Celastrol targets mitochondrial respiratory chain complex I to induce reactive oxygen species-dependent cytotoxicity in tumor cells

H1299 and HepG2 cells were obtained from the American Tissue Culture Collection (Manassas, VA). Cells were cultured with Dulbecco's modified Eagle medium (Gibco, Grand Island, NY) containing 10% fetal bovine serum (FBS, HyClone, Logan, UT). Celastrol was purchased from Calbiochem (San Diego, CA) and dissolved in DMSO. N-acetylcysteine and 17-allylamino-17-demethoxygeldanamycin were purchased from Sigma (St Louis, MO). SP600125 was purchased from LC laboratories (Worburn, MA). Z-VAD-FMK was purchased from R&D Systems (Minneapolis, MN, USA).

Cells were collected by trypsinization and incubated with 0.4% trypan blue for 3 minutes at room temperature. The viable (unstained) cells were counted and used to calculate viability.

50,000 cells were fixed with 70% ethanol containing 1% FBS at -20°C overnight, and then incubated with RNase A (20 μg/mL) at 37°C for 30 minutes, stained with propidium iodide (100 μg/mL) for 10 minutes, and then analyzed by flow cytometry (FACSCalibur, BD, USA) and ModFit LT software (FACSCalibur, BD, USA). For each measurement, 20,000 cells were analyzed and the representative measurements were shown.

Apoptosis and necrosis were determined by a Annexin V-FITC Kit (Roche, Indianapolis, IN). H1299 and HepG2 cells were collected by trypsinization. After washed with PBS, cells were stained with annexin V-FITC and propidium iodide, and then the apoptosis and necrosis were determined by flow cytometry (FACSCalibur) and CELLQuest software (FACSCalibur). For each measurement, 20,000 cells were analyzed and the representative measurements were shown.

The mitochondrial membrane potential was measured with a Mitochondrial Membrane Potential Assay Kit with JC-1 (Beyotime biotechnologies, Jiangsu, China) according to the manufacturer's instruction. Briefly, 50,000 cells were collected by trypsinization and incubated with JC-1 for 20 minutes at 37°C in the dark. The stained cells were washed with ice-cold working solution twice and then analyzed by flow cytometry (FACSCalibur) and CELLQuest software (FACSCalibur). 20,000 cells were analyzed in each measurement. JC-1 aggregates in the polarized mitochondrial matrix and forms J-aggregates, which emit red fluorescence at 595 nm when excited at 525 nm. However, JC-1 cannot aggregate in the depolarized mitochondrial matrix and exists as JC-1 monomers, which emit green fluorescence at 525 nm when excited at 485 nm. Mitochondrial depolarization is indicated by a decrease in the red/green fluorescence intensity ratio.

ROS were measured by a Reactive Oxygen Species Assay Kit (Applygen Technologies, Beijing, China). Briefly, cells were incubated with 3 μM DCFH-DA for 40 minutes at 37°C in the dark, and ROS were determined by fluorescence microscopy or flow cytometry (FACSCalibur) at an excitation wavelength of 480 nm and an emission wavelength of 525 nm. More than 3 fields were observed by fluorescence microscope and 20,000 stained cells were analyzed with flow cytometry in each measurement. The ROS fold was calculated based on the mean geometry fluorescence determined by flow cytometry and shown as a histogram.

Immunoprecipitation and western blotting were performed according to the method described by Yu [34]. For immunoprecipitation experiments, cells were lysed in TNESV buffer containing 50 mM Tris (pH 7.5), 2 mM EDTA, 100 mM NaCl, 2% Nonidet P-40 (NP-40), 1 mM Na3VO4 and cocktail (1 tablet/10 mL solution, Roche, Indianapolis, USA) at 4°C for 30 minutes. Cell lysates were centrifuged at 10,000 g to remove cellular debris. Protein concentration in the lysate was quantified with a Bicinchoninic Acid (BCA) Protein Assay Kit (Pierce, Rockford, IL, USA). Approximately 500 μg total proteins was incubated with 2 μg heat shock protein 90 antibody (Stressgen, Victoria, British Columbia, Canada) at 4°C for 12 h, after which 20 μL of protein A/G Plus Agarose (Santa Cruz, CA, USA) was added into the mixture and incubated for 2 h at 4°C. Agarose-antibody-protein complexes were washed one time with lysis buffer and two times with PBS. The immune complexes were resuspended in 20 μL of Laemmli Buffer (Bio-Rad Laboratories, CA, USA) and boiled for 5 minutes. The samples were analyzed by western blotting.

For western blotting experiments, cells were lysed in Laemmli Buffer (Bio-Rad Laboratories) and protein concentration in the lysate was quantified by a BCA Protein Assay Kit (Pierce,). Fifty micrograms of total protein was subjected to SDS-PAGE and transferred onto a PVDF membrane. The membrane was blotted with primary antibodies for 12-15 h at 4°C and incubated with horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature. Proteins were detected using a SuperEnhanced Chemiluminescence Detection Kit (Applygen Technologies). The antibodies used in the study were anti-PARP, cleaved caspase 3, caspase 9, phospho-JNK, JNK, AKT, epidermal growth factor receptor, heat shock protein 90, CDK4 and Thioredoxin (Cell Signaling, Beverly, MA); anti-catalase (Abcam, Cambridge, MA); anti-Bad (Transduction Laboratories, KY); anti-cytochrome c, superoxide dismutase and Cdc37 (Santa Cruz,) and anti-β-actin (Oncogene, Boston).

The mitochondria were isolated with a Mitochondrial Isolation Kit (Applygen Technologies). Fifty million cells were resuspended with ice-cold Mito-Cyto isolation buffer and homogenized with the grinder. Homogenate was centrifuged at 800 g for 10 minutes at 4°C. The supernatants were collected in a new tube, and centrifuged at 10,000 g for 10 minutes at 4°C. The supernatant and pellet were saved as cytosolic fraction and intact mitochondria, respectively. The intact mitochondria were lysed with Laemmli Buffer (Bio-Rad Laboratories) to extract mitochondrial protein. The alteration of Bad and cytochrome c in mitochondria and cytoplasm were analyzed by western blotting.

The activity of MRC complexes was determined with Mitochondrial Respiratory Chain Complexes Activity Assay Kits (Genmed Scientifics, Shanghai, China). Briefly, the isolated mitochondria were resuspended with the Mito-Cito buffer (Applygen Technologies), frozen at -70°C and thawed at 37°C three times to extract the mitochondrial proteins. The protein concentration in the lysate was determined using the BCA Protein Assay Kit (Pierce, Rockford) and diluted to 0.1 μg/μL. The absorbance was determined on a Smartspec™ Plus spectrophotometer (Bio-Rad). The activity of complex I-linked NADH-ubiquinone reductase was determined by measuring the reduction of ubiquinone to ubiquinol, which leads to a decrease in absorbance of NADH at 340 nm. The activity was measured with or without rotenone, a specific inhibitor of NADH-ubiquinone reductase. The specific activity of complex I was calculated by subtracting the rotenone-nonsensitive activity from the total activity and is expressed as μM NADH mg/min. Complex II-linked succinate-ubiquinone reductase activity was determined by measuring the reduction of 2,6-dichlorophenolindophenol (DCIP), which can be monitored at 600 nm. The activity is expressed as μM DCIP mg/min. Complex III-linked ubiquinol cytochrome c reductase activity was determined by monitoring the reduction of cytochrome c by the electrons donated from ubiquinol, which can be monitored at 550 nm. The activity was measured with or without antimycin A, a specific inhibitor of ubiquinol cytochrome c reductase. The specific activity of complex III was calculated by subtracting the antimycin A-nonsensitive activity from the total activity and is expressed as μM CoQH2 mg/min. Complex IV-linked cytochrome c oxidoreductase activity was determined by measuring the oxidation of cytochrome c, which can be monitored at 550 nm. The activity was expressed as μM Cyt c mg/min. All measurements were performed in triplicate.

To determine the role of ROS in mediating celastrol-induced cytotoxicity, we first measured ROS levels in H1299 and HepG2 cells after celastrol exposure. As shown in Figure 1A, celastrol increased ROS levels in a dose-dependent manner in both H1299 and HepG2 cells. Celastrol also reduced cell viability in both H1299 and HepG2 cells in a dose-dependent manner (Figure 1B). Celastrol arrested cell cycle in both cell lines, and the G2-M phase ratio rose from 15 ± 1.6% to 41 ± 3.1% in H1299 cells and 15 ± 1.8% to 34 ± 3.5% in HepG2 cells after 24 h of treatment with 6 μM celastrol (Figure 1C). The results of annexin V-FITC and propidium iodide (PI) staining showed that celastrol induced apoptotic and necrotic cell death in a dose-dependent manner, and the percent of cell death (apoptosis and necrosis) was 41 ± 4.1% in H1299 cells and 22 ± 2.5% in HepG2 cells treated with 6 μM celastrol for 24 h (Figure 1D). Therefore, ROS levels positively correlate with cell death and growth arrest induced by celastrol. Moreover, scavenging of ROS by the antioxidative agent N-acetylcysteine (NAC) inhibited celastrol-induced decrease in cell viability, cell cycle arrest and cell death (Figure 1). As ROS have been reported to mediate the cytotoxicity induced by some cytotoxic agents [22, 25], we assumed that ROS played a critical role in mediating celastrol-induced cytotoxicity as well.

Celastrol has been identified as a novel HSP90 inhibitor and mediates cytotoxicity by facilitating the degradation of HSP90 client proteins [7, 9, 15‐17]. Thus, we determined whether ROS were involved in mediating the decline of HSP90 client proteins induced by celastrol. Levels of the HSP90 client proteins including AKT, epidermal growth factor receptor (EGFR) and CDK4 were all decreased following celastrol treatment in H1299 cells (Figure 2A). However, the decrease of HSP90 client proteins induced by celastrol was completely inhibited by NAC (Figure 2A). To determine whether the decline of HSP90 client proteins induced by celastrol was dependent on apoptosis, we analyzed the effect of celastrol on HSP90 client protein expression in the absence and presence of Z-VAD-FMK (Z-VAD), a pan-caspase inhibitor. As shown in Figure 2B, 50 μM Z-VAD significantly decreased the cleavage of PARP (apoptosis marker) induced by celastrol, indicating that Z-VAD inhibited celastrol-induced apoptosis. However, Z-VAD had no significant effect on celastrol-induced decrease of HSP90 client proteins including EGFR, AKT and CDK4, suggesting that celastrol-induced decrease of HSP90 client proteins is not dependent on apoptosis. By using a GST pull-down assay, a previous study has reported that celastrol could disrupt the HSP90/Cdc37 complex [15]; however, we did not observe that the interaction between endogenous HSP90 and Cdc37 was disrupted by celastrol in H1299 cells (Figure 2C). In addition, we compared the effect of NAC in reversing the decrease of HSP90 client proteins induced by 17-allylamino-17-demethoxygeldanamycin (17-AAG), a classic HSP90 inhibitor, with that induced by celastrol. Both 17-AAG and celastrol induced the depletion of HSP90 client proteins including AKT and EGFR whereas NAC blocked celastrol-induced depletion of ATK and EGFR but had no effect on 17-AAG-induced HSP90 client protein degradation (Figure 2D). These data indicate that celastrol-induced inhibition of HSP90 chaperone function is mediated by ROS.

ROS play a critical role in mediating the cytotoxicity induced by celastrol, but the targets by which celastrol induces ROS accumulation are unknown. To identify the targets for celastrol-induced ROS accumulation, we first measured ROS levels in H1299 cells at different time points after celastrol exposure. As shown in Figure 3A, celastrol induced time-dependent ROS accumulation in H1299 cells, and ROS levels increased 7.7 ± 0.7 times after 1 h of celastrol treatment. The downregulation of antioxidative proteins results in ROS accumulation [24]; therefore, we investigated the effect of celastrol on the expression of antioxidative proteins, including catalase, SOD, and Trx. Celastrol had no significant effect on SOD but slightly downregulated catalase and Trx expression (Figure 3B). However, the downregulation of catalase and Trx was completely inhibited by NAC (Figure 3B). Therefore, celastrol does not cause ROS accumulation by downregulating the antioxidative proteins SOD, catalase, and Trx.

Inhibiting the activity of MRC complexes promotes ROS production [25]. Therefore, we speculated that celastrol may induce ROS accumulation by inhibiting MRC complex activity. We detected the effect of celastrol on different MRC complexes. Celastrol completely inhibited MRC complex I activity but increased MRC complex IV activity (Figure 3C), indicating that complex I, but not complex IV, may be the target of celastrol in mediating ROS accumulation. The activity of MRC complex II and III decreased in H1299 cells treated with celastrol; however, NAC reversed the inhibition (Figure 3C), indicating that complex II and III are not likely targets of celastrol for ROS accumulation. In addition, MRC complex I activity was completely inhibited as early as 30 minutes after celastrol exposure (Figure 3D), which precedes ROS accumulation. In contrast, the activity of MRC complex II, III and IV did not significantly decrease after 30 or 60 minutes of treatment with celastrol (Figure 3D), further confirming that complex I is the target of celastrol in promoting ROS production.

Because celastrol induced apoptosis in a dose-dependent manner (Figure 1D), we further investigated the effect of celastrol on mitochondria. The mitochondrial membrane potential (MMP) was measured using a JC-1 fluorescent probe, and the JC-1 red/green fluorescence intensity ratio was used to represent MMP. As shown in Figure 4A, celastrol treatment resulted in a dose-dependent decrease in MMP in H1299 cells. The JC-1 red/green fluorescence intensity ratio in H1299 cells treated with 6 μM celastrol for 24 h decreased to 2 ± 0.2%, indicating that celastrol significantly induced dysfunction of mitochondria. As before, NAC inhibited the decrease of MMP induced by celastrol (Figure 4A), indicating that ROS were involved in mediating mitochondrial depolarization. Moreover, as shown in Figure 4B, NAC could completely block the cleavage of PARP, caspase 9, and caspase 3 in H1299 cells treated with celastrol. These data are consistent with celastrol induced mitochondrial-mediated apoptosis through ROS accumulation.

As the stress-activated protein kinase, JNK can be activated by oxidative stress induced by accumulated ROS [35, 36]. Activated JNK initiates the mitochondrial apoptotic pathway by mediating the activation and mitochondrial translocation of proapoptotic proteins including Bax and Bad [37‐39]. We determined the effect of celastrol on JNK activation. As shown in Figure 4C, celastrol increased the level of JNK phosphorylation in a dose-dependent manner whereas NAC completely inhibited celastrol-induced JNK phosphorylation, demonstrating that JNK was activated by celastrol-induced ROS accumulation. The results of annexin V-FITC and PI staining showed that JNK inhibitor SP600125 (SP) significantly inhibited celastrol-induced apoptosis and necrosis. The percentage of cell death (apoptosis and necrosis) was 43 ± 4.1% in cells treated with celastrol and 11 ± 1.2% in cells treated with celastrol and SP (Figure 4D).

We further determined the mechanism by which JNK mediates celastrol-induced cell death. As shown in Figure 5A, SP did not suppress celastrol-induced ROS accumulation, indicating that SP does not inhibit celastrol-induced cell death by suppressing ROS accumulation. However, SP significantly inhibited celastrol-induced depolarization of mitochondria. The JC-1 red/green fluorescence intensity ratio was 7 ± 1.4% and 68 ± 4.2% in H1299 cells treated with 6 μM celastrol alone and 6 μM celastrol with SP for 24 h, respectively (Figure 5B). Moreover, SP inhibited celastrol-induced JNK phosphorylation and cleavage of PARP, caspase 9 and caspase 3 (Figure 5C). Furthermore, SP significantly inhibited mitochondrial translocation of Bad and cytoplasmic release of Cytochrome c induced by celastrol (Figure 5D). Therefore, ROS-induced JNK activation played a crucial role in initiating mitochondria-mediated apoptosis of H1299 cells treated with celastrol.