The targeted therapeutics sorafenib and vorinostat interact in a synergistic fashion to kill carcinoma cells by activating CD95, and this drug combination is entering phase I evaluation. In this study, we determined how CD95 is activated by treatment with this drug combination. Low doses of sorafenib and vorinostat, but not the individual drugs, rapidly increased reactive oxygen species (ROS), Ca2+, and ceramide levels in gastrointestinal tumor cells. The production of ROS was reduced in Rho zero cells. Quenching ROS blocked drug-induced CD95 surface localization and apoptosis. ROS generation, CD95 activation, and cell killing was also blocked by quenching of induced Ca2+ levels or by inhibition of PP2A. Inhibition of acidic sphingomyelinase or de novo ceramide generation blocked the induction of ROS; however, combined inhibition of both acidic sphingomyelinase and de novo ceramide generation was required to block the induction of Ca2+. Quenching of ROS did not affect drug-induced ceramide/dihydro-ceramide levels, whereas quenching of Ca2+ reduced the ceramide increase. Sorafenib and vorinostat treatment radiosensitized liver and pancreatic cancer cells, an effect that was suppressed by quenching ROS or knockdown of LASS6. Further, sorafenib and vorinostat treatment suppressed the growth of pancreatic tumors in vivo. Our findings show that induction of cytosolic Ca2+ by sorafenib and vorinostat is a primary event that elevates dihydroceramide levels, each essential steps in ROS generation that promotes CD95 activation. Cancer Res; 70(15); 6313–24. ©2010 AACR.
In the United States, hepatoma and pancreatic carcinomas have 5-year survival rates of <10% and <5%, respectively (1, 2). These statistics emphasize the need to develop novel therapies against these lethal malignancies.
The extracellular signal-regulated kinase 1/2 (ERK1/2) pathway is frequently dysregulated in neoplastic transformation (3–5). The ERK1/2 module comprises, along with c-Jun NH2-terminal kinase (JNK1/2) and p38 mitogen-activated protein kinase (MAPK), members of the MAPK super family. These kinases are involved in responses to diverse mitogens and stresses, and have also been implicated in survival processes. Activation of the ERK1/2 pathway is generally associated with survival, whereas induction of JNK1/2 and p38 MAPK pathways generally signals apoptosis. Although the mechanisms by which ERK1/2 activation promote survival are not fully characterized, several antiapoptotic effector proteins have been identified, including increased expression of antiapoptotic proteins such as c-FLIP (6–11).
Sorafenib is a multikinase inhibitor that was originally developed as an inhibitor of Raf-1, but which was subsequently shown to inhibit multiple other kinases, including class III tyrosine kinase receptors such as platelet-derived growth factor (PDGF), vascular endothelial growth factor (VEGF) receptors 1 and 2, c-Kit, and FLT3 (12–14). The antitumor effects of sorafenib in renal cell carcinoma and in hepatoma have been ascribed to antiangiogenic actions of this agent through inhibition of the growth factor receptors (15–17). Several groups have shown in vitro that sorafenib kills human leukemia cells at concentrations below the maximum achievable dose (Cmax) of 15 to 20 μmol/L, through a mechanism involving the downregulation of the antiapoptotic BCL-2 family member MCL-1 (18, 19). In these studies sorafenib-mediated MCL-1 downregulation occurred through a translational rather than a transcriptional or posttranslational process that was mediated by endoplasmic reticulum (ER) stress signaling (20, 21). This suggests that the previously observed antitumor effects of sorafenib are mediated by a combination of inhibition of Raf family kinases, receptor tyrosine kinases that signal angiogenesis, and the induction of ER stress signaling.
Histone deacetylase inhibitors (HDACI) represent a class of agents that act by blocking histone deacetylation, thereby modifying chromatin structure and gene transcription. HDACIs promote histone acetylation and neutralization of positively charged lysine residues on histone tails, allowing chromatin to assume a more open conformation, which favors transcription (22). HDACIs also induce acetylation of other nonhistone targets, actions that may have plieotropic biological consequences, including inhibition of HSP90 function, induction of oxidative injury, and upregulation of death receptor expression (23–25). With respect to combinatorial drug studies with a multikinase inhibitor such as sorafenib, HDACIs are of interest in that they also downregulate multiple oncogenic kinases by interfering with HSP90 function, leading to proteasomal degradation of these proteins. Vorinostat (Zolinza) is a hydroxamic acid HDACI that has shown preliminary preclinical evidence of activity in hepatoma and other malignancies with a Cmax of ∼9 μmol/L (26–28).
We have recently published that sorafenib and vorinostat interact to kill in a wide range of tumor cell types through activation of the CD95 extrinsic apoptotic pathway (29, 30). The present studies have extended in greater molecular detail our analyses to understanding how sorafenib and vorinostat interact to promote CD95 activation.
Materials and Methods
Sorafenib tosylate (Bayer) and vorinostat (Merck) were provided by the Cancer Treatment and Evaluation Program, National Cancer Institute/NIH. Commercially available validated short hairpin RNA molecules to knock down RNA/protein levels were from Qiagen: CD95 (SI02654463; SI03118255). The kit to assay PP2A activity was purchased from Millipore. Reagents and performance of experimental procedures were described in refs. 20, 21, 29–37.
Cell line authentication
HEPG2, HEP3B, Mia PaCa2, and PANC1 cells were purchased from the American Type Culture Collection on a regular basis (fresh cryo-preserved vials of cells have been purchased at least once every 6 mo). Cells have not been authenticated in the corresponding author's laboratory.
In vitro cell treatments
Cells were isolated at the indicated times and subjected to trypan blue cell viability assay by counting in a light microscope, or alternatively, Annexin V/propidium iodide assays were carried to confirm our cell viability data (29–32).
Transfection of cells with small interfering RNA or with plasmids
Assessment of reactive oxygen species generation and cytosolic Ca2+ levels
Cancer cells were plated in 96-well plates. Cells were preincubated with dihydro-DCF (5 mmol/L for 30 min). Fluorescence measurements were obtained 0 to 30 minutes after drug addition with a Vector 3 plate reader. Data are presented corrected for basal fluorescence of vehicle-treated cells at each time point and expressed as a fold increase in reactive oxygen species (ROS) levels. Carcinoma cells, seeded in 96-well plates, with fura-2 acetoxymethylester as an indicator. The ratio of Fura-2 acetoxymethylester emissions, when excited at the wavelengths of 340 and 380 nm, was recorded, and analysis software were used to process and statistically analyze data.
Mass spectrometric analysis of sphingolipids and metabolites
Cells were washed extensively with PBS and detached by trypsinization. An aliquot of cells was taken for protein determination. For the remainder of cells, internal standards were added; lipids were extracted; and individual ceramide acyl chain species were quantified by liquid chromatography, electrospray ionization tandem mass spectrometry (29).
Recombinant adenoviral vectors; infection in vitro
Mia Paca2 cells were injected s.c. (1 × 107), and 14 days later, tumor volume was determined by calipers. Animals were segregated into four groups of approximate mean tumor volume ± SEM and then treated by oral gavage with vehicle or sorafenib+vorinostat. Animals are treated once per day for 5 consecutive days. Twenty-four hours after the first drug administration, animals are placed in a shielded container and either mock exposed or have the flanks irradiated (2 Gy). Forty-eight hours after the first irradiation, a second irradiation (2 Gy) is performed. Tumors were subjected to two cycles of drug treatment/radiation exposure (total dose 4 × 2 Gy). Tumor mass is determined at least every 3rd day after the initiation of drug treatment.
Comparison of the effects of various treatments was performed using ANOVA and the Student's t test. Differences with a P value of <0.05 were considered statistically significant. All experiments were performed on at least two separate occasions: each cell death or ceramide measurement experiment had between 3 and 4 separate independent experimental data points per experiment, and each ROS or Ca2+ measurement had at least 8 to 12 independent data points per experiment. Thus, experiments shown are the means of multiple individual data points per experiment from multiple experiments (±SEM).
Treatment of cells with low concentrations of sorafenib and vorinostat, but not the individual agents, promoted activation of CD95 and the generation of ROS (Fig. 1A and B). The rapid activation of CD95 was not reduced by incubation with a neutralizing antibody to inhibit FAS ligand (data not shown). The production of ROS was quenched using N-acetyl cysteine or MnTBAP (Fig. 1C). Mitochondria-deficient rho zero HuH7 cells were generated (29–32); cells lacking functional mitochondria exhibited a significantly reduced ability to generate ROS in response to sorafenib and vorinostat exposure (Fig. 1D).
Quenching of drug-induced ROS suppressed sorafenib and vorinostat toxicity (Fig. 2A and B). Quenching of ROS also suppressed CD95 activation (Fig. 2B, inset). Molecular quenching of ROS using thioredoxin (TRX) suppressed ROS generation, CD95 activation, and drug toxicity, whereas molecular enhancement of ROS generation through expression of a mutant-inactive TRX protein enhanced ROS generation, although not CD95 activation, and promoted drug toxicity (Fig. 2C). HuH7 hepatoma cells do not express CD95 and are relatively resistant to sorafenib and vorinostat-induced cell killing (29). Transfection of HuH7 cells with a wild-type CD95-YFP construct, but not a construct lacking two known sites of regulatory tyrosine phosphorylation CD95-YFP FF, facilitated sorafenib and vorinostat toxicity (Fig. 2D). Sorafenib and vorinostat treatment promoted tyrosine phosphorylation and cell surface localization of wild-type CD95-YFP but not of CD95-YFP YY-FF (data not shown; Fig. 2D).
Low concentrations of either sorafenib (1–3 μmol/L) or vorinostat (250–500 nmol/L) did not strongly increase either ROS or Ca2+ levels, whereas we did observe, at 9 to 12 μmol/L sorafenib and >1 μmol/L vorinostat concentrations, measurable effects of the individual drugs on ROS and Ca2+ levels (data not shown). Treatment of tumor cells with low doses of sorafenib and vorinostat enhanced cytosolic Ca2+ levels (Fig. 3A). Quenching of ROS did not block drug-induced cytosolic Ca2+ levels (Fig. 3B). Quenching of Ca2+, using either BAPTA-AM or expression of Calbindin D28, blocked drug-induced ROS induction (data not shown; Fig. 3C). Expression of Calbindin D28 blocked endogenous CD95 surface localization, CD95 tyrosine phosphorylation, and CD95 association with caspase-8 (Fig. 3D, top section). Expression of Calbindin D28 blocked sorafenib and vorinostat toxicity (Fig. 3D, bottom graph). Knockdown of CD95 neither blocked drug-induced cytosolic Ca2+ levels nor blocked capacitative Ca2+ entry into cells (Supplementary Figs. S1–S3).
Prior studies using sorafenib and vorinostat had implicated ceramide as a mediator of CD95 activation. Loss of acidic sphingomyelinase expression (ASMase −/−) caused a partial although statistically significant (P < 0.05) reduction in drug-induced cytosolic Ca2+ levels (Fig. 4A). In contrast to the partial effect observed on reducing Ca2+ levels, deletion of ASMase expression abolished drug-induced ROS production (Fig. 4B). Knockdown of ASMase or inhibition of the de novo ceramide synthase pathway each caused a partial reduction in drug-induced cytosolic Ca2+ levels, and complete blockade of ceramide generation abolished Ca2+ induction (Fig. 4C). Combined knockdown of ASMase expression and inhibition of de novo ceramide synthesis in hepatoma cells abolished sorafenib- and vorinostat-induced ROS levels (Fig. 4D).
We next wished to place the sequence of ceramide generation within the context of elevated ROS and Ca2+ levels. Knockdown of CD95 protected tumor cells from sorafenib+vorinostat toxicity (Fig. 5A). CD95 activation and sorafenib+vorinostat toxicity were blocked by knockdown of ASMase and by inhibition of de novo ceramide synthesis (Fig. 5B). Knockdown of ceramide synthase 6 (LASS6) expression abolished drug-induced activation of CD95 and significantly reduced drug toxicity (Fig. 5C). Inhibition of ROS generation did not alter sorafenib+vorinostat-induced C16 ceramide or dihydro-ceramide levels (Fig. 5D). However, inhibition of drug-induced cytosolic Ca2+ levels or knock down of LASS6 significantly reduced the amount of C16 dihydro-ceramide generated following drug treatment (data not shown; Fig. 5D). In parallel studies using SW620 colon cancer cells that lack LASS6 expression, transfected to express LASS6, we noted that reexpression of ceramide synthase 6 restored drug-induced ROS generation levels (Supplementary Fig. S4).
As sorafenib+vorinostat exposure generated ceramide, and the protein serine/threonine phosphatase PP2A can play a pivotal role in ceramide-dependent signaling, as well as in the regulation of ROS generation, we determined whether modulation of PP2A function affected the actions of this drug combination (29, 37). Sorafenib and vorinostat activated PP2A in a LASS6-dependent fashion (Fig. 6A). Expression of a wild-type PP2A inhibitory protein, whose function can be blocked by ceramide or a mutant inhibitor protein, whose function cannot be altered by ceramide, suppressed drug combination–induced ROS generation and cell death (Fig. 6B and C). Phosphorylation of mitochondria-localized signal transducers and activators of transcription 3 (STAT3) S727 plays a key role in regulating the rate of mitochondrial respiration and de facto mitochondrial ROS production, and STAT3 S727 is a PP2A substrate (38). Sorafenib and vorinostat treatment reduced STAT3 S727 phosphorylation by ∼50% within 2 hours (Fig. 6D). Expression of a dominant-negative form of STAT3 S727A enhanced sorafenib and vorinostat lethality, whereas expression of activated STAT3 S727D suppressed drug toxicity (Fig. 6D).
Sorafenib and HDACI drug combination therapy is entering clinical phase I evaluation. Radiotherapy is known to generate ROS and ceramide, and it is a primary therapeutic modality for pancreatic cancer. Sorafenib and vorinostat treatment radiosensitized pancreatic and liver cancer cells in vitro (Supplementary Figs. S5 and S6). Radiosensitization was blocked by expression of TRX or knockdown of LASS6 expression (Supplementary Fig. S5). Treatment of animals carrying MiaPaca2 tumors with sorafenib and vorinostat significantly reduced tumor growth in vivo, which validates this drug combination as a possible therapeutic for pancreatic cancer (P < 0.05; Supplementary Fig. S7). Tumor growth was blunted by radiation exposure; however, radiation exposure did not significantly further enhance the antitumor effects of sorafenib and vorinostat treatment on tumor regrowth. Further studies will be required to determine whether the in vivo administration of radiotherapy combined with drug treatment is schedule dependent.
The present studies attempted to determine in detail the molecular mechanisms by which sorafenib and vorinostat interacted to activate CD95 and promote drug-induced toxicity.
Low concentrations of sorafenib and vorinostat interacted in a greater than additive manner to increase ROS levels. The induction of ROS occurred in cells lacking expression of CD95, strongly arguing against death receptor–induced activation of NADPH oxidases in this process. In rho zero cells, the induction of ROS was reduced, arguing that mitochondria represent the major source of drug-induced ROS generation. Quenching of ROS blocked CD95 activation and drug combination–induced cell death. Sorafenib and vorinostat treatment increased cytosolic Ca2+ levels in an ROS-independent fashion, and quenching of Ca2+ blocked the increase in ROS levels, CD95 activation, and tumor cell killing. Inhibition of ASMase or of de novo ceramide generation almost abolished drug combination–induced ROS levels and CD95 activation, but individually blockade of either pathway only caused a partial suppression in drug-induced cytosolic Ca2+. Quenching of Ca2+, or knockdown of ceramide synthase 6, reduced drug-induced ceramide levels. Thus, our data argue for drug combination–induced Ca2+ signaling (primary) and ceramide generation (secondary) as two interdependent signaling processes that regulate each other, and that both signals act in concert to promote ROS generation, which is essential for drug-induced CD95 activation and tumor cell death (Supplementary Fig. S8).
Bile acids can promote ligand-independent, ASMase, and ceramide-dependent activation of CD95 in hepatocytes (11, 34–39). The generation of ceramide has been shown by many groups to promote ligand-independent activation of growth factor receptors through the clustering of these receptors and other signal facilitating proteins into lipid rich domains (39). The six known ceramide synthase genes (LASS) are localized in the ER, and different LASS proteins have been noted to generate different chain length ceramide forms (40). Based on our data, with increases in C16 dihydro-ceramide levels, it was probable that vorinostat and sorafenib were modulating the activities of LASS6 and LASS5, respectively (40). We identified ceramide synthase 6 (LASS6) as an essential enzyme in the drug-induced induction of ceramide, in CD95 activity, and in tumor cell killing. Treatment of cells with vorinostat increased the acetylation of LASS6, suggesting that one mechanism of drug-induced LASS6 activity may be mediated in part through this process.7 It is well known that other nonhistone proteins are functionally regulated by reversible acetylation, notably NFκB and HSP90 (22–25). Additional studies will be required to define the precise site of acetylation in LASS6 and to discover whether it truly is a site of regulatory acetylation.
The generation of ROS was significantly reduced in rho zero cells, and we have documented a key role for mitochondria in the production of ROS in response to various agents (e.g., ref. 41). In the present studies, ROS production was a tertiary effect dependent on the actions of ceramide and Ca2+. It has been widely documented that the release of Ca2+ from the ER in to the cytosol rapidly alters Ca2+ levels in mitochondria, and that changes in Ca2+ fluxes within mitochondria can stimulate ROS production (e.g., ref. 42). Ceramide levels, independently of Ca2+ fluxes, are causal in regulating mitochondrial ROS production, and we noted that whereas inhibition of both ASMase and de novo ceramide generation was required to suppress Ca2+ signaling, inhibition of either de novo or ASMase actions blocked ROS production. There are several probable modes of ceramide action at the mitochondrion that could alter respiratory chain complex activities, including modulation of membrane fluidity as well as activation of PP2A (37, 43–45). The mitochondrial respiratory chain complexes, as well as BCL-2 family proteins, are heavily Ser/Thr phosphorylated, and ceramide has the potential to regulate their function through PP2A. In this regard, drug treatment reduced STAT3 S727 phosphorylation, and inhibition of PP2A activity blocked sorafenib+vorinostat-induced ROS and suppressed drug toxicity. The additional molecular targets of PP2A beyond STAT3 in the regulation of mitochondrial ROS generation will be explored in future studies.
Ligand-independent activation of CD95 is a complex process, including tyrosine phosphorylation, altered subcellular localization and formation of the DISC, i.e., association with FADD and procaspase-8 (37, 38). In our system, drug-induced activation of CD95, as defined by formation of the DISC or by plasma membrane localization of CD95, was blocked by quenching of Ca2+, suppressing ceramide formation or by quenching ROS production. Tyrosine phosphorylation of CD95 was essential for drug-induced membrane localization and DISC formation, but tyrosine phosphorylation of CD95 was only suppressed by quenching Ca2+. Thus, at least two independent signals downstream of drug-induced Ca2+ signaling act to promote ligand-independent activation of CD95 in GI tumor cells. The regulation of CD95 tyrosine phosphorylation has been proposed to be catalyzed by ERBB1 and by Src family tyrosine kinases (38).
An additional question that remains unresolved is how sorafenib and vorinostat treatment promote Ca2+ signaling, and what the putative primary effectors for these drugs are. Two likely targets of sorafenib and vorinostat may be LASS6 and the IP3 receptors that regulate ER Ca2+ fluxes. At present, little is known about how the ceramide synthase proteins are enzymatically regulated, and it is possible that sorafenib reduces phosphorylation at a regulatory site or, alternatively, that vorinostat promotes LASS6 acetylation, which facilitates enzyme activation. Whether Ca2+-calmodulin signaling directly or through CaM kinase actions modulates ceramide synthase function is unknown. It is also possible that the drug combination causes inactivation of ceramidase enzyme activities that act to lower ceramide levels. IP3 levels are regulated by phospholipase C and phosphodiesterase enzymes; at present, the effect of sorafenib and vorinostat exposure on the activities of peritoneal lymphocytes (PLC) enzymes is unclear, and data from many groups arguing that sorafenib inhibits receptor tyrosine kinases such as the PDGF and VEGF receptors, and c-Kit would a priori predict for reduced activity of PLC enzymes that would reduce IP3 levels (14, 15).
Sorafenib and HDACI combination therapy is entering phase I evaluation at both Virginia Commonwealth University and at Medical University of South Carolina in a variety of solid and liquid tumor types. We have noted that sorafenib and vorinostat combination therapy shows in vitro efficacy against multiple pancreatic cancer cell lines (29, 30, 46). The present studies showed that sorafenib and vorinostat treatment suppressed the growth of pancreatic tumor cells in vivo, potentially validating this drug combination as a new therapeutic for pancreatic cancer. Further studies, however, will be required to determine whether sorafenib and HDACI combination therapy can radiosensitize GI tumors in a sequence-dependent fashion in vivo.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant Support: This work was funded by PHS grants R01-DK52825, P01-CA104177, and R01-CA108520 (P. Dent).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Received March 23, 2010.
- Revision received May 24, 2010.
- Accepted June 3, 2010.
- ©2010 American Association for Cancer Research.