Article pubs.acs.org/jpr
Sorafenib-Induced Mitochondrial Complex I Inactivation and Cell Death in Human Neuroblastoma Cells Vibeke Hervik Bull,† Krishnaraj Rajalingam,*,‡,§ and Bernd Thiede*,†,§ †
The Biotechnology Centre of Oslo, University of Oslo, P.O. Box 1125 Blindern, 0317 Oslo, Norway Emmy Noether Group of the DFG, Institute of Biochemistry II, Medical Faculty of the Goethe University, University Hospital Building 74/75, Theodor-Stern-Kai 7, 60528 Frankfurt am Main, Germany
‡
S Supporting Information *
ABSTRACT: Sorafenib is a multikinase inhibitor that is approved for use against renal cell and hepatocellular carcinoma. We found that sorafenib potently induced cell death in human neuroblastoma cells. To understand the molecular basis of sorafenib-mediated cell death in human SH-SY5Y cells, we performed a temporal quantitative proteome analysis. The results showed significant quantitative changes of 193 unique proteins. Bioinformatics-assisted pathway analysis of the regulated proteins revealed that mitochondrial proteins, especially components of the electron transport chain and the mitochondrial ribosomes, were significantly affected upon exposure to sorafenib. The observed down-regulation of the respiratory chain complex I (NADH dehydrogenase) was accompanied with loss of mitochondrial transmembrane potential (Δψm) and complete impairment of complex I enzyme activity. The destabilization of complex I subunits was consistent, rapid, and independent of caspase activation as well as Bcl-2 overexpression. This study provides an overview of the molecular machinery driving sorafenib-mediated cell death in neuroblastoma cells and suggests that sorafenib could be a potential therapeutic drug for the treatment of neuroblastoma. KEYWORDS: apoptosis, complex I, mitochondria, neuroblastoma, SILAC, sorafenib
■
INTRODUCTION Sorafenib (Nexavar, BAY 43-9006) is a newly discovered drug that has been approved by the FDA for use against renal cell carcinoma (RCC) and hepatocellular carcinoma (HCC). In addition, it has reached phase I and II trials for colorectal carcinomas and breast cancer, respectively. Phase II/III studies of RCC patients showed that sorafenib increase the progressionfree survival 2−4-fold compared with placebo;1,2 however, less than 1% of sorafenib-treated patients had complete regression, and this issue is now addressed by testing sorafenib treatment in combination with other drugs. Sorafenib-induced side effects are common, but normally tolerable, and include diarrhea, skin rash, fatigue, hand-foot skin reactions, alopecia, and nausea. Serious but rare side effects include hypertension and cardiac ischemia.2 Sorafenib was developed to target CRAF, a central member of the Ras/mitogen activated protein kinase (Ras/MAP kinase) signaling pathway, which regulates important cellular processes.3,4 Additionally, sorafenib has been reported to be a multikinase inhibitor targeting several cell surface tyrosine kinase receptors involved in tumor angiogenesis, such as vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF).3 Furthermore, it was also found to inhibit growth factor Fms-related tyrosine kinase 3 ligand (Flt-3) and cytokine receptor c-Kit. Neuroblastoma is one of the most abundant childhood cancers, occurring in approximately one out of 7000 children. © 2012 American Chemical Society
It is a solid extra-cranial cancer originating from the sympathetic nervous system that results in paraspinal located tumors in the abdomen or chest. They often present abnormal DNA ploidy (near triploid or tetraploid) and DNA rearrangements, such as amplification of MYCN gene as well as expression of different Trk neurothropin receptors.5 Neuroblastomas are highly vascular, and as sorafenib directly targets factors that regulate angiogenesis, it could be a valuable therapeutic agent. Although several direct targets of sorafenib are known, the exact mechanism for sorafenib-induced cell death is not well understood. The process of apoptotic programmed cell death is mainly executed through two protein families, the Bcl-2 family and the proteolytic caspases reviewed in ref 6. Intrinsic death stimuli, such as UV radiation, stress, and chemotherapeutic drugs, typically activate the Bcl-2 family proteins that regulate the integrity of the mitochondrial membrane. Permeabilization of the mitochondrial outer membrane results in release of a number of proapoptotic factors from the mitochondrial intermembrane space and marks the point of no return for the process.7 Conversely, transmembrane death-receptor proteins are activated by extracellular ligands, which result in activation of the caspase cascade and apoptosis.8 The two signaling pathways eventually accumulate in common downstream events including Received: August 17, 2011 Published: January 23, 2012 1609
dx.doi.org/10.1021/pr200790e | J. Proteome Res. 2012, 11, 1609−1620
Journal of Proteome Research
Article
before protein concentrations were determined by measuring the absorbance at 280 nm using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). After determination of protein concentration, 1% Triton X-100 was added the lysates, followed by additional homogenization and centrifugation. For stable isotope labeling with amino acids in cell culture (SILAC) neuroblastoma cells (SH-SY5Y) were maintained in RPMI-1640 culture medium without arginine or lysine (Invitrogen, Oslo, Norway) supplemented with 5% dialyzed fetal calf serum (Invitrogen) and penicillin (100 U/mL) (Gibco BRL) and 100 mg/L of arginine-12C6 monohydrochloride and lysine-12C6 monohydrochloride (light) or arginine-13C6 monohydrochloride and lysine-13C6 monohydrochloride (heavy) (Sigma-Aldrich, Oslo, Norway) at 37 °C in 5.0% CO2. After four population doublings of the cells, the incorporation of arginine and lysine isotopes was checked according to Schmidt et al.9 After five cell population doublings, cells grown in light-arginine/lysine-containing medium were incubated with sorafenib (15 μM) for 12 and 24 h, before harvesting of heavyarginine/lysine-labeled control cells and light arginine/lysinelabeled sorafenib-treated cells.
activation of initiator and effector caspases, as well as morphological changes such as DNA condensation and fragmentation, cell shrinking, and blebbing. We observed that sorafenib induces cell death in the neuroblastoma cell line SH-SY5Y. To investigate the intracellular effects of sorafenib in this cell line, we combined quantitative proteome analysis using SILAC in combination with high resolution mass spectrometry. Different bioinformatic tools were applied to predict functional association between quantitatively modified proteins. This approach resulted in new insights, anchored with statistical evidence, of which cellular processes and molecular signaling pathways were affected by sorafenib. In particular, validation analyses confirmed that sorafenib-induced complex I inactivation is accompanied with loss of mitochondrial membrane potential and cell death.
■
MATERIALS AND METHODS RPMI-1640 without phenyl red (11835-063) and tetramethylrhodamine ethyl ester (TMRE) (T669) were purchased from Invitrogen, (Oslo, Norway); cycloheximide (01810), ammonium diethyldithiocarbamate (DDC) (359548), anti-SOD2, and antiMRPL18 from Sigma (Oslo, Norway); sorafenib (BAY 43-9006) from Alexis Biochem (San Diego, USA) and LC Laboratories (Woburn, MA, USA); Z-VAD-(OMe)-FMK (N-150) from Bachem AG (Bubendorf, Switzerland); Complex I enzyme activity assay kit (MS130) from MitoSciences (Eugene, OR, USA); and Annexin V-FITC apoptosis detection kit (ALX850-020-KI02) from Enzo Life Sciences (Plymouth Meeting, PA, USA). Effectene transfection agent and Qproteome mitochondria kit were from Qiagen (Oslo, Norway). Affinity-purified antibodies against Bad, Bid, Bim, Bcl-xL, caspase-10, cytochrome c, phospho-Erk1/2 (Thr202/Tyr204), poly (ADP-ribose) polymerase-1 (PARP1), and GAPDH were purchased from Cell Signaling (Beverly, MS, USA), while caspase-7 and cleavedcaspase-3 antibodies were purchased from Merck KGaA (Darmstadt, Germany). Antibodies against caspase-8 p18 (H-134), caspase-2L (C-20), and Ndufa6, p21 (C19) were bought from Santa Cruz (Heidelberg, Germany). Myeloid leukemia cell differentiation protein (Mcl-1) and X-linked inhibitor of apoptosis protein (XIAP) were obtained from BD Biosciences (Heidelberg, Germany), while antibody against β-tubulin was bought from Millipore. Caspase-6, -8, and -9 were from a cleaved Caspase Antibody Sampler Kit (AP1026) from Calbiochem, (Darmstadt, Germany). HRP-conjugated anti-mouse and anti-rabbit IgG were purchased from Promega (Madison, WI, USA). GFP-Bcl-2 plasmid was obtained from Clark Distelhorst.
Transient Transfection
GFP-tagged plasmids (pEGFP-C1 and GFP-Bcl-2) were transfected with effectene transfection reagent according to manufacture's recommendations with some modifications. The DNA amount was increased to 4-fold, and effectene amount was doubled compared to the recommended doses. This was done to increase transfection efficiency. Immunoblotting
Protein samples were separated by SDS-PAGE and transferred onto PVDF membranes (Immobilon P, Millipore, Oslo, Norway) or nitrocellulose transfer membrane (Whatman Protran BA83) using a Mini Trans-Blot cell (Bio-Rad, Munich, Germany). After blocking in TBST (10 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.4) containing 5% nonfat dried milk, membranes were incubated with affinity-purified antibodies O/N at 4 °C. HRP-conjugated anti-mouse or antirabbit IgG was used as secondary antibody. Membranes were developed by SuperSignal West Pico Chemiluminescent (VWR, Oslo, Norway). Mass Spectrometry Analysis
Each lane of Coomassie G-250 stained SDS-PAGE gels was divided into 24 bands for in-gel digestion using 0.1 μg of trypsin in 20 μL of 50 mM ammonium bicarbonate, pH 7.8.10 An LC−MS system consisting of a Dionex Ultimate 3000 nano-LC system (Sunnyvale CA, USA) connected to a linear quadrupole ion trap−Orbitrap (LTQ Orbitrap XL) mass spectrometer (ThermoElectron, Bremen, Germany) equipped with a nanoelectrospray ion source was used to analyze the tryptic peptides. For liquid chromatography separation, an Acclaim PepMap 100 column (C18, 3 μm, 100 Å) (Dionex, Sunnyvale CA, USA) capillary of 12 cm bed length was used with a flow rate of 300 nL/min. Two solvents A (0.1% formic acid) and B (aqueous 90% acetonitrile in 0.1% formic acid) were used in the nano column. The gradient went from 7% to 35% B in 77 min, and from 35% to 50% in 10 min. The mass spectrometer was operated in the data-dependent mode to automatically switch between Orbitrap-MS and LTQ-MS/MS acquisition. Survey full scan MS spectra (from m/z 300 to 2,000) were acquired in the Orbitrap with resolution R = 60,000 at m/z 400 and allowed
Cell Culture and Incubation with Sorafenib
Human neuroblastoma cells (SH-SY5Y cells; approved by the Helsedirektoratet, Oslo, Norway) were grown in a monolayer in RPMI-1640 (Invitrogen, Oslo, Norway) supplemented with 5% fetal calf serum (Invitrogen, Oslo, Norway), 0.2 M glutamine and penicillin and maintained in a humidified incubator at 37 °C in a 5% CO2 environment. For apoptosis induction, SH-SY5Y cells were exposed to 15 μM sorafenib for 0, 12, and 24 h before harvesting. Protein synthesis was inhibited by 50 μM cycloheximide. Cell pellets were resuspended in lysis buffer (25 mM Tris-HCl, 50 mM KCl, 3 mM EDTA, 5 mM β-mercaptoethanol, pH 7.1) supplemented with protease inhibitors (3 mM benzamidine, 10 μM leupeptin, and 1 mM PMSF) and phosphatase inhibitors (30 mM NaF, 1 mM Na3VO4, 20 mM Na4P2O7) and homogenized using a pestle pellet. Lysates were centrifuged for 10 min at 16,000g and 4 °C, 1610
dx.doi.org/10.1021/pr200790e | J. Proteome Res. 2012, 11, 1609−1620
Journal of Proteome Research
Article
Complex I Activity Assay
sequential isolation of the most intense ions, up to six, depending on signal intensity, for fragmentation on the linear ion trap using collisionally induced dissociation at a target value of 10,000 charges. For accurate mass measurements the lock mass option was enabled in MS mode and the polydimethylcyclosiloxane (PCM) ions generated in the electrospray process from ambient air were used for internal recalibration during the analysis.11 Other instrument parameters were set as previously described.12 Raw data were processed using DTA SuperCharge v.2.0a7 to generate mgf files. The data from each lane were merged into two separate mgf files, bands 1−12 in one and bands 13−24 in another, due to the limited capacity of MSQuant.
Complex I activity was measured using complex I activity dipstick assay from MitoSciences (Eugene, OR, USA) according to the manufactures descriptions. The kit is developed to detect the oxidation of NADH to NAD+ catalyzed by complex I. Measurement of Mitochondrial Membrane Potential and Apoptosis
For measurement of changes in mitochondrial membrane potential (Δψm), SH-SY5Y cells were harvested before suspension and incubation in prewarmed RPMI 1640 media containing 50 μM of the red-orange fluorescent dye tetramethylrhodamine ethyl ester (TMRE) for 30 min in 37 °C in 5% CO2 in the dark. To detect apoptotic cells, SH-SY5Y cells were incubated with Annexin V-FITC for 15 min in the dark, washed once with PBS, and resuspended in propidium iodide (PI)containing buffer. Fluorescence was measured by flow cytometry using a FACSCanto II (BD Biosciences).
Protein Identification, Quantification, and Bioinformatic Data Analysis
Proteins were identified using an in-house version of the protein identification software Mascot version 2.2.13 Mass spectra were searched against the Swiss-Prot database (20081212, human, 20411 sequences). Mass accuracy was set to 10 ppm for MS mode and to 0.6 Da for MS/MS mode, and one missed tryptic cleavage site was allowed in the search. In addition to 13C6 isotopes of lysine and arginine, methionine oxidation, acetylation at the protein N-terminus, pyroglutamate formation of Nterminal glutamine, deamidation of asparagines and glutamines, and propionamide formation of cysteines were allowed as variable modifications. Automatic decoy database searches were performed in Mascot and revealed a false discovery rate for peptide matches above an identity threshold of less than 4% for the seven data sets. The raw data, including peak lists and experimental information, are made available at Proteomics IDEntifications (PRIDE) database for all seven parallels (accession numbers: 13698−13711). The search result file (HTML) from the Mascot search engine and the raw spectrum files were used for protein quantification using MSQuant version 2.0a81,14 based on precursor ion intensities. For the entries where the ratios of the different peptides within a protein varied by more than 20%, the data were checked manually and poor quality spectra were removed. For every biological replicate, the quantifications were normalized against the median to correct for systematic errors. The adjusted data from each replicate were merged, and an average ratio was calculated for every protein before further data analysis. Acceptance criteria for protein identifications were a score above the Mascot identity threshold (set at 95% confidence level), at least two peptides identified, each with an ion score of at least 20. In addition, the protein had to be identified in at least three out of four (0 vs 24 h sorafenibincubated cells) or two of three (0 vs 12 h sorafenib-incubated cells) biological replicates. Finally, at least two peptides had to be quantified for at least one of these parallels. Grubbs test was used to remove outliers (p < 0.05) and to calculate mean and standard deviation (SD) resulting in mean ± 1.96 SD as limit for regulated proteins. We performed comprehensive bioinformatic analyses using DAVID (http://david.abcc.ncifcrf.gov)15,16 and Cytoscape/Biological Network Gene Ontology (BiNGO)17,18 for GO term and pathway analysis, and FunCoup (http://funcoup.sbc. su.se)19 for functional coupling between the regulated proteins to detect which biological processes and intracellular pathways that were affected by sorafenib treatment. The p-values obtained from the DAVID and Cytoscape/BiNGO analysis were adjusted for multiple hypotheses testing using the Benjamini−Hochberg false discovery rate correction.
FACS Measurement
TMRE and PI were measured in the FL-2 channel, whereas Annexin V-FITC was measured in the FL-1 channel of a FACSCanto II (BD Biosciences). All data were analyzed with FlowJo (Tree Star) software, where populations were gated for live or live and apoptotic cell populations for Δψm and apoptosis measurements, respectively.
■
RESULTS
Sorafenib Induces Apoptotic Signaling Events in Human Neuroblastoma Cells
Human neuroblastoma cells (SH-SY5Y) treated with sorafenib for 24 h showed the classical morphology of apoptotic cell death including cell rounding, shrinkage, and blebbing (Figure 1A). To elucidate which apoptotic signaling pathways were activated in response to sorafenib, established apoptotic markers, including caspases and members of the Bcl-2 protein family, were analyzed by immunoblotting. This analysis revealed that initiator caspases -2, -9, and -10, but not -8 were processed after exposure to sorafenib for 24 h (Figure 1B). Furthermore, activation of effector caspases-3, -6, and -7, as well as cleavage of the caspase substrate PARP1, were detected within 24 h. In addition, the caspase inhibitor XIAP was down-regulated after 24 h of exposure to sorafenib. Notably, caspase-7 and -10 were the only caspases that were activated already after 12 h. Caspase8 and -10 are considered to be induced primarily by the extrinsic apoptotic signaling pathway, while caspase-9 is activated upon mitochondrial outer membrane permeabilization and cytochrome c release. Caspase-7 is a known substrate of the early activated initiator caspase-10.20 The Bcl-2 family proteins are the main regulators of the intrinsic signaling pathway and are responsible for the integrity of the mitochondrial outer membrane. Consistent with previous results of sorafenib-induced cell death in various cell types,21 we found that the anti-apoptotic Bcl-2 family members Bcl-xL, Bcl-2, and Mcl-1 were down-regulated also in neuroblastoma cells (Figure 1B). Furthermore, the proapoptotic BH3-only proteins Bim and Bid were up-regulated upon exposure to sorafenib, suggesting a possible regulatory role in the apoptotic pathway. Bad levels were not affected. P21, an essential cell cycle repressor involved in p53-induced apoptosis,22 was up-regulated already after 12 h of incubation with sorafenib. Finally, reduction in the phosphorylation/activation of the RAF downstream target Erk1/2 (Thr202/Tyr204) confirmed that sorafenib 1611
dx.doi.org/10.1021/pr200790e | J. Proteome Res. 2012, 11, 1609−1620
Journal of Proteome Research
Article
Figure 1. Sorafenib-induced apoptosis in human neuroblastoma cells. (A) Human neuroblastoma cells (SH-SY5Y) incubated with sorafenib (sor) for 24 h underwent apoptotic cell death with classical morphology changes including cell blebbing and shrinking. (B) SH-SY5Y cells incubated with sorafenib for 12 and 24 h were blotted against apoptotic markers. Sorafenib induced activation of caspase (Casp)-2, -3, -6, -7, -9, and -10 but not -8 in human SH-SY5Y neuroblastoma cells. Antibodies against cleaved caspases are indicated with an asterisk. Caspase substrate PARP1 was cleaved, and caspase inhibitor XIAP was downregulated after 24 h. Anti-apoptotic Bcl-2 family proteins Bcl-xL, Bcl-2, and Mcl-1 were down-regulated, whereas propapototic Bim-EL and full length (FL) Bid levels increased upon exposure to sorafenib and Bad was unchanged. The levels of cell cycle arrest inducer p21 were increased already after 12 h. Phosphorylation of Raf kinase downstream target Erk1 and Erk2 (Thr202/Tyr204) was reduced after incubation with sorafenib. β-Tubulin (β-Tub) was used as loading control. (C) SH-SY5Y cells treated with sorafenib with or without caspase inhibitor zVAD-fmk were stained with Annexin V and PI and measured by flow cytometry. Caspase inhibition partially inhibited sorafenib-induced cell death.
regulation. Cells incubated with sorafenib were harvested simultaneously with untreated cells labeled with heavy amino acids and were mixed in a protein ratio of 1:1. Three and four biological replicates treated with sorafenib for 12 and 24 h, respectively, were run on SDS-PAGE for protein separation. In-gel trypsin digestion was followed by liquid chromatography coupled mass spectrometry (LC−MS) analysis. Using a nano-LC-LTQ-Orbitrap system, 1,500−2,000 proteins were identified from each lysate and resulted in identification of approximately 2,900 different proteins in total. In order to minimize the number of false positive protein identifications and quantifications, the data for each protein entry had to fulfill a relatively strict set of criteria and resulted in 193 significantly regulated unique proteins (Supplementary Table 1). More than 40% of the regulated proteins were found to be significantly altered at both time points. Notably, the threshold for significant regulation varied for the two time points because the statistical analysis was performed individually.
efficiently inhibited MAPK pathway (Figure 1B). To test if caspase activation is required for mediating sorafenib-mediated cell death, we pretreated SH-SY5Y cells with the broadspectrum caspase inhibitor zVAD-fmk simultaneously with sorafenib. Cell death was monitored by measuring exposure of phosphatidylserine on the outer cell membrane (Annexin V-FITC binding) and membrane damage (PI permeabilization). After 20 h, 56% of sorafenib-treated cells were double positive indicating late apoptosis or secondary necrosis. This population is reduced to 24% when the cells were pretreated with zVAD-fmk, showing that zVAD-fmk can partially inhibit sorafenib-induced cell death (Figure 1C). Temporal Quantitative Proteome Analysis of SH-SY5Y Cells Exposed to Sorafenib
Even though the interest in sorafenib has accelerated since the clinical approval of the drug in 2005 for advanced renal cell carcinoma, the molecular responses initiated by the drug are not well characterized. To uncover the molecular machinery driving this process, we performed a quantitative analysis of the proteomic changes occurring upon exposure to sorafenib of human neuroblastoma cells. The quantitative comparison was performed by growing SH-SY5Y cells in media containing light (12C6) or heavy isotopes (13C6) of the amino acids arginine and lysine. Total incorporation of the heavy arginine and lysine and exclusion of arginine to proline conversion was confirmed by rapid determination of amino acid incorporation as previously described.9 SH-SY5Y cells incorporated with the light arginine and lysine isotopes were incubated with sorafenib for 12 and 24 h to investigate the affects of sorafenib at different time points of
Mitochondrial proteins were significantly affected upon exposure to sorafenib
To examine which cellular processes and protein signaling pathways were significantly affected by sorafenib, the 193 regulated proteins were analyzed for annotation enrichment and functional coupling using data analyzing tools such as DAVID/ BiNGO and FunCoup. DAVID and BiNGO were used to detect which biological processes contained more enriched biological processes than what was expected from a random list of genes/proteins. Gene Ontology (GO) term analysis of the term “cellular compartment” revealed that a large fraction of the regulated proteins 1612
dx.doi.org/10.1021/pr200790e | J. Proteome Res. 2012, 11, 1609−1620
Journal of Proteome Research
Article
(39%) were associated with mitochondria. In total, 72 GO terms were found to be overrepresented for the 193 proteins
(Figure 2A, small inset and Supplementary Table 2), and 24 of these were directly related with mitochondria (Figure 2A,
Figure 2. Enrichment of mitochondrial proteins in the sorafenib-induced proteome. (A) A statistical analysis of the gene ontology (GO) term “cellular compartment” was performed using BiNGO plugin in Cytoscape. A significant portion of the proteins regulated after exposure to sorafenib was associated with mitochondria (zoomed area); 72 different GO terms were defined as overrepresented (small inset). The node size corresponds to the number of proteins that are assigned with the individual term. The terms with a p-value below 5.00 × 10−2 were defined as significant (yellow), and a darker color represents a lower p-value (dark orange
