Hsp90 inhibitor geldanamycin attenuates the cytotoxicity of sunitinib in cardiomyocytes via inhibition of the autophagy pathway

Abstract
Sunitinib malate (sunitinib) is an orally available, multitargeted tyrosine kinase inhibitor with antitumor and antiangiogenic activities. Although sunitinib is effective for the treatment of patients with gastrointestinal stromal tumor, advanced renal cell carcinoma, or pancreatic neuroendocrine tumor, adverse cardiac events associated with sunitinib administration have been reported. Here, we examined the effect of geldanamycin, an inhibitor of heat shock protein (Hsp) 90, on sunitinib-induced cytotoxicity in cardiomyocytes. First, we found that treatment with geldanamycin or other Hsp90 inhibitors (tanespimycin, ganetespib, or BIIB021) significantly attenuated sunitinib- induced cytotoxicity in rat H9c2 cardiomyocytes, suggesting a drug-class effect of Hsp90 inhibitors. We then examined the mechanisms underlying sunitinib- induced cytotoxicity and found that sunitinib induced autophagy in H9c2 cells and that pretreatment with geldanamycin inhibited the induction of autophagy by promoting degradation of the autophagy-related proteins Atg7, Beclin-1, and ULK1. Pharmacological assessment with autophagy inhibitors confirmed that geldanamycin attenuated the cytotoxicity of sunitinib by interfering with autophagy. In addition, we found that the molecular chaperone Hsp70, which is induced by geldanamycin, was not involved in the attenuation of sunitinib- induced cytotoxicity. Finally, to provide more clinically relevant data, we confirmed that geldanamycin attenuated sunitinib- induced cytotoxicity in human induced pluripotent stem cell-derived cardiomyocytes. Together, these data suggest that geldanamycin attenuates sunitinib- induced cytotoxicity in cardiomyocytes by inhibiting the autophagy pathway. Thus, the further investigation of combination or sequential treatment with an Hsp90 inhibitor and sunitinib is warranted as a potential strategy of attenuating the cardiotoxicity associated with sunitinib administration in the clinical setting.

1.Introduction
Combination or sequential treatment with multiple molecular-targeted agents is a useful strategy for enhancing the efficacy and safety of cancer therapies when monotherapy alone is insufficient to inhibit the growth or survival of cancer cells, or manage adverse reactions (Li et al., 2014; Zhang et al., 2016). For example, combination or sequential treatment can enhance antitumor activity by simultaneously inhibiting different cell signaling pathways with multiple agents, and it can overcome drug resistance by acting on cancer cells or their microenvironment via multiple modes of action. Also combination treatment can reduce drug toxicity without reducing drug efficacy, because of enhanced combination activity even with lower drug dosages compared with each monotherapy, and drugs with desirable cytoprotective effects can be added to the regimen (Sysa-Shah et al., 2014).
Sunitinib malate (sunitinib) is an orally available, multitargeted tyrosine kinase inhibitor that has antitumor and antiangiogenic activities via inhibition of vascular endothelial growth factor receptors 1–3, platelet-derived growth factor receptor α/β, mast/stem cell growth factor receptor Kit, colony-stimulating factor 1 receptor, receptor-type tyrosine-protein kinase FLT3, and proto-oncogene tyrosine-protein kinase receptor Ret (Faivre et al., 2007). Sunitinib is approved for the treatment of gastrointestinal stromal tumor, advanced renal cell carcinoma, and pancreatic neuroendocrine tumor; however, high incidences of serious cardiovascular adverse events such as hypertension, left ventricular systolic dysfunction, and congestive heart failure have been reported in the clinical setting (Demetri et al., 2006; Chu et al., 2007; Motzer et al., 2007). Therefore, additional clinical management such as careful cardiac monitoring, dose reduction, and drug intervention to prevent or treat adverse events is often required for patients undergoing sunitinib treatment. One potential means of attenuating the cardiotoxicity of sunitinib could be combination or sequential treatment with a cardioprotective agent that antagonizes the molecular mechanism underlying the cardiotoxicity caused by sunitinib. Although there are many hypotheses about the mechanisms of sunitinib- induced cardiotoxicity (Kerkela et al., 2009; Hasinoff, 2010; Zhao et al., 2010), the precise cardiotoxicity mechanism remains unclear.

Heat shock protein (Hsp) 90 inhibitors are a promising class of molecular-targeted agents for the treatment of cancers (Garcia-Carbonero et al., 2013). Hsp90 inhibitors, such as the naturally occurring benzoquinone ansamycin antibiotic geldanamycin (DeBoer et al., 1970), inhibit the molecular chaperone function of Hsp90 (Stebbins et al., 1997; Ying et al., 2012), which induces dissociation of client proteins, including oncogenic proteins, from Hsp90, resulting in their degradation via the ubiquitin-proteasome pathway. In addition to their antitumor activity, Hsp90 inhibitors may also protect cardiac cells from certain types of damage such as ischemia and cardiac hypertrophy on the basis of their mode of action (Morris et al., 1996; Lee et al., 2010). The unique drug action of Hsp90 inhibitors makes them particularly attractive for use in combination or sequential regimens with anticancer agents such as tyrosine kinase inhibitors, which disrupt cellular processes such as proliferation, differentiation, migration, invasion, and survival by preventing phosphorylation of tyrosine kinases (Vaishampayan et al., 2010; Courtin et al., 2016). Combination or sequential treatment with sunitinib and an Hsp90 inhibitor may enhance antitumor activity through inhibition of cancer-related signaling pathways by different mechanisms of action (Floris et al., 2011). However, there is no report to examine the cytoprotective effect of Hsp90 inhibitors on sunitinib- induced cytotoxicity in cardiac cells.Here, we examined the effect of geldanamycin on sunitinib- induced cardiotoxicity in rat embryonic ventricle-derived H9c2 cells. H9c2 cells are a commonly used model of human cardiomyocytes because they have cardiac and skeletal muscle ion channels and express metabolic enzymes similar to those of human cardiomyocytes (Will et al., 2008; Lou et al., 2015). We found that geldanamycin attenuated sunitinib- induced cytotoxicity in H9c2 cells by disrupting the autophagy pathway. We also found that Hsp70, a molecular chaperone induced by geldanamycin, did not contribute to the cytoprotective effect of geldanamycin. To provide more clinically relevant data, we confirmed that geldanamycin also attenuated sunitinib- induced cytotoxicity in human induced pluripotent stem cell (hiPSC)-derived cardiomyocytes.

2.Materials and Methods
Sunitinib, tanespimycin, ganetespib, and BIIB021 were purchased from Selleck Chemicals (Houston, TX). Geldanamycin was purchased from Tokyo Chemical Industry (Tokyo, Japan). All of the compounds listed above were dissolved in dimethyl sulfoxide (DMSO) to make stock solutions. 3-Methyladenine (3-MA) was purchased from Merck Millipore (Darmstadt, Germany). Bafilomycin A1 was purchased from Santa Cruz Biotechnology ( Dallas, TX). Chloroquine diphosphate (CQ) and ammonium chloride (NH4Cl) were purchased from Wako Pure Chemical Industries (Osaka, Japan).H9c2 rat cardiomyocytes were obtained from the American Type Culture Collection (Manassas, VA). H9c2 cells were cultured in Dulbecco’s Modified Eagle’s Medium (Wako Pure Chemical Industries) supplemented with 10% fetal bovine serum (Sigma-Aldrich, St. Louis, MO) and 1% penicillin- streptomycin (Wako Pure Chemical Industries) at 37 °C under a 5% CO2 atmosphere.H9c2 cells stably overexpressing Hsp70 were established by using the PiggyBac Transposon System (System Biosciences, Palo Alto, CA). Briefly, H9c2 cells were seeded on a 6-well plate at a density of approximately 2.0 × 105 cells/well one day before plasmid DNA transfection. Transfection was performed in Opti-MEM Reduced Serum Medium (Thermo Fisher Scientific, Waltham, MA) by using Lipofectamine 3000 Reagent (Thermo Fisher Scientific) in accordance with the manufacturer’s protocol. After 24 h incubation, the medium was replaced with Dulbecco’s Modified Eagle’s Medium and puromycin-resistant H9c2 cells were selected in thepresence of puromycin (3 µg/mL; Thermo Fisher Scientific).hiPSC-derived cardiomyocytes (iCell Cardiomyocytes) were obtained from Cellular Dynamics International (Madison, WI).

Cells were pre-cultured on gelatin (Sigma-Aldrich)-coated 6-well plates in iCell Cardiomyocytes Plating Medium (Cellular Dynamics International) at a density of approximately 2.0 × 106 cells/well and incubated for 48 h at 37 °C under a 5% CO2 atmosphere in accordance with the manufacturer’s protocol. The medium was replaced with iCell Cardiomyocytes Maintenance Medium (Cellular Dynamics International); afterwards, half the volume of medium was replaced every 2 days. At 7 days after plating, hiPSC-derived cardiomyocytes were detached by using 0.25% trypsin-EDTA (Sigma-Aldrich) and re-plated on fibronectin (Thermo Fisher Scientific)-coated 24- or 96-well plates in iCell Cardiomyocytes Maintenance Medium at a density of approximately 1.2 × 105 or 1.8 × 104 cells/well, respectively. Half the volume of medium was replaced every 2 days and the cells were maintained at 37 °C under a 5% CO2 atmosphere until assay at day 7 after re-plating.The PiggyBac Transposon vector pPBcmv- mcs was constructed from the PB-CMV-MCS-EF1-Puro Vector PB510B-1 (System Biosciences). An EcoRI-SalI fragment (1.2 kb), which contains the EF1 promoter of the puromycin resistance gene, was removed. A sequence encoding rat heat shock 70 kDa protein 1A (Hspa1a) (NCBI Reference Sequence: NM_031971) was synthesized by the GeneArt Gene Synthesis service (Thermo Fisher Scientific). An IAG2AP DNA fragment containing the internal ribosome entry site sequence followed by the hmAG1- T2A-Puromycin resistance gene was constructed by using the In-Fusion cloning method (Takara Bio, Shiga, Japan) (Zhu et al., 2007). pPBcmv-rHsp70-IAG2AP wasconstructed by connecting each DNA fragment by using the In-Fusion cloning method.H9c2 cells and hiPSC-derived cardiomyocytes were treated with sunitinib, Hsp90 inhibitors (geldanamycin, tanespimycin, ganetespib, or BIIB021), or autophagy inhibitors (3-MA, bafilomycin A1, NH4Cl, or CQ) as follows.

For Hsp90 inhibitor treatment, the culture medium was replaced with medium containing 0.5 μM of one of the Hsp90 inhibitors, and the cells were then incubated for 6 h. For autophagy inhibitor treatment, the culture medium was replaced with medium containing 20 mM 3-MA, 200 nM bafilomycin A1, 20 mM NH4Cl, or 10 μM CQ, whichwere concentrations determined from previous publications (Ha et al., 2012; Zhang et al., 2012;Liu et al., 2015) and in- house preliminary studies to be sufficient to inhibit autophagy in H9c2cells, and the cells were then incubated for 1 h. For autophagic flux analysis with CQ treatment, the culture medium was replaced with medium containing 10 μM CQ, and the cells were then incubated for 6 h. For sunitinib treatment, the culture medium was replaced with medium containing sunitinib (1–20 μM) with or without Hsp90 inhibitors or autophagy inhibitors, and the cells were then incubated for 6 h. Geldanamycin co-treatment was conducted by pretreating the cells with geldanamycin for 6 h followed by treatment with geldanamycin plus sunitinib treatment for 6 h. In all experiments, the final DMSO concentration was below 0.1% (v/v).Transfection of H9c2 cells with rat siRNA targeting Atg7, Beclin-1, or ULK1 was performedby using ON-TARGETplus siRNA (GE Healthcare Bio-Sciences, Pittsburgh, PA) in accordancewith the manufacturer’s protocol. Briefly, H9c2 cells were seeded in a T-75 flask atModified Eagle’s Medium for the following experiments.Cells grown in 96-well plates or 35 mm dishes to approximately 80% confluency were transferred to an incubator preset at 43 °C under a 5% CO2 atmosphere for 1 h, and then were incubated at 37 °C under a 5% CO2 atmosphere for 3–24 h.The cytotoxicity of compounds was evaluated by determination of total intracellular ATP content or the concentration of the cytosolic enzyme lactate dehydrogenase (LDH) in the culture medium (Adderley and Fitzgerald, 1999; Hasinoff et al., 2008; Cohen et al., 2011).

In this study, both methods were performed in accordance with the manufacturers’ protocols. Briefly, total intracellular ATP content was quantified by using a CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI) and an EnVision Multilabel Plate Reader (Perkin Elmer, Norwalk, CT), and cell viability (%) was defined as the percent ATP content compared with that in DMSO-treated control cells. The amount of LDH in the culture medium was determined by using an LDH Cytotoxicity Detection Kit (Takara Bio) and a SpectraMax 190 Microplate reader (Molecular Devices, Sunnyvale, CA) with the bundled SoftMax Pro software (version 4.8,Molecular Devices) or an EnVision Multilabel Plate Reader at a wavelength of 490 nm; reference measurements were made at a wavelength of 650 nm. Loss of membrane integrity (%) was defined as the percent LDH in the culture medium compared with the maximum releasable amount of LDH in the cells.Viability of geldanamycin-treated cells was determined using a Cell Counting Kit-8, which estimates the number of living cells by measuring the activities of dehydrogenases in cells (Dojindo, Kumamoto, Japan). Briefly, cells (2.5–3.0 × 103 cells/well) were cultured in 96-well plates for 48 h and were then incubated with geldanamycin (0.01–30 μM) for 24 h. The solution of the Cell Counting K it-8 was added to each well in accordance with the manufacturer’s protocol and absorbance was measured with an EnVision Multilabel Plate Reader at a wavelength of 450 nm; reference measurements were taken at a wavelength of 650 nm.After drug, siRNA, or heat-shock treatment, cells were washed with Dulbecco’s Phosphate-Buffered Saline (DPBS) and lysed with radioimmunoprecipitation assay buffer (Thermo Fisher Scientific) containing Protease Inhibitor Cocktail (Roche Diagnostics, Mannheim, Germany). The protein concentration of each lysate was determined by using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). After sample preparation for sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) by using 4× Laemmli Sample Buffer (Bio-Rad Laboratories, Hercules, CA) containing 10% 2- mercaptoethanol (Wako Pure Chemical Industries), cell lysate was loaded onto a 4% to 20% gradient polyacrylamide gel(Bio-Rad Laboratories) or a 7.5% polyacrylamide gel (ATTO Corporation, Tokyo, Japan), and SDS-PAGE was performed with a Tris/Glycine/SDS running buffer (Bio-Rad Laboratories).

Proteins were then transferred onto a polyvinylidene difluoride membrane. The membrane was blocked in 5% skim milk (Wako Pure Chemical Industries) dissolved in Tris-buffered saline (Takara Bio) with 0.1% Tween 20 (TBS-T) for 1 h at room temperature, and incubated with primary antibodies overnight at 4 °C. Following three TBS-T washes, the membrane was incubated with horseradish peroxidase-conjugated secondary antibody to detect the primary antibody for 1 h at room temperature. Secondary antibody was visualized by using ECL Prime or ECL Select Western Blotting Detection Reagent (GE Healthcare Bio-Sciences), and signals were captured with an ImageQuant LAS-4000 luminescent image analyzer (Fujifilm, Tokyo, Japan) and the bundled ImageQ uant TL software (version 8.1.0.0; GE Healthcare Bio-Sciences). The density of protein bands was quantified by using ImageQuant TL software. Relative fold-changes of the density of the protein bands were calculated as reported previously (Chung et al., 2015). Briefly, the protein bands of interest were normalized using β-actin as the loading control, and the ratio of each protein band to DMSO control was calculated.The primary antibodies used were Hsp70/Hsp72 antibody (Enzo Life Sciences, Farmingdale, NY), anti- Atg1/ULK1 antibody (Sigma-Aldrich), LC3B antibody, Atg7 antibody, Beclin-1 antibody, Hsp90 antibody, and β-actin antibody (Cell Signaling Technology, Danvers, MA). The secondary antibodies used were horseradish peroxidase-conjugated anti- mouse IgG secondary antibody and horseradish peroxidase-conjugated anti-rabbit IgG secondary antibody (Cell Signaling Technology).Under normal cell culture conditions, microtubule-associated protein light chain 3 (LC3) mostly exists as a cytosolic form (LC3-I). Following the induction of autophagy, some of the LC3 is conjugated to phosphatidylethanolamine, and lipidated LC3 (LC3-II) is then recruited to the autophagosomal membrane. Conversion of LC3-I to LC3-II was detected by western blotting with anti- LC3B antibody as previously reported (Mizushima and Yoshimori, 2014). The densities of the LC3-II protein bands were normalized to those of β-actin, and the ratio of the density of each protein band to that of DMSO control was calculated and used as an index of autophagicactivity, because the amount of LC3-II is well correlated with the number of autophagosomes.Autophagic flux was evaluated by measuring the amount of LC3-II in the presence or absenceautophagic degradation.

After drug treatment, total RNA from H9c2 cells was prepared by using an RNeasy Mini Kit (Qiagen, Hilden, Germany) in accordance with the manufacturer ’s protocol. The amount and purity of the total RNA isolated was determined in accordance with a previous report (Uesugi et al., 2014). Briefly, a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific) was used to measure absorbance at wavelengths of 280 nm and 260 nm; an A260:A280 ratio of approximately 2.0 was considered to indicate a pure RNA sample. In addition, a 2100Bioanalyzer and RNA Nano LabChips (Agilent Technologies, Santa Clara, CA) were used to obtain the RNA integrity number from the pattern of total RNA electrophoresed as a metric of RNA degradation (Schroeder et al., 2006). An RNA integrity number of 9.4 to 9.6 was considered to indicate pure total RNA.Gene expression levels were determined by DNA microarray analysis. Total RNA (100 ng) was converted to cyanine-3-labeled complementary RNA (cRNA) by using a Low Input Quick Amp Labeling Kit, One-Color (Agilent Technologies); the cRNA was hybridized to a SurePrint G3 Rat Gene Expression v2 8×60K Microarray (Agilent Technologies). Hybridization images were scanned with a DNA Microarray Scanner (Agilent Technologies) and quantified by using Feature Extraction software (version 11, Agilent Technologies). Raw data were normalized by using the quantile function in R software (version 3.1.0). P-values for all statistical tests were determined by using Welch’s t-test, followed by adjustment for multiple comparisons by using a false discovery rate approach (Benjamini- Hochberg procedure) in R software (version 3.1.0). A gene was considered to be differentially expressed when its absolute fold-change relative to the control value was ≥2.0 and it had a false discovery rate P-value of <0.05.

Pathway analysis was conducted by using Ingenuity Pathway Analysis (Qiagen).H9c2 cells (approximately 6.0 × 104 cells/well) were cultured in 24-well glass-bottom culture plates (AGC Techno Glass, Shizuoka, Japan) for 24 h. After drug treatment, cells were washed twice with DPBS and fixed in 4% paraformaldehyde (Wako Pure Chemical Industries) for 30 min and permeabilized with DPBS containing 0.2% Triton X-100 (Wako Pure Chemical Industries) for 5 min at room temperature. Cells were treated with Block Ace (DS PharmaBiomedical, Osaka, Japan) in DPBS (blocking buffer) for 1 h and probed overnight at 4 °C with anti-LC3 antibody (MBL, Aichi, Japan) in blocking buffer. After three 5 min washes with DPBS containing 0.05% Tween 20 (PBS-T), cells were incubated for 1 h at room temperature with Alexa Fluor 555 Donkey Anti- Rabbit IgG (H+L) (Thermo Fisher Scientific) and Hoechst 33342 (1 µg/mL; Sigma-Aldrich) in blocking buffer. The cells were then washed three times with PBS-T and rinsed with DPBS. Stained cells were imaged under a confocal microscope (TCS SP8; Leica Microsystems, Wetzlar, Germany).All data are expressed as means ± standard error of the mean (SEM); n is the number ofindependent experiments conducted. GraphPad Prism 7 software (version 7.02; GraphPad Software, San Diego, CA) was used for statistical analysis. Statistical differences between two groups were determined by using the F-test followed by the two-tailed unpaired Student’s t-test (P ≥ 0.05 in F-test) or Welch’s t-test (P < 0.05 in F-test). A P- value less than 0.05 was considered statistically significant.

3.Results
To evaluate the toxicity of sunitinib in cardiomyocytes, H9c2 cells were exposed to sunitinib for 6 h and the total amount of intracellular ATP and the amount of LDH in the culture medium were quantified (Fig. 1A). In the ATP assay, sunitinib decreased cell viability in a concentration-dependent manner at concentrations of 1 to 20 μM; cell viability was decreased by over 80% when the cells were exposed to 20 μM sunitinib. In the LDH assay, exposure to 15 or 20 μM sunitinib decreased membrane integrity by more than 50%.Next, to assess the protective effect of geldanamycin against sunitinib-induced cytotoxicity, we exposed H9c2 cells to sunitinib with geldanamycin pretreatment or co-treatment and then determined the total intracellular ATP content (Fig. 1B). Irrespective of the amount of sunitinib the cells were exposed to (range, 5–20 μM), 0.5 μM geldanamycin pretreatment or co-treatment significantly attenuated the sunitinib- induced decrease in cell viability compared with DMSO control. In particular, pretreatment or co-treatment with geldanamycin was remarkably effective at attenuating the decrease in cell viability induced by 20 μM sunitinib (cell viability: DMSO control, 15.78% ± 2.86%; geldanamycin pretreatment, 70.41% ± 2.74%, P < 0.001; geldanamycin co-treatment, 73.63% ± 1.97%, P < 0.001). We also confirmed that exposure to geldanamycin alone did not affect cell viability in H9c2 cells at concentrations of 0.01 to 10 µM, as assessed by a viable cell counting assay, and did not affect the total amount of intracellular ATP in the experimental conditions. (Figs. S1, S2).To investigate whether the attenuation of sunitinib- induced cytotoxicity is a class effect among Hsp90 inhibitors, we examined the cytotoxicity of sunitinib in cells pretreated with tanespimycin, ganetespib, and BIIB021. We found that these Hsp90 inhibitors also significantly attenuatedsunitinib- induced cytotoxicity in H9c2 cells compared with DMSO control (Fig. 1C).

These results suggest that the attenuation of sunitinib- induced cytotoxicity in H9c2 cells is a class effect of Hsp90 inhibitors.It is reported that autophagy likely plays an important role in sunitinib- induced cytotoxicity in H9c2 cells (Zhao et al., 2010). Therefore, to determine whether autophagy was induced under our experimental conditions, we used immunocytochemistry methods to examine subcellular localization of LC3 as a marker of autophagosomes (Fig. 2A). After 6 h sunitinib treatment, LC3 puncta, the presence of which indicates autophagic activity, were clearly detected within the H9c2 cells. In contrast, in DMSO control cells, LC3 was diffusely localized throughout the cytoplasm.To confirm induction of autophagy by sunitinib, we performed a western blotting based autophagic flux assay. Exposure of H9c2 cells to CQ or sunitinib alone significantly increased the amount of LC3-II compared with DMSO control (Fig. 2B). Moreover, co-treatment with CQ and sunitinib further increased the amount of LC3-II compared with that in H9c2 cells exposed to CQ or sunitinib alone. Together, these results indicate that sunitinib induces autophagic flux and it is likely to be involved in the cytotoxicity of sunitinib in H9c2 cells.Geldanamycin is reported to block autophagy by promoting degradation of the autophagy-related protein Beclin-1 (Xu et al., 2011). Therefore, we hypothesized thatpretreatment with geldanamycin attenuates sunitinib- induced cytotoxicity in H9c2 cells by suppressing sunitinib-induced autophagy. To test this hypothesis, first we exposed H9c2 cells to sunitinib with or without geldanamycin pretreatment and assessed immunocytochemically if the development of LC3 puncta was affected by geldanamycin (Fig. 3A). Although geldanamycin pretreatment alone did not affect the intracellular localization of LC3, it clearly inhibited sunitinib- induced production of LC3 puncta.Next, we evaluated the effect of geldanamycin on sunitinib- induced autophagy by examining the conversion of LC3-I to LC3-II in cells treated with sunitinib with or without geldanamycin pretreatment by western blotting. Compared with DMSO treatment alone, geldanamycin pretreatment slightly but significantly decreased the amount of LC3-II (0.84 ± 0.01, P < 0.001).

Geldanamycin pretreatment also significantly attenuated the sunitinib- induced increase of LC3-II compared with DMSO pretreatment (sunitinib 10 μM, P < 0.001; sunitinib 15 μM, P < 0.001; sunitinib 20 μM, P < 0.01) (Fig. 3B).To investigate how geldanamycin inhibited autophagy, we performed western blotting to examine the expression levels of Atg7, Beclin-1, and ULK1, which are key regulators of the induction of autophagy. Geldanamycin significantly decreased the expression levels of Atg7, Beclin-1, and ULK1 compared with DMSO control (Fig. 3C). The relative fold-change in the expression of Atg7 was 0.77 ± 0.02 (P < 0.001), that of Beclin-1 was 0.75 ± 0.03 (P < 0.01), and that of ULK was 0.44 ± 0.04 (P < 0.001). These results suggest that geldanamycin inhibited sunitinib- induced autophagy by promoting degradation of autophagy-related proteins.To confirm that the protective effect of geldanamycin against sunitinib- induced cytotoxicitydepended on autophagy inhibition, we examined the effect of the autophagy inhibitors 3-MA, bafilomycin A1, and NH4Cl, each of which inhibits autophagy via a different mechanism, on sunitinib- induced cytotoxicity. All three inhibitors inhibited autophagy in H9c2 cells under basal conditions (Fig. 4A). Viability of cells exposed to sunitinib was higher when they were also treated with autophagy inhibitors compared with DMSO control (Fig. 4B). Among these inhibitors, 3-MA had the greatest protective activity, in particular in cells exposed to 20 μMsunitinib (cell viability: DMSO control, 3.74% ± 0.71%; 3-MA, 95.02% ± 0.81%; P < 0.001). Inaddition, when siRNA knockdown of the autophagy-related proteins Atg7, Beclin-1, or ULK1,which were decreased by geldanamycin treatment, was performed, knockdown of ULK1 but notthat of Atg7 or Beclin-1 significantly inhibited sunitinib- induced autophagy, and autophagyinhibition by ULK1 knockdown significantly attenuated the sunitinib-induced decrease in cellviability (Fig. S3).

These results suggested that sunitinib- induced cytotoxicity in H9c2 cells was a result of increased autophagy and that inhibition of autophagy attenuated sunitinib- induced cytotoxicity.One of the major cellular responses caused by Hsp90 inhibitors such as geldanamycin is the degradation of Hsp90 client proteins and the induction of molecular chaperones such as Hsp70. To examine the general response of H9c2 cells to geldanamycin, we used DNA microarray analysis. Gene expression profiling of H9c2 cells resulted in 1662 gene probes that had at least a two- fold difference in expression with a false discovery rate P-value of <0.05 between geldanamycin- treated and non-treated H9c2 cells (863 and 799 probes for genes with increasedand decreased expression, respectively; Table S1). We analyzed the list of these gene probes using the Canonical Pathways analysis feature of the Ingenuity Pathway Analysis software. Unfolded protein response was the most relevant pathway for geldanamycin treatment (Table 1). The genes assigned to this canonical pathway are shown in Table 2. Hspa1a, which is a member of the heat shock protein 70 family, had the highest positive fold-change value among the genes examined. We then determined the protein expression levels for Hsp70 and Hsp90 by western blotting and found that the expression of Hsp70 was increased after exposure to geldanamycin for 6 h and was maintained at 24 h (Fig. 5A). Although Hsp90 appeared to be slightly increased after exposure to geldanamycin for 12 h, the level of protein expression was high even under basal condition.It is reported that pharmacological induction of Hsp70 by Hsp90 inhibitors protects against apoptotic cell death induced by doxorubicin (Demidenko et al., 2006). To investigate whether the induction of Hsp70 involved the attenuation of sunitinib-induced toxicity in H9c2 cells, we used two approaches to induce the expression of Hsp70. In H9c2 cells subjected to heat-shock treatment (43 °C for 1 h), Hsp70 expression was markedly increased from 3 to 24 h after treatment, with maximal expression at 6 h (Fig. 5B). A cytotoxicity assay revealed that viability of cells exposed to sunitinib at 6 h after heat-shock treatment was comparable with that in control cells not exposed to heat-shock treatment (Fig. 5C).

In H9c2 cells genetically engineered to express high levels of Hsp70 (Fig. 5D), viability of cells stably overexpressing Hsp70 and mock cells exposed to sunitinib did not differ significantly (Fig. 5E). These results indicated that the induction of Hsp70 is not involved in the geldanamycin- mediated attenuation of sunitinib- induced cytotoxicity in H9c2 cells.To investigate whether geldanamycin also protects against sunitinib- induced cytotoxicity in a more clinically relevant human cardiomyocyte model, we used hiPSC-derived cardiomyocytes; these cells have sarcomeric structures, electrophysiological properties, spontaneous contractility, and no proliferative capacity, which resemble phenotypes of human cardiomyocytes (Ma et al., 2011). In hiPSC-derived cardiomyocytes, geldanamycin pretreatment significantly attenuated the decrease in cell viability induced by sunitinib compared with DMSO control (Fig. 6A). Although exposure to geldanamycin alone did not affect the conversion of LC3-I to LC3-II compared with DMSO control, the increase in LC3-II induced by sunitinib exposure was significantly attenuated by geldanamycin pretreatment (Fig. 6B). These results indicate that geldanamycin attenuated sunitinib- induced cytotoxicity in clinically relevant human cells as it did in rat H9c2 cells.

4.Discussion
Here, we demonstrated that the Hsp90 inhibitor geldanamycin attenuated sunitinib- induced toxicity in rat H9c2 cells via inhibition of the autophagy pathway by promoting the degradation of autophagy-related proteins. These results suggest that combination or sequential treatment with Hsp90 inhibitors may be a useful strategy of reducing the cardiotoxic effects of sunitinib.In this study, we confirmed that sunitinib is cytotoxic to rat H9c2 cells (Fig. 1A) and hiPSC-derived cardiomyocytes (Fig. 6A), which is consistent with previous studies using other cardiomyocyte models such as neonatal rat ventricular myocytes and human cardiac myocytes (Hasinoff et al., 2008; Doherty et al., 2013). We found that cell viability, as determined from the cellular ATP content, was decreased by sunitinib in a concentration-dependent manner, implying that sunitinib disrupts cellular energy metabolism. Several potential mechanisms underlying sunitinib- induced cardiotoxicity via adverse effects on cellular metabolism have been reported. For example, sunitinib has been shown to inhibit several off-target proteins, including AMP-activated protein kinase (AMPK), which plays a role in cellular energy homeostasis, at clinically relevant concentrations (Fabian et al., 2005). In addition, sunitinib has been shown to directly inhibit AMPK activity in the murine heart and in isolated mouse cardiomyocytes (Kerkela et al., 2009). Inhibition of AMPK may damage cardiomyocytes because AMPK is crucial for ATP production and pro-survival signaling in heart cells (Terai et al., 2005; Dyck and Lopaschuk, 2006). However, studies in neonatal rat myocytes and hiPSC-derived cardiomyocytes have suggested that inhibition of AMPK is not a major component of sunitinib- induced cardiotoxicity (Hasinoff et al., 2008; Cohen et al., 2011). This contradictory data may be a result of the different species of animals and types of cells used in the experiments.

Alternatively, autophagy may play an important role in sunitinib- mediated cardiotoxicity(Zhao et al., 2010). Autophagy is essential for degrading and recycling long- lived cytosolic proteins and organelles, and therefore autophagy plays an important role in energy homeostasis. Thus, dysregulation of autophagy in cardiomyocytes may be the mechanism underlying the development of cardiovascular dysfunction and disease (Nakai et al., 2007; Lavandero et al., 2015). In the present study, we confirmed that sunitinib abnormalized autophagy in rat H9c2 cells (Fig. 2) and hiPSC-derived cardiomyocytes (Fig. 6B). Although sunitinib is known to be sequestered in lysosomes (Gotink et al., 2011), our autophagic flux assay using the well-characterized lysosomotropic agent CQ indicated that sunitinib induces autophagic flux rather than inhibits autophagosome- lysosome fusion or autolysosome function. Ribosomal S6 kinase family, which is a family of off-target proteins reported for sunitinib (Fabian et al., 2005), is related not only to the regulation of survival signaling through inhibitory phosphorylation of the pro-apoptotic factor Bcl2-associated agonist of cell death, BAD, but also the induction of autophagy. In addition, sunitinib inhibits phosphorylation of the protein kinases Akt and mTOR, which results in reduced phosphorylation of ribosomal protein S6 kinase beta-1 (Saito et al., 2012). Thus, because mTOR signaling regulates autophagy through phosphorylation of the autophagy-initiating kinase ULK1, which constitutes a complex with Atg13 and FIP200 that is crucial for the initiation of autophagy, direct inhibition of mTOR signaling could be one of the mechanisms through which sunitinib induces autophagy in cardiomyocytes.Inhibition of autophagy via treatment with autophagy inhibitors or siRNA- mediatedknockdown of ULK1 confirmed that sunitinib- induced autophagy was the mechanism underlyingthe cytotoxicity of sunitinib in H9c2 cells (Figs. 4, S3). We also pharmacologically inhibitedautophagy by using CQ and found that sunitinib in combination with CQ was significantly morecytotoxic than sunitinib alone (Fig. S4).

The reason why concomitant treatment with sunitinibimpairing lysosomal function, in H9c2 cells (Ross et al., 2000). We found that 3-MA, which is an inhibitor of phosphatidylinositol 3-kinase, reduced sunitinib- induced cytotoxicity more than did bafilomycin A1 and NH4Cl, which are a specific inhibitor of vacuolar type H+ ATPase and a lysosomotropic agent, respectively, and inhibit lysosome function and autophagosome- lysosome fusion. This suggests that the attenuation of sunitinib-induced cytotoxicity may be achieved efficiently by disruption of upstream phases of the autophagy pathway that involve phosphatidylinositol 3-kinase, such as phagophore formation. Targeted inhibition of autophagy by using small- molecule autophagy inhibitors is a promising means of attenuating sunitinib- induced cardiotoxicity.In this study, we found that several Hsp90 inhibitors, in particular geldanamycin, attenuated sunitinib- induced cytotoxicity in H9c2 cells. Geldanamycin treatment markedly promoted the degradation of Atg7, Beclin-1, and ULK1, which are key regulators of autophagy initiation, and inhibited autophagy induction (Fig. 3). This is consistent with previous reports demonstrating that knockdown of the genes encoding these proteins decreases autophagic activity by inhibiting the formation of autophagosomes (Chan et al., 2007). Thus, geldanamycin likely attenuates sunitinib- induced cytotoxicity in H9c2 cells by inhibiting autophagy via inhibition of the function of autophagy regulator proteins, especially those involved in autophagy initiation,which is supported by our present results (Fig. S3) and a previous report in which suppression ofautophagy by using Beclin-1 siRNA reduced sunitinib- induced cytotoxicity in H9c2 cells (Zhaoet al., 2010).In the present study, we used two types of cardiomyocytes, H9c2 cells, and hiPSC-derivedconditions such as an increased reliance on fatty acid oxidation.Geldanamycin activates heat shock factor 1 and induces production of heat shock proteins such as Hsp70, which is known to protect against cellular stress and damage (Griffin et al., 2004;Tarone and Brancaccio, 2014). For example, Hsp70-overexpressing transgenic mice are protected against cardiac dysfunction induced by doxorubicin, and pharmacological induction of Hsp70 by Hsp90 inhibitors has been shown to protect against doxorubicin- induced cell death (Demidenko et al., 2006; Naka et al., 2014).

We established two different types of H9c2 cells expressing Hsp70 and performed a cytotoxicity assay to examine the contribution of inducible Hsp70 to the attenuation of sunitinib-induced cytotoxicity; however, no cytoprotective effects of Hsp70 were detected under any of the experimental conditions (Fig. 5). This is probably because the mechanisms underlying the cytotoxic effects of doxorubicin and sunitinib are different. A major mechanism through which doxorubicin exerts its cytotoxic effects on cardiomyocytes is oxidative stress, and Hsp70 has been shown to protect against cell damage and apoptosis by decreasing the production of reactive oxygen species (Vayssier and Polla, 1998; Octavia et al., 2012). In contrast, sunitinib has been shown not to induce the generation of reactive oxygen species in human cardiomyocytes (Doherty et al., 2013).It has been reported that inhibiting autophagic activity during sunitinib treatment enhances its antitumor activity compared with monotherapy in some cancer cells (Ikeda et al., 2013; DeVorkin et al., 2016). In the present study, inhibition of autophagy with geldanamycin protected cardiomyocytes from the cytotoxicity caused by sunitinib. Although the precise mechanisms underlying these two different effects of the inhibition of autophagy remain unclear, at present it can be understood that autophagy is a double-edged sword that has positive or negative effects depending on the conditions of autophagy induction, levels of autophagic flux, and cell type. Further research is required to better understand the different effects of sunitinib treatment and inhibition of autophagic activity in cancer cells and cardiac cells.

In conclusion, the results presented here suggest that combination or sequential treatment with Hsp90 inhibitors and sunitinib may be a promising strategy of improving cancer management by attenuating the cardiotoxic effects of sunitinib. Further investigation into this novel therapeutic strategy for cancer treatment is Tanespimycin warranted.