast, fluorescence signals showed that low level of mitoxantrone was accumulated in resistant NCI-H460/MX20 and HEK/ABCG2-R482 cells. Furthermore, a membrane distribution pattern was shown in HEK/ABCG2-R482 cells. Pretreatment of the resistance cells with bafetinib also significantly increased the fluorescence intensity in resistant cells, but not significant change the intracellular accumulation pattern in the parental cells.

Effects of bafetinib on the cellular accumulation of [ 3 H]-paclitaxel or [ 3 H]-mitoxantrone.
[ 3 H]-paclitaxel was used to further access and compare the quantitative accumulation of paclitaxel in parental and ABCB1-mediated MDR cells with or without inhibitor. As shown in Fig. 2E, intracellular [ 3 H]-paclitaxel level in SW620/Ad300 cells was approximately 50-fold lower than that in the SW620 cells after 2 h incubation.
Furthermore, as shown in Fig. 3E,F, untreated NCI-H460/MX20 cells lost 49% of the normalized intracellular Bafetinib did not change the expression level and the intracellular localization of ABCB1 or ABCG2 in cells overexpressing ABCB1 or ABCG2 respectively. ABCB1 and ABCG2 expression levels were demonstrated with Western blotting analysis. As previously reported, the monoclonal antibody C219 specifically detects a 170-kDa ABCB1 in SW620/Ad300 cells and the monoclonal antibody BXP-21 detects a 72-kDa ABCG2 in NCI-H460/MX20 cells 26 . Samples were taken at a 24 h interval after exposure to 3 μ M bafetinib. As shown in Fig. 4A,B, the treatment of 3 μ M bafetinib up to 72 hours did not significantly change the expression level of ABCB1 or ABCG2 in SW620/Ad300 or NCI-H460/MX20 cells respectively at any time points. Beta-actin at 42 kDa was used as loading control for Western blotting analysis.
As shown in Fig. 4C,D by immunofluorescence staining, ABCB1 transporters showed a membrane-located pattern in SW620/Ad300 and HEK/ABCB1 cells. The incubation of these cells with 3 μ M bafetinib did not significantly alter the subcellular distribution pattern of ABCB1 when compared to control at any time point at 24 h interval. ABCG2 transporters showed a similar membrane-located pattern in NCI-H460/MX20 cells (Fig. 4E) and HEK/ABCG2-R482 cells (Fig. 4F), while few transporters are located in the cytoplasm near nuclei. These distribution patterns are not altered by treatment of 3 μ M bafetinib. Bafetinib stimulated ATPase activity of ABCB1 and inhibited ATPase activity of ABCG2.
ABCB1-or ABCG2-mediated ATP hydrolysis in the presence of bafetinib at various concentrations from 0 to 40 μ M was measured to assess the effect of bafetinib on the ATPase activity of ABCB1 or ABCG2, respectively. Bafetinib stimulated the ATPase activity of ABCB1 in a concentration-dependent manner, with a maximal stimulation of 2-fold of the basal activity as shown in Fig. 5A,B. Interestingly, bafetinib inhibited the ATPase activity of ABCG2 in a concentration-dependent manner, with a maximal inhibition of 47% of the basal activity (Fig. 5C,D). Figure 5 demonstrates that the concentration of bafetinib required to obtain 50% of maximal stimulation or inhibition is 0.64 μ M or 2.97 μ M for ABCB1 or ABCG2 respectively.

Discussion
As previously reported, several small molecule cell signaling inhibitors, particularly tyrosine kinase inhibitors, are potent inhibitors of ABC transporters 3,11,27 . Therefore, it is of great value and interest to us to discover novel inhibitors of ABC transporter from these kinase inhibitors. As cancer cells may develop MDR via different types of ABC transporters, inhibitor for single ABC transporter may not be effective. For example colon cancer cell may become MDR via ABCB1 (SW620/Ad300 cells) or via ABCG2 (S1-M1-80 cells). Therefore, in this study, we aimed to find the 'hits' for both ABCB1 and ABCG2. With the 3D structures of human homology ABCB1 and ABCG2, we were able to perform a structure-based in silico screening to filter possible inhibitors through large ligand database. Our top scoring ligands screened through our virtual screening protocol included previously reported potent ABCB1 and ABCG2 inhibitors imatinib 28 and a weak inhibitor NVP-BHG712 29 . flunarizine 30 , and ABCB1 substrate baicalin 31 . These results indicated great potential and validity of our screening protocol. Among the candidates, bafetinib at its non-toxic concentration of 3 μ M was demonstrated as a potent inhibitor for both ABCB1 and ABCG2. Our MTT assays showed that doxorubicin-selected SW620/Ad300 cells also acquired resistance towards paclitaxel. This MDR cells had shown to be sensitized toward doxorubicin or paclitaxel by pre-incubation of bafetinib in a concentration-dependent manner. Beside the fact that bafetinib is reported as a potential ABCB1 substrate, the effects of bafetinib were clearly related to the ABCB1 overexpressed in SW620/Ad300 through our study for the following reasons. First, bafetinib, at its non-toxic concentration, did not alter the IC 50 values of doxorubicin or paclitaxel on parental SW620 cells, in which ABCB1 is not expressed 32,33 . Second, bafetinib at its non-toxic concentration did not sensitize SW620/Ad300 cells towards cisplatin, which is not a substrate of ABCB1. Last, bafetinib exhibited similar reversal effects on ABCB1-transfected HEK/ABCB1 cells, in which ABCB1 is the sole contributor for multidrug resistance. Therefore, bafetinib reversed MDR in SW620/Ad300 and HEK/ABCB1 cells by targeting at ABCB1 transporters.
Likewise, effect of bafetinib on ABCG2-overexpressing MDR cells NCI-H460/MX20 was related to the ABCG2 transporter. The decreases of IC 50 values in parental NCI-H460 cells may due to moderate level expression of ABCG2 as previously reported 26,34 . Moreover, site-mutation on target might produce significant differences in the effectiveness of inhibitor, such as T315I mutation will inactive bafetinib at Bcr-Abl 23 . Mutation at position 482 in ABCG2 has been previously demonstrated for affecting inhibitor efficacy such as novobiocin 35 . Our results showed that bafetinib could inhibit the activity of both wild-type (R482) and two mutant variant (R482G and R482T) of ABCG2. Also, our docking simulation of bafetinib at ABCG2 did not predict great contribution of Arg 482 in bafetinib binding. Furthermore, our study indicated a trend of correlation exists between the potency of doxorubicin or mitoxantrone and the intracellular accumulation of the compounds. Consistent with previous report, doxorubicin was mainly accumulated in the cellular nuclei in parental SW620 and HEK293/pcDNA3.1 cells 36 . The low level of doxorubicin accumulated in MDR cells may account for its higher IC 50 values. Also, light fluorescence detected doxorubicin near cellular membrane suggested that doxorubicin was pumped out of the cells. Bafetinib treatment significantly increased the intracellular doxorubicin accumulation, providing clues for the mechanisms of its reversal effects. The level of doxorubicin accumulation was further confirmed and quantified using another ABCB1 substrate [ 3 H]-paclitaxel. Moreover, by plotting the efflux time-course, it is clear that the resistance in MDR cells is on account of an accelerated efflux and bafetinib, as well as verapamil, acts by inhibiting the drug efflux. Similarly, mitoxantrone distributed in both cellular nuclei and cytoplasm in parental NCI-H460 and HEK293/pcDNA3.1 cells 37 . The membrane accumulated pattern of mitoxantrone in the MDR cells, especially transfected HEK/ABCG2-R482, suggested a process of drug efflux. Treatment of bafetinib in MDR cells was able to reverse the intracellular mitoxantrone accumulation to the level which is similar within the parental cells. Therefore, we concluded that bafetinib reversed the ABCB1-or ABCG2-mediated MDR by inhibiting the transported mediated drug efflux, thereby increasing the intracellular drug accumulation.
Connections between ABC transporters and other cell signaling transducer proteins, such as PKA and PKC, have been previously reported 38 . As membrane-bound transporter, ABCB1 or ABCG2 might lose their function by being down-regulated or translocated into the cytoplasm upon treatment, which could have potentiated the reversal effects of bafetinib. These hypotheses were examined by our Western blotting and immunofluorescence studies. No change in protein expression level or in subcellular location was observed when ABCB1-or ABCG2-overexpressing cells were treated with bafetinib at 3 μ M for 24, 48 and 72 h, respectively. Overall, the reversal effects of bafetinib on ABCB1 or ABCG2 is therefore related to its direct inhibition of function of these ABC transporters. It was hypothesized that bafetinib could interact with the ATP-binding pocket in NBD of ABC transporters, similar to its binding to the ATP-binding pocket on the BCR-Abl kinase. Although inhibitory ATPase activity on ABCG2 as increasing concentration of bafetinib was observed, our docking simulation predicted low score of bafetinib at ABCG2 ATP binding site (− 5.660 kcal/mol) 39 . On the other hand, increased ATPase activity on ABCB1 upon increasing concentration of bafetinib suggested that bafetinib might competitively inhibit ABCB1 by binding to the drug-binding pocket in TMD. Since NBDs of ABCB1 and ABCG2 exhibited high levels of sequence homology and possessed the similar conserved secondary structure topology as previous reported 40 , it is believed that bafetinib bound and stabilized with TMD of ABCG2 instead of being transported. Therefore, bafetinib may serve as an actual inhibitor of ABCG2 while as an competitive substrate of ABCB1. The difference in TMDs of ABCB1 and ABCG2 accounts for their difference in substrates recognition as well as their responses to bafetinib binding.
In conclusion, this study report that bafetinib reverse ABCB1-and ABCG2-mediated MDR by directly inhibiting their efflux functions. These results tentatively suggest that bafetinib could be used in patients with MDR to augment the clinical response to chemotherapeutic agents that are substrates of ABCB1 and ABCG2.
Virtual screening and molecular docking. Selleck chemicals ligand 2D structures database was downloaded from ZINC 41 (San Francisco, CA) and were prepared using our previous molecular modeling protocols 33 . The human ABCB1 homology model based on refined mouse ABCB1 (RCSB ID: 4M1M) was kindly provided by S. Aller and the grid of 25 Å was refined as previously described 19 . Our previous grid on human homology ABCG2 on centroid of Arg482 was used for docking on ABCG2 42 . Glide HTVS (high throughput virtual screening) docking protocol was followed to dock all prepared 5076 structures into ABCB1 and ABCG2 (Schrödinger, LLC, New York, NY). Top scoring (Glide score) 10% ligands of ABCB1 or ABCG2 were selected respectively and were subjected to Glide SP (standard precision) docking. Ligands with Glide SP score more than −8.0 kcal/mol or −7.0 kcal/mol were kept respectively for ABCB1 and ABCG2. 79 ligands were selected as overlaps between ABCB1 and ABCG2 SP docking results and then submitted to Glide XP (extra precision) docking (Schrödinger, LLC, New York, NY). The top 10 results were examined visually and 8 compounds were acquired for reversal MTT assays. All computations were carried out on a 6-core Intel Xeon Processor with a Mac OS. A schematic view of virtual screening processes was shown in Fig. 1A.
Scientific RepoRts | 6:25694 | DOI: 10.1038/srep25694 Cytotoxicity by MTT assay. Modified MTT colorimetric assay was used to detect the cytotoxicity of anticancer drugs with or without modulator agents. 5 × 10 3 cells were seeded evenly into each well in 96-well microplates and cultured overnight. Bafetinib and parallel control modulators were added 1 h prior. Chemotherapeutic drugs were then added into the designated wells by concentration gradient. After 72 h of incubation, 20 μ L of MTT solution (4 mg/mL) was added into each well, then the microplate was further incubated for 4 h. Subsequently, the medium was aspirated and 100 μ L of DMSO was added to dissolve the formazan crystal in each well. The absorbance was determined at 570 nm by the OPSYS microplate reader (Dynex Technology, Chantilly, VA). The IC 50 values were calculated from the survival curves using modified Bliss method. Verapamil was used as a positive control for ABCB1 overexpressing cell lines. FTC was used as a positive control for ABCG2 overexpressing cell lines.
Doxorubicin or mitoxantrone intracellular accumulation and effects of bafetinib. Cells were seeded on sterile coated coverslips and were allowed to grow overnight. On the following day, cells were washed with phosphate-buffered saline (PBS), incubated with or without 3 μ M of Bafetinib for 1 h before being treated with 5 μ M doxorubicin or mitoxantrone for 2 h, and then examined using fluorescence microscopy. SW620, HEK293/pcDNA3.1, SW620/Ad300, and HEK/ABCB1 cells received doxorubicin treatment. NCI-H460, HEK293/pcDNA3.1, NCI-H460/MX20 and HEK/ABCG2-R482 were tested for mitoxantrone accumulation.  . Serial aliquots at 0, 30, 60 and 120 min in efflux phase were taken and analyzed as previously described 33 . To study the effect of bafetinib on the time course of [ 3 H]-paclitaxel efflux, cells were pre-exposed to 1 μ M or 3 μ M bafetinib for 1 hour before being treated with [ 3 H]-paclitaxel containing medium. Then bafetinib and paralled control inhibitor were also added to each phase, respectively. Same protocol was followed by using 0.01 μ M [ 3 H]-mitoxantrone to study ABCG2-mediated drug efflux.
Preparation of total cell lysates and Western blotting. Cell extracts were prepared by incubating on ice with lysis buffer (10 mM Tris, 1 mM EDTA, 0.1% SDS, 150 mM NaCl, 1% Triton-X and protease inhibitor cocktail) for 20 min, followed by centrifugation at 12,000 × g at 4 °C for 5 min. The supernatant containing total cell lysate was collected and protein concentration was determined by bicinchoninic acid (BCA TM ) based protein assay (Thermo Scientific, Rockford, IL). Equal amount of total cell lysates were loaded and separated by SDS-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane. For ABCB1-overexpressing cells, the blot was then probed with the primary antibody C219 (dilution 1:200) and primary antibody BA3R (dilution 1:500). For ABCG2-overexpressing cells, the blot was probed with primary antibody BXP-21 (dilution 1:500) and primary antibody BA3R (dilution 1:500). Membranes were then further incubated with HRP-conjugated secondary antibody. The signal was detected using enhanced chemiluminescence and exposure of X-ray film. ABCB1 and ABCG2 ATPase assays. The ABCB1-and ABCG2-associated ATPase activities were measured by PREDEASY ATPase Kits with modified protocols. Basically, the ABCB1-or ABCG2-membranes were thawed and diluted. Sodium orthovanadate (Na 3 VO 4 ) was used as an ATPase inhibitor. Various concentrations of bafetinib were incubated with membranes for 5 min, then the ATPase reactions were initiated by adding 5 mM Mg 2+ ATP. After incubation at 37 for 40 min with brief mixing, luminescence signals of Pi were initiated and measured. The changes of relative light units were determined by comparing Na 3 VO 4 -treated samples with bafetinib-treated groups.
Immunofluorescence of ABCB1 or ABCG2. Cells were seeded on sterile coated coverslips and were allowed to grow overnight, followed by treatment with 3 μ M bafetinib for 24, 48 and 72 h respectively. The cells were then washed with ice-cold PBS, fixed in 4% paraformaldehyde, permeatabilized by 0.1% Triton X-100 and then blocked by 6% BSA for 1 h. Subsequently, cells were incubated with either monoclonal antibody P7965 against ABCB1 (1:400, for SW620/Ad300 and HEK/ABCB1 cells) or monoclonal antibody BXP-21 against ABCG2 (1:400, for NCI-H460/MX20 and HEK/ABCG2-R482) overnight, followed by Alexa Fluor 488 conjugated secondary antibody (1:2000) for 1 h. Propidium iodide/RNase solution was applied to counterstain the nuclei. Cells were mounted onto glass slides and then examined under fluorescence microscopy.
Fluorescence microscopy. Images were collected using a Nikon TE-2000S fluorescence microscope with an APO 60× oil immersion lens (Nikon). Propidium iodide, mitoxantrone and doxorubicin fluorescence were excited and collected with Nikon TRITC HYQ excitation filter combination. The fluorescence of mitoxantrone was represented using pseudo color (blue) to represent the physical blue color of mitoxantrone liquid. Alexa Fluor Scientific RepoRts | 6:25694 | DOI: 10.1038/srep25694 488 fluorescence was excited and collected with Nikon B1-E filter combinations. The same fluorescence settings (excitation, lamp power, detector gain and exposure time) were used to image different samples.
Statistics. All data are expressed as mean ± S.D. from three or more experiments and statistically evaluated by one-way ANOVA. Differences were considered significant when p < 0.05.