Induction of Apoptosis in Human Cancer Cells by ... - Semantic Scholar

5 downloads 0 Views 863KB Size Report
Sep 3, 2008 - CAN for as long as 18 hours ( Supplementary Figure 2, available online); thus, the reduction in HIF-1 α protein levels was likely to be.
ARTICLE

Induction of Apoptosis in Human Cancer Cells by Candidaspongiolide, a Novel Sponge Polyketide Daniela Trisciuoglio, Badarch Uranchimeg, John H. Cardellina, Tamara L. Meragelman, Shigeki Matsunaga, Nobuhiru Fusetani, Donatella Del Bufalo, Robert H. Shoemaker, Giovanni Melillo

Background

Candidaspongiolide (CAN), a novel polyketide from a marine sponge, is the active component of a mixture that was found to be potently cytotoxic in the National Cancer Institute’s 60-cell-line screen.

Methods

Effects of CAN on U251 glioma and HCT116 colorectal cancer cells and on normal fibroblasts were assessed using radiolabeling studies to measure protein synthesis, clonogenic assays to measure cell survival, flow cytometry of annexin V– and propidium iodide–stained cells to measure apoptosis, and western blots in the presence or absence of specific inhibitors to assess accumulation and phosphorylation of potential downstream target proteins.

Results

CAN inhibited protein synthesis and potently induced apoptosis in both U251 and HCT116 cells, the latter in part by a caspase 12–dependent pathway. For example, 25%–30% of U251 or HCT116 cells became apoptotic after 24 hours of treatment with 100 nM CAN. CAN also rapidly induced sustained phosphorylation of eukaryotic translation initiation factor-2 (eIF2)-␣ at Ser51 and of the translation elongation factor eEF2 at Thr56, which could contribute to its dose-dependent inhibition of protein synthesis. Stable expression of dominant-negative eIF2␣ was sufficient to prevent CAN-induced eIF2␣ phosphorylation and induction of apoptosis but insufficient to prevent inhibition of protein synthesis. CAN induction of eIF2␣ phosphorylation did not occur by a classic endoplasmic reticulum stress pathway. However, an inhibitor of and smallinterfering RNAs to the double-stranded RNA–dependent protein kinase PKR prevented CAN-mediated eIF2␣ phosphorylation and apoptosis, respectively. Although CAN inhibited protein synthesis in both cancer cells and normal human fibroblasts, it induced eIF2␣ phosphorylation and apoptosis only in cancer cells.

Conclusions

CAN triggers PKR/eIF2␣/caspase 12–dependent apoptosis and inhibits protein synthesis in cancer cells but only inhibits protein synthesis in normal cells. J Natl Cancer Inst 2008;100:1233–1246

Candidaspongiolide (CAN) is a novel natural product recently reported from extracts of marine sponges, Candidaspongia sp., from Australia and Papua New Guinea (1,2). Sponge extracts containing CAN were identified as potent cytotoxins that preferentially killed melanoma and glioma cells in the National Cancer Institute’s (NCI) 60-cell-line screen (2). This cytotoxicity was tracked to a series of related polyketide macrolides, candidaspongiolide (structure shown in Supplementary Figure 1, available online), and a group of derived fatty acid esters. CAN belongs to the small family of tedanolide macrolides, which includes tedanolide (3), 13deoxytedanolide (13-DT) [(4); Supplementary Figure 1, available online], and tedanolide C (5), all potent compounds that are cytotoxic in the nanomolar to subnanomolar range against various cancer cell lines (3,5). 13-DT is the best characterized representative of the group (4). It binds to the 60S ribosomal subunit and thereby induces a “ribotoxic stress” response that triggers phosphorylation of the stress-activated protein kinases p38 and jun NH2-terminal kinase (JNK) and inhibits protein synthesis in eukaryotic cells (6–8). However, the ribotoxic stress response may not account fully for the cytotoxicity of tedanolide macrolides and jnci.oxfordjournals.org

that mechanism remains to be defined. Despite considerable focus on the synthesis of this class of compound (9), a continuing lack of supply from either natural or synthetic sources has severely hampered efforts to develop these potential antitumor agents. Recently, we identified the CAN mixture as a potent inhibitor of hypoxia-inducible factor-1 (HIF-1) transcriptional activity in a cellbased high-throughput screen and confirmed inhibition of HIF-1 in three cancer cell lines (Supplementary Figure 2, A and B, available

Affiliations of authors: Tumor Hypoxia Laboratory, SAIC-Frederick, Inc. (DT, BU, GM), and Screening Technologies Branch, Developmental Therapeutics Program (JHC, TLM, RHS), National Cancer Institute at Frederick, Frederick, MD; Laboratory of Marine Biochemistry, University of Tokyo, Tokyo, Japan (SM, NF), Experimental Chemotherapy Laboratory, Regina Elena Cancer Institute, Rome, Italy (DT, DDB). Correspondence to: Giovanni Melillo, MD, DTP-Tumor Hypoxia Laboratory, Bldg 432, Rm 218, National Cancer Institute, Frederick, MD 21702 (e-mail: [email protected]). See “Funding” and “Notes” following “References.” DOI: 10.1093/jnci/djn239 Published by Oxford University Press 2008.

JNCI

|

Articles 1233

CONT E X T A N D C A VEAT S Prior knowledge Candidaspongiolide (CAN), a tedanolide macrolide, was discovered to be the active compound in an extract from marine sponges that killed glioma and melanoma cells in the NCI 60-cell-line screen. The mechanism of cytotoxicity was not known, although related compounds have been reported to inhibit protein synthesis. Study design Cell and molecular assays were done to examine the effects of CAN on protein synthesis and apoptosis pathways in U251 glioma cells and HCT116 colorectal cancer cells, as well as in normal cells. Contribution CAN inhibited protein synthesis and induced apoptosis in the tested cancer cell lines by a mechanism that included activation of caspase 12 and protein kinase PKR and inhibitory phosphorylation of eukaryotic initiation factor-2 (eIF2)-␣. CAN also appeared able to inhibit protein synthesis by additional pathway(s). In the normal cell lines, CAN inhibited protein synthesis but did not activate the PKR/eIF2␣ pathway or apoptosis. Implications The in vitro activity of CAN raises the possibility that it will have antitumor activity against tumors in vivo. Limitations Further testing of CAN’s safety and efficacy in treating tumors in mouse models will be necessary before its usefulness in humans can be evaluated. From the Editors

online). We interpreted the reduction in HIF-1␣ protein levels by CAN as a result of the known ability of tedanolide macrolides to inhibit protein synthesis. However, we also noticed that HIF-1␤ and ␤-actin, proteins with a longer half-life than HIF-1␣, were not similarly affected, even when cells were exposed to as much as 100 nM CAN for as long as 18 hours (Supplementary Figure 2, available online); thus, the reduction in HIF-1␣ protein levels was likely to be attributable to the short half-life of the protein. Based on these preliminary results, we decided to further investigate whether CAN, like other tedanolide macrolides, inhibited protein synthesis overall and whether that mechanism and/or others might be responsible for CAN’s potent cytotoxic activity in the NCI 60-cell-line screen. Protein translation is often increased by genetic alterations in human cancers and by dysregulation of signal transduction pathways that contribute to cancer progression (10,11). Translation initiation is the main step in the regulation of protein synthesis, and several factors regulate this critical process (12). Eukaryotic translation initiation factor-2 (eIF2) mediates binding of initiator transfer RNAs to the small ribosomal unit and is a key regulator of translation initiation (13). Phosphorylation of the ␣ subunit of eIF2 is a well-documented mechanism for the inhibition of protein synthesis (14). Among other kinases, both double-stranded RNAdependent protein kinase (PKR) and PKR-like endoplasmic reticulum kinase (PERK) can phosphorylate and inhibit eIF2␣ to inhibit most translation during conditions of cellular stress. The eukaryotic initiation factor 4E (eIF4E) binding proteins (4E-BPs), which sequester eIF4E from eIF4G and control recognition of 1234 Articles

|

JNCI

mRNA via the 5′ cap, are also important regulators of translation initiation (15–17). The kinase mammalian target of rapamycin (mTOR) can activate 5′ cap–dependent translation by phosphorylating 4E-BP1 and preventing it from binding and inhibiting eIF4E (18). Apoptosis, or programmed cell death, is also frequently dysregulated in cancer cells (19). In the extrinsic pathway of apoptosis, extracellular ligands activate death receptors such as tumor necrosis factor receptor (TNFR) and Fas at the cell surface, resulting in the activation of the cysteine protease caspase 8. In the intrinsic pathway of apoptosis, apoptotic stimuli such as radiation, toxins, viral infections, hypoxia, or withdrawal of growth factors result in changes at the mitochondrial membrane that cause release of proteins, including cytochrome c, and subsequent activation of caspase 9 via Apaf-1. Both caspases 8 and 9 activate the downstream “executioner” caspase 3, which causes endonuclease activation, leading to chromatin condensation; cytoskeletal reorganization, leading to formation of apoptotic cell bodies; and externalization of phosphatidylserine to the outer plasma membrane, allowing detection of apoptotic cells using annexin V. Stress conditions, when they are unresolved by measures that are meant to protect cells, such as translational inhibition, can lead to apoptosis by additional pathways that are incompletely understood. For example, in the presence of unresolved endoplasmic reticulum (ER) stress due to conditions such as hypoxia, several pathways can lead to apoptosis (20,21). In one such pathway, the ER protein IRE1 recruits c-jun NH2-terminal inhibitory kinase and TNFR-associated factor-2, which leads to the activation of apoptosis-signaling kinase 1, and JNK. In another pathway, caspase 12 is activated at the ER membrane; it then activates caspase 9 to activate caspase 3 in an apoptotic pathway that does not involve mitochondria or death receptors. In yet another pathway, the ER protein PERK, while inactivating eIF2␣ and thereby inhibiting most translation, promotes translation of the transcription factor ATF4, which induces the proapoptotic transcription factor C/EBP-homologous protein (CHOP). Alternatively, in virus-infected cells or in response to lipopolysaccharide, serum deprivation, or TNF-␣ treatment, among other stressors, the interferon-inducible double-stranded RNA–regulated protein kinase PKR can mediate the induction of apoptosis (22,23). PKRmediated apoptosis appears to often proceed through phosphorylation of eIF2␣, which can elevate ATF4 and CHOP protein levels. However, PKR-mediated apoptosis may also involve other pathways, including the nuclear factor-кB pathway, which can be proapoptotic under certain conditions, and/or the FADD/caspase 8 and Apaf-1/caspase 9 pathways. In this study, we examined the effects of CAN on both global protein synthesis, as a measure of translation, and on apoptosis in human cancer cells. We then looked at CAN-mediated apoptosis in the presence of inhibitors of caspases 3, 8, 9, and 12 to determine which pathways might be necessary for CAN-mediated cytotoxicity. Next, we looked at the p38 and JNK pathways and at eIF2␣ phosphorylation to determine whether they might mediate the translational and/or apoptotic activities of CAN. The effects of CAN on the expression or activation of various proteins modulated by ER stress, including ATF4, CHOP, and PERK, as well as its effect on PKR activity, were examined. A nonphosphorylatable Vol. 100, Issue 17

|

September 3, 2008

mutant of eIF2␣ and inhibitors of PKR were then introduced into U251 cells to determine whether these could block CAN-mediated apoptosis. Last, we examined whether CAN was able to inhibit translation and induce apoptosis in normal cells.

Materials and Methods Cell Lines and Culture Conditions Cell lines were obtained from the Developmental Therapeutics Program of the National Cancer Institute. U251 human glioma cells were cultured in RPMI 1640 medium supplemented with 5% heatinactivated fetal bovine serum (FCS), 50 IU/mL penicillin, 50 µg/mL streptomycin, and 2 mM glutamine. MCF7 human breast adenocarcinoma (used only in Supplementary Figure 2) and HCT116 human colorectal carcinoma cells were cultured in RPMI 1640 medium supplemented with 10% FCS. AG06858 and AG068103, normal human fibroblast cell lines, were cultured in Dulbecco’s minimal essential medium with 10% FCS. Cells were normally grown at 37°C in 5% CO2 and ambient oxygen (normoxic) conditions. For those experiments requiring hypoxia, hypoxic conditions were achieved in an InVivo2 400 hypoxic workstation (Ruskinn Technologies, Cincinnati, OH) set to deliver a 1% oxygen atmosphere in the presence of 5% CO2 at 37°C. Exponentially growing HCT116, U251, and MCF7 cells were seeded at a density of 106 cells per 100-mm dish and incubated for 24 hours in complete medium. The next day, cells were exposed for 6 or 18 hours to normoxia or hypoxia in the presence or absence of different doses of CAN. Reagents CAN was isolated at the Developmental Therapeutics Program, NCI-Frederick, by a process that included two-dimensional nuclear magnetic resonance and mass spectrographic analysis (2). In our studies we used the purified compound. CAN was dissolved in dimethyl sulfoxide and stored in aliquots at –20°C. CAN was routinely added to cells 30 minutes before incubation under hypoxia or before the addition of other agents, unless otherwise indicated. The translational inhibitor bis-cycloheximide oxaldihydrazone (CHX, 40 µg/mL for 2 hours), protesome inhibitor carbobenzoxyleucyl-leucyl-leucinal-ZLLal (MG132, 10 µM), PKR inhibitor 2aminopurine (2-AP) (5 mM), p38 kinase inhibitor SB202190 (10 µM), and JNK inhibitor SP600125 (10 µM) were obtained from Sigma Chemical Company (St Louis, MO). All inhibitors were routinely added to the cells 30 minutes before the addition of CAN. For apoptosis assays, HCT116 cells were treated with 50 nM CAN for 48 hours in the absence or presence of 10 µM Ac-DEVD-fmk (caspase 3 inhibitor), 10 µM Ac-ZIETD-fmk (caspase 8 inhibitor), 10 µM Ac-ZLEHD-fmk (caspase 9 inhibitor), 10 µM Ac-ZATAD-fmk (caspase 12 inhibitor), or 10 µM Z-VAD-fmk (pan-caspase inhibitor) before annexin V and propidium iodide (PI) staining analysis by flow cytometry (see below). Specific inhibitors of caspase 3, caspase 8, and caspase 9 and the total caspase inhibitor Z-VAD-fmk were obtained from Calbiochem (San Diego, CA). The specific inhibitor of caspase 12 was obtained from Biovision (Mountain View, CA). Cells treated with thapsigargin (2 µM) and tunicamycin (20 µM) were used as positive controls for caspase 12 activation. In contrast, camptothecin (1 µM), which did jnci.oxfordjournals.org

not cause activation of caspase 12, was used as negative control. Thapsgargin, tunicamycin, and camptothecin were all obtained from Sigma. Clonogenic Cell Survival Assay For the clonogenic assay, cells that had been grown to confluence in six-well plates were incubated with different concentrations of CAN for 6 hours. Immediately after treatment with CAN, cells were plated in 60-mm culture dishes at a density of 500 cells per dish in RPMI medium containing 10% FCS and then kept in a humidified incubator at 37°C and 5% CO2 for 2 weeks. Colonies were fixed and stained with 1% crystal violet (Sigma) in 10% methanol and counted. Percentage of colonies arising from surviving treated cells was calculated relative to colonies arising from untreated control cells. Generation of U251 Cells Expressing Dominant-Negative eIF2␣ A nonphosphorylatable mutant form of eIF2␣ (S51A), cloned at the HindIII site of plasmid pRc/CMV (Invitrogen, Carlsbad, CA), was a kind gift of C. Koumenis (24). Exponentially growing U251 cells were seeded (2 × 105 cells per well in six-well plates) and incubated for 24 hours in complete medium. The next day, cells were transfected with 1 µg of plasmid pCMV-neo or pCMV-eIF2␣-S51A using Effectene (Qiagen, Hilden, Germany). After 48 hours, cells were detached with trypsin, counted, and plated in RPMI complete medium on 100-mm dishes at three different cell densities: 10 000, 1000, and 100 cells per plate. After another 24 hours, 800 µg/mL G418 (Geneticin, Invitrogen) was added to the medium, and the cells were incubated in the presence of the drug (replaced with fresh medium every 4 days) for a total of 2 weeks. Individual G418resistant clones containing fewer than 200 cells per clone were isolated and expanded. The pooled clones were routinely cultured in the presence of G418 (100 µg/mL) to minimize the expansion of revertant cells. Small-Interfering RNA Transfection Small-interfering RNAs (siRNAs) that silence the expression of protein kinase RNA-activated (PKR siRNA, No. 11185) or endoplasmic reticulum (ER) stress–activated protein kinase (PERK siRNA, No. 18102) were obtained from Ambion (Austin, TX), and control (nonsilencing) siRNA was obtained from Qiagen. U251 cells were transfected with 100 nM siRNA. Transfection was performed using oligofectamine reagent (Invitrogen) according to the manufacturer’s instructions. Following transfection, cells were allowed to recover for 48 hours before subsequent treatment. Protein Extraction and Western Blot Analysis For western blot analysis, HCT116, U251, and MCF7 cells that had been grown in 100-mm dishes were incubated with CAN or inhibitors for the indicated times. Cells were then washed, harvested, and washed twice with ice-cold phosphate-buffered saline (PBS), and proteins were extracted in 100 µL of ice-cold RIPA lysis buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1% NP40, 0.25% sodium deoxycholate, 1 mM PMSF, with 1 × complete mini protease inhibitor [Boehringer, Mannheim, Germany]). Extracts were kept on ice for 15 minutes and centrifuged at JNCI

|

Articles 1235

10 000 × g for 15 minutes at 4°C, and the supernatants containing the clarified protein extracts were stored at ⫺80°C. Thawed protein extracts were subjected to electrophoresis on 4%–20% sodium dodecyl sulfate–polyacrylamide gels (80 µg protein per lane). Proteins were then electroblotted onto PVDF membranes (Invitrogen), which were incubated with any of the following primary antibodies: mouse monoclonal antibody to HIF-1␣ (1:250 dilution, BD Biosciences, Franklin Lakes, NJ), rabbit polyclonal antibody to caspase 12 (1:1000, BD Biosciences); rabbit polyclonal antibodies to caspase 3, poly (ADP-ribose) polymerase (PARP), phospho-eEF2 (Thr56), total eEF2, phospho-eEF2k (Ser366), total eEF2k, total eIF4E, phospho-4E-BP1 (Thr37/46), total 4EBP1, phospho-eIF-2␣ (Ser51), total eIF2␣, phospho-JNK (Thr183/Tyr185), total JNK, phospho-p38 mitogen-activated protein kinase (MAPK) (Thr180/Tyr182), total p38 MAPK, and phospho-PKR (Thr451) (all used 1:1000, Cell Signaling Technology, Beverly, MA); rabbit polyclonal antibodies to PERK, ATF4, or GADD153/CHOP, or mouse monoclonal antibody to PKR (all used 1:500, Santa Cruz Biotechnology, Santa Cruz, CA); or mouse monoclonal antibody to actin (1:1000, ICN Biomedicals, Irvine, CA). Anti-mouse immunoglobulin G (IgG)-horseradish peroxidase (HRP)–conjugated, anti-goat IgG-HRP–conjugated, or anti-rabbit IgG-HRP–conjugated antibodies were used as secondary antibodies at 1:10 000 dilution. Antibody binding was visualized by enhanced chemiluminescence according to manufacturer’s specification and recorded on autoradiography film (Amersham Biosciences, Freiburg, Germany). Flow Cytometric Analysis of Apoptosis and Cell Cycle Apoptosis was assayed by staining cells with the annexin V– fluorescein isothiocyanate (FITC) apoptosis kit (BD Bioscience) according to the manufacturer’s instructions. Cells double-stained for both annexin V and PI were analyzed by flow cytometry using a FACScan (Becton Dickinson GmbH, Heidelberg, Germany). PI was used in conjunction with annexin V–FITC to distinguish cells in the earlier stages of apoptosis (annexin V-FITC positive, PI negative) from those in later stages of apoptosis or that were already dead (annexin V–FITC positive, PI positive). (Gated cells were plotted on a dot-plot showing annexin V staining and PI staining.) To measure the percentage of apoptotic cells, cells floating in the culture supernatants were collected by centrifugation and pooled with adherent cells recovered from the plates, fixed in cold 70% ethanol, and stained with PI (0.5 mg/mL in PBS). Cell cycle analysis was performed as previously described (25). Caspase 12 activation was assessed using a fluorescent substrate kit (Biovision). The assay is based on detection of cleavage of the substrate ATAD-AFC (ATAD: acetyl-alanine-threoninealanine-aspartic acid; AFC: 7-amino-4-trifluoromethyl coumarin). Briefly, U251 cells that had been grown in 100-mm dishes and were incubated with CAN or inhibitors for the indicated times were washed, harvested, and washed twice more with icecold PBS. Then, 105 cells were incubated in a test tube with 5 µL of ATAD-AFC substrate in PBS for 1 hour at 37°C and 5% CO2, then washed again with PBS, and immediately analyzed by flow cytometry. Samples were gated using forward scatter and side scatter to exclude cell debris not within normal cell size using CellQuest software (BD Biosciences). 1236 Articles

|

JNCI

Analysis of Protein Synthesis U251 cells that had been grown in 100-mm dishes and were then incubated with CAN or inhibitors for the indicated times were labeled with a mixture of [35S]-methionine and [35S]-cysteine ([35S]Met/Cys; 100 µCi/mL, 1000 Ci/mmol; Amersham/GE Healthcare, Piscataway, NJ) for 20 minutes in Met/Cys-free minimal essential medium (Invitrogen). After washing with PBS, cell extracts were prepared by lysing the cells in RIPA lysis buffer. The [35S] incorporation was measured by trichloroacetic acid (TCA) precipitation. In brief, equal amounts of cell lysates were added to 0.5 mL of 0.1 mg/mL bovine serum albumin containing 0.02% sodium azide and placed on ice. Ice-cold 20% TCA (0.5 mL) was added, and the mixture was vortexed vigorously and incubated for 30 minutes on ice. The samples were washed twice with 1 mL of ice-cold 10% TCA and twice with 1 mL of 100% ethanol. The pellets were solubilized in 1 mL of 1 M Tris–HCl, pH 8.0, and boiled for 5 minutes, and the radioactivity was measured in a scintillation counter (Packard Instrument Co., Meriden, CT). Results were normalized as the percent increase or decrease of [35S]-Met/Cys incorporation in treated cells compared with that in untreated control cells. Statistical Analysis Differences in survival, apoptosis, caspase 12 activation, and inhibition of protein synthesis between groups were analyzed with a twosided unpaired Student t test by use of GraphPad Prism 3.00 (GraphPad Software, San Diego, CA). Results were considered to be statistically significant if P < .05. Experiments were usually replicated three times unless otherwise indicated. Protein synthesis data represent values derived from triplicate samples and are shown as means with 95% confidence intervals (CIs).

Results CAN-Mediated Inhibition of Protein Synthesis To determine whether CAN inhibits overall protein translation in cancer cells, we first measured its effects on de novo protein synthesis in U251 glioma cells. U251 cells were treated with 10 or 50 nM (Figure 1, A and B) CAN for times ranging from 4 to 18 hours and were metabolically labeled with [35S]-Met/Cys during the last hour of incubation. Incorporation of [35S]-Met/Cys label into TCAprecipitable protein was measured to reflect the relative level of protein translation. Cyclohexamide (40 µg/ml), a known inhibitor of protein synthesis, was used as a positive control and inhibited protein synthesis by 88.2% (95% CI = 86.1% to 90.4%) compared with that in untreated control cells at 2 hours. Protein synthesis was decreased by 35.2% (95% CI = 33.3% to 37.3%) and 67.5% (95% CI = 62.3% to 72.6%) at 6 hours and 18 hours, respectively, in U251 cells treated with 10 nM CAN. By contrast, protein synthesis was inhibited by 58.7% (95% CI = 50.4% to 67.1%) and 89.5% (95% CI = 85.7% to 93.3%) at 4 hours and 18 hours, respectively, in U251 cells treated with 50 nM CAN compared with untreated controls. Thus, CAN inhibited protein synthesis in a dose- and time-dependent fashion in U251 cells. We next assessed whether inhibition of protein synthesis by CAN was reversible. U251 cells were treated with 10 or 50 nM CAN for 6 hours, washed with PBS, and incubated in fresh culture media without CAN for an additional 16 hours before lysis. One Vol. 100, Issue 17

|

September 3, 2008

Figure 1. Inhibition of protein synthesis by candidaspongiolide (CAN). A) Time course of new protein accumulation in U251 cells treated with candidaspongiolide (CAN). U251 cells were treated with 10 nM (open bars) or 50 nM (gray bars) CAN for the indicated times. One hour before harvesting cell lysates, [35S]-Met/Cys was added to the media, and the relative amount of [35S]-Met/ Cys incorporation was measured by trichloroacetic acid precipitation of radiolabeled total protein to measure inhibition of protein synthesis. A 2-hour treatment with 40 µg/mL cyclohexamide was used as positive control. Results are expressed as percent of [35S]Met/Cys incorporation relative to that of untreated control cells (equal to 100%). Data are the means of four replicates from a single experiment with upper 95% confidence interval (CI) (P < .001) for all CAN-treated samples relative to control. B) Reversibility of CAN-mediated inhibition of protein synthesis. U251 cells were treated with the indicated concentrations of CAN for 6 hours (gray bars), at which point cells were washed and incubated for additional 16 hours in fresh media without CAN (open bars). Results are expressed as percent of [35S]-Met/Cys incorporation relative to that of untreated control cells (equal to 100%). Data are the means of three replicates from a single experiment with upper 95% CI. C) Reversible inhibition of hypoxiainduced expression of hypoxia-inducible factor-1␣ (HIF-1␣). U251 cells

hour before harvesting cell lysates, [35S]-Met/Cys was added to the media and then TCA-precipitable counts were measured from lysates. As shown in Figure 1, B, CAN-induced inhibition of protein translation was almost completely reversed by a 16-hour wash-out when the lower (10 nM), but not the higher (50 nM), concentration was used. Consistent with this result, we also found that CAN-mediated inhibition of hypoxia-induced HIF-1␣ protein expression was reversible in a dose-dependent fashion. As shown in Figure 1, C, a 6-hour treatment of hypoxic U251 cells with CAN inhibited HIF-1␣ expression, particularly at doses of CAN more than 10 nM. The inhibition was reversible by a 16-hour wash-out in cells treated with up to 25 nM CAN but not in cells treated with 50 nM CAN, indicating a clear difference between the impact of low vs high doses of CAN in terms of inhibition of protein translation. CAN-Mediated Induction of Apoptosis in Human Cancer Cells Next, we investigated whether CAN inhibition of protein translation might be associated with induction of programmed cell death in cancer cells. U251 glioma cells were treated with CAN for 6 hours at concentrations of up to 100 nM, and cell survival and apoptosis were assessed by clonogenic assay (Figure 2, A) and annexin V binding (Figure 2, B), respectively. CAN induced cell death in a dose-dependent fashion, with an IC50 of 20 nM (Figure 2, A). In addition, CAN induced apoptotic cell death in a dose- and time-dependent manner as measured by the percentage of annexin V–positive cells detected by flow cytometry following 6- or 24-hour incubations with increasing concentrations of CAN (Figure 2, B). We observed little if any increase in the percentage of apoptotic cells, relative to that in untreated control cells, in U251 cells treated for 6 hours with CAN at as high as 100 nM. However, we observed a substantial and dose-dependent increase in annexin V–positive jnci.oxfordjournals.org

were cultured for 6 hours in the absence of CAN under normoxic conditions or in the presence of the increasing concentrations of CAN under hypoxic conditions. Alternatively, cells were treated with CAN for 6 hours and then washed and incubated for additional 16 hours in the absence of CAN. Western blots were performed on total cell lysates using an antiHIF-1␣–specific antibody or an antibody to ␤-actin (as a loading control).

U251 cells after 24 hours of CAN treatment. Up to 30% of the cell population became apoptotic at the highest concentration used (100 nM). Similar results were obtained in HCT116 colon carcinoma cells (Supplementary Figure 3, A and B, available online). Finally, U251 cells were treated in the absence or presence of increasing concentrations of CAN for 24 hours, at which point the cell cycle distribution was analyzed (Supplementary Figure 4, available online). A dose-dependent increase of the sub-G1 peak, consistent with apoptosis, was observed in U251 cells treated with increasing concentrations of CAN. In addition, U251 cells accumulated in S phase with a concomitant decrease of cells in G0/G1 (Supplementary Figure 4, available online). We conclude that CAN induced dose-dependent apoptotic cell death in U251 and HCT116 cancer cells. Mechanism of CAN-Induced Apoptosis: Caspase 12 To further characterize the signaling pathway(s) that are involved in CAN-induced apoptosis, U251 and HCT116 cells were treated for 24 hours in the absence or presence of increasing concentrations of CAN, and cleavage of the caspase substrate PARP was analyzed. CAN treatment resulted in the processing of full-length PARP (110 kD) to its 84 kD cleaved form in both HCT116 and U251 cells (Figure 3, A), suggesting that CAN activates a caspase-dependent apoptotic program. Next, we used specific caspase inhibitors to identify the caspases involved in the induction of apoptosis by CAN. HCT116 cells were treated with 50 nM CAN for 48 hours in the absence or presence of a pan-caspase inhibitor or inhibitors of caspases 3, 8, 9, or 12 before annexin V staining and flow cytometry. Figure 3, B shows that CAN induced apoptosis, as reflected by 41% annexin V–positive cells in HCT116 cells treated with CAN for 48 hours. By contrast, only 10% of untreated cells were annexin V positive. The broad-spectrum caspase inhibitor Z-VAD-fmk almost completely inhibited apoptosis induced by CAN, resulting in 15% JNCI

|

Articles 1237

Figure 2. Dose-dependent inhibition of clonogenic survival and induction of apoptosis by candidaspongiolide (CAN) in the U251 cell line. A) Survival assay of CAN-treated cells. U251 cells were treated for 6 hours with increasing concentrations of CAN. Cells were then harvested, plated at a density of 500 cells per dish in 60-mm dishes with medium containing 10% FCS and incubated at 37°C, 5% CO2, for 2 weeks. Colonies were then fixed and stained with crystal violet and counted. Surviving fractions were calculated as the ratio of the absolute number of colonies arising from surviving-treated cells over the number of colonies arising from surviving-untreated control cells. Representative plates (top panel) and survival curves (bottom panel) are shown. Data are expressed as the means of three replicate experiments with 95% confidence intervals (CIs) (error bars). P < .001 for samples treated with 10, 25, 50, and 100 nM of CAN relative to control. B) Apoptosis assay of CAN-treated cells. U251 cells were treated for 6 hours (open bars) or 24 hours (solid bars) with increasing concentrations of CAN, and apoptosis was assessed by flow cytometry after annexin V and propidium iodide staining. Values represent the percentages of annexin V–positive cells at each CAN concentration and are the means of four replicate experiments with upper 95% CI. P = .02 for samples treated with 10 nM of CAN and P < .01 for samples treated with 10, 25, 50, and 100 nM of CAN relative to control.

annexin V–positive cells (Figure 3, B). The addition of inhibitors of caspase 8 (Ac-ZIETD-fmk) or caspase 9 (Ac-ZLEHD-fmk) had little if any effect on the induction of apoptosis by CAN. By contrast, the caspase 12 inhibitor Ac-ZATAD-fmk and to a lesser extent the caspase 3 inhibitor Ac-DEVD-fmk substantially inhibited CANinduced apoptosis, resulting in 19% and 26% annexin V–positive cells, respectively. These results indicate that CAN induced apoptosis in HCT116 and U251 cells by a caspase 12–dependent pathway and, at least in part, by a caspase 3–dependent pathway. Caspase 12 has been implicated in ER stress–dependent apoptosis, although its role in the induction of programmed cell death in human cells is still controversial (26). To further probe the potential involvement of caspase 12 in the induction of apoptosis by CAN, U251 and HCT116 cells were treated for 24 hours with 0–50 nM CAN before electrophoresis of cell lysates and immunoblotting. Blots were probed with specific antibodies that recognize procaspase 12, which is cleaved when it is activated, or that recognize caspase 3 fragments. As shown in Figure 3, C, CAN induced a substantial reduction in levels of the uncleaved procaspase 12 protein, consistent with its dose-dependent cleavage and activation. Concomitant with the decrease in procaspase 12 levels, the levels of 17–19 kD caspase 3 fragments increased, consistent with the activation of this effector caspase. In addition, functional activation of caspase 12 was assessed in U251 cells by measuring the conversion of a cell-permeable substrate, ATAD-AFC, which indicates activation of caspase 12. U251 cells, with or without CAN, were incubated in the presence of ATAD-AFC for 1 hour at 37°C, at which point fluorescence, indicating cleavage of the substrate, was analyzed by flow cytometry. A substantial increase in fluorescence was observed in cells treated with CAN for 24 hours (Figure 3, D), similar to that in cells treated with thapsigargin and tunicamycin, which are known caspase 12 activators. In contrast, treatment of U251 cells with 1238 Articles

|

JNCI

camptothecin, a topoisomerase I inhibitor, did not result in caspase 12 activation (Figure 3, D). These results demonstrate that CAN induces apoptosis in at least two human cancer cell lines by a caspase-dependent pathway that apparently involves caspase 12 and caspase 3. Mechanism of CAN-Induced Activities: Activation of p38 and JNK? Inhibition of protein synthesis by other members of the tedanolide macrolide family, including 13-DT, has been associated with the activation of a ribotoxic stress response, a stress response that is induced by toxic substances and stimulates intracellular signaling pathways (27). 13-DT, which is at least 10-fold more potent than CAN in cytotoxicity assays (7), induces rapid phosphorylation of the stressactivated protein kinase/JNK and p38 MAPKs, two major down stream mediators of the ribotoxic stress response in NIH3T3 cells (7). To investigate whether CAN also induces phosphorylation of the p38 and JNK stress-activated kinases, U251 cells were incubated with 10 nM 13-DT or 100 nM CAN, and p38 and JNK phosphorylation were measured by western blot analysis (Figure 4, A). 13-DT induced rapid and transient phosphorylation of p38 and JNK, which occurred as early as 15 minutes after treatment and then substantially decreased by 60 minutes. In contrast, CAN induced a modest and transient phosphorylation of p38 and JNK that was observed after 30 minutes but was no longer detectable after 60 minutes. This pattern suggests that activation of p38 and JNK kinases may be involved in the early sensing of cellular stress but is unlikely to account for the sustained effects of CAN (eg, induction of apoptosis). To test whether activation of p38 and JNK is required for the induction of apoptosis by CAN, U251 cells were treated with 100 nM CAN or 10 nM 13-DT for 24 hours in the absence or presence of pharmacological inhibitors of the activation of p38 Vol. 100, Issue 17

|

September 3, 2008

Figure 3. Induction of caspase 12–dependent apoptosis by candidaspongiolide (CAN). A) Western blot analysis of poly (ADP-ribose) polymerase (PARP) cleavage in U251 and HCT116 cells. Cells were treated for 24 hours with increasing concentrations of CAN. Lysates were separated by electrophoresis, and blots were probed with anti-PARP antibodies. B) Flow cytometric analysis of annexin V–positive HCT116 cells after treatment with caspase inhibitors. Cells were treated for 48 hours with 50 nM CAN in the absence or presence of the following caspase inhibitors at 10-µM concentration: Ac-DEVD-fmk (caspase 3), Ac-ZIETDfmk (caspase 8), Ac-ZLEHD-fmk (caspase 9), Ac-ZATAD-fmk (caspase 12), and Z-VAD-fmk (pan-caspase inhibitor). Percentages of annexin V–positive/propidium iodide–negative cells (bottom) and propidium iodide–positive cells (top) from one representative experiment are shown. C) Western blot analysis of (inactive) pro-caspase 12 and (activated) caspase 3 levels in U251 and HCT116 cells treated for 24 hours in the absence or presence of increasing concentration of CAN. ␤-Actin is shown as loading control. D) Assay of caspase 12 activation. U251 cells were treated for 24 hours in the absence (-) or presence of 50 nM CAN, 2 µM thapsigargin (Th), 20 µM tunicamycin (Tu), or 1 µM camptothecin (CPT). Cells were then incubated with FITC-ATAD-fmk for 1 hour and immediately analyzed for caspase 12 activation by flow cytometry. Results are expressed as mean intensity of fluorescence and are the means of three replicates from a single experiment with upper 95% confidence interval.

(SB202190) and JNK (SP600125). Under these conditions, both SB202190 and SP600125 selectively inhibited activation of the corresponding target kinase by hydrogen peroxide, a known activator of stress kinases (data not shown). As illustrated in Figure 4, B, treatment of U251 cells with SB202190 or SP600125 alone did not affect the percentage of cells that appeared to be apoptotic relative to the untreated control (less than 7% in either case). Interestingly, induction of apoptosis by CAN (approximately 31% annexin V–positive cells) or 13-DT (approximately 28%

annexin V–positive cells) was not affected by pretreatment of the cells with either inhibitor, suggesting that activation of p38 and JNK is not required for induction of cell death by CAN and 13-DT. CAN-Induced Phosphorylation of eIF2␣ To better understand the mechanism by which CAN inhibited protein translation, U251 cells were treated with increasing concentrations of CAN and the levels of expression and/or phosphorylation of

Figure 4. Induction of p38 and jun NH2terminal kinase (JNK) phosphorylation by candidaspongiolide (CAN). A) Time course of p38 and JNK phosphorylation. U251 cells were cultured in the absence or presence of 100 nM CAN or 10 nM 13-deoxytedanolide (13-DT) for 0–60 minutes. Levels of phospho-p38 (Thr180/Tyr182), phospho-JNK (Thr183/Tyr185), total p38, and total JNK proteins were then assessed by western blot analysis. ␤-Actin levels are shown as a loading control. B) Apoptosis assays with p38 and JNK inhibitors. U251 cells were treated with 100 nM CAN or 10 nM 13-DT for 24 hours or left untreated and were cultured in the absence or presence of SB202190, a pharmacological inhibitor of p38 activation, or SP600125, an inhibitor of JNK activation. Apoptosis was assessed by flow cytometry after annexin V and propidium iodide staining. Percentages of annexin V–positive cells are indicated.

jnci.oxfordjournals.org

JNCI

|

Articles 1239

Figure 5. Candidaspongiolide (CAN) induction of the phosphorylation of eukaryotic initiation factor-2 (eIF2)-␣ and eukaryotic elongation factor-2 (eEF2) in U251 and HCT116 cells. A) Phosphorylation of translation regulatory proteins. U251 cells were treated with the indicated concentration of CAN for 16 hours. Levels of phospho4EBP1 (Thr37/46), total 4EBP1, total eIF4E, phospho-eIF2␣ (Ser51), total eIF2␣, phospho-eEF2 (Thr56), and total eEF2 were analyzed by western blot analysis. Levels of ␤-actin are shown as loading control. B) Time course of eIF2␣ inactivation. U251 cells were treated with 10 nM (upper panel) or 50 nM (lower panel) of CAN for the indicated times (ranging from 2 to 48 hours), and levels of total and phosphorylated eIF2␣ were analyzed by western blot analysis. C) eIF2␣ phosphorylation in HCT116 cells. HCT116 cells were treated with increasing concentrations of CAN for 16 hours, and levels of total and phosphorylated eIF2␣ were analyzed by western blot analysis.

downstream targets of mTOR activity, which regulate cap-dependent protein translation, were analyzed. As shown in Figure 5, A, one such target, 4EBP1, was phosphorylated in untreated U251 cells, presumably because of a loss of function of the PTEN tumor suppressor gene present in these cells (28). No changes in either phosphorylation or expression of 4EBP1 were observed in U251 cells treated with 10 or 50 nM CAN. In addition, no changes in the total levels of eIF4E, another downstream target of mTOR, were observed. These results suggest that the effects of CAN on protein synthesis did not involve inhibition of the 4EBP1–eIF4E pathway. We then tested the levels of expression and phosphorylation of other factors known to be involved in translation initiation and elongation. eIF2 is an important regulatory protein for mRNA translation, and phosphorylation of the ␣ subunit of eIF2 is a welldocumented mechanism for inhibition of protein synthesis (12). CAN induced eIF2␣ phosphorylation at Ser51 in a dosedependent manner (Figure 5, A). In addition, CAN induced a marked increase in the phosphorylation of the protein elongation factor eEF2 at Thr56 (Figure 5, A), a phosphorylation site associated with inhibition of its activity (15). Phosphorylation of eEF2 at Thr56 was both time and dose dependent (Supplementary Figure 5, available online). We further investigated the kinetics of eIF2␣ phosphorylation induced by CAN. U251 cells were exposed to 10 or 50 nM CAN for 2–48 hours, and the phosphorylation of eIF2␣ was monitored by western blotting. At 10 nM, CAN induced eIF2␣ phosphorylation that was detectable at 6 hours, more pronounced at 16 hours, and persisted until at least 48 hours (Figure 5, B). At 50 nM, CAN induced much greater phosphorylation of eIF2␣, which was clearly detectable by 4 to 6 hours and was maintained up to at least 48 hours (Figure 5, B). CAN at 10 and 50 nM also induced substantial phosphorylation of eIF2␣ in HCT116 cells (Figure 5, C). Taken together, our results demonstrate that CAN induced sustained phosphorylation of eIF2␣ in U251 and HCT116 cells. They also suggest that phosphorylation and inactivation of eIF2␣ and eEF2 may be responsible for the inhibition of protein synthesis caused by CAN. 1240 Articles

|

JNCI

Mechanism of CAN-Induced eIF2␣ Phosphorylation: Classic ER Stress? Phosphorylation of eIF2␣ is a key event that mediates inhibition of translation during the ER stress and the unfolded protein response (UPR) (20,21,24,29). In addition, our finding that CAN induced caspase 12–dependent apoptosis raised the possibility that CAN might induce a classic ER stress response. We therefore examined the effects of CAN on the expression of proteins that are modulated during ER stress, including the bZIP transcriptional activators ATF4 and CHOP (20). Thapsigargin, an ER Ca2+-ATPase inhibitor that is known to induce ER stress and eIF2␣ phosphorylation, was used as positive control. Consistent with earlier reports (24), thapsigargin induced eIF2␣ phosphorylation and ATF4 and CHOP expression in a dose-dependent fashion in U251 cells (Figure 6, A). By contrast, CAN induced only eIF2␣ phosphorylation and did not affect the levels of ATF4 and CHOP, suggesting that distinct pathways may mediate the ER response and eIF2␣ phosphorylation induced by CAN. PERK is part of a signaling pathway that phosphorylates and inhibits eIF2␣ in response to stimuli that induce ER stress, such as glucose deprivation, severe hypoxia, and thapsigargin treatment. In the presence of ER stress conditions, misfolded proteins accumulate in the ER and activate distinct signalling pathways that mediate survival and/or cell death (20,21,29). To investigate whether PERK is required for the phosphorylation of eIF2␣ induced by CAN, U251 cells were transfected with a siRNA targeting PERK (siPERK) or a scrambled negative control RNA (siNC), and cells were tested for eIF2␣ phosphorylation in response to CAN or thapsigargin 48 hours after transfection. As demonstrated in Figure 6, B, transfection of U251 cells with siRNA directed against PERK almost completely abrogated PERK expression relative to cells transfected with control scrambled siRNA. CAN was equally able to induce eIF2␣ phosphorylation in U251 cells transfected with either siNC or siPERK, demonstrating that CAN induction of eIF2␣ phosphorylation is PERK independent. By contrast, thapsigargin, which is known to Vol. 100, Issue 17

|

September 3, 2008

Figure 6. Effect of candidaspongiolide (CAN) on classic endoplasmic reticulum stress response pathways. A) Effects of CAN on unfolded protein response pathway components. U251 cells were treated with the indicated concentrations of CAN or thapsigargin (Thaps) for 16 and 4 hours, respectively. Cells were then lysed, and levels of total and phosphorylated (pSer51) eukaryotic initiation factor-2 (eIF2)-␣, total ATF4, and GADD153/ CHOP were assayed by western blot analysis. ␤-Actin is shown as a loading control. B) Effect of reduced levels of PERK on eIF2␣ phosphorylation. U251 cells were transiently transfected with negative control smallinterfering RNA (siNC) or siRNA targeting PERK (siPERK). After 48 hours, cells were treated with the indicated concentrations of CAN or thapsigargin for 16 and 4 hours, respectively, and levels of PERK, phosphoeIF2␣ (Ser51), and total eIF2␣ were assayed by western blotting. ␤-actin levels are shown as loading control.

induce the UPR pathway, induced eIF2␣ phosphorylation in the presence but not the absence of PERK. These results suggest that CAN induced eIF2␣ phosphorylation by a biochemical pathway that is distinct from the UPR. Requirement of eIF2␣ Phosphorylation for CAN-Mediated Induction of Apoptosis To further investigate the role of eIF2␣ in the inhibition of protein synthesis and the induction of apoptosis by CAN, we generated U251 cells expressing a mutant allele of eIF2␣ that encodes a protein with a single amino acid substitution at position 51 (S51A). This mutated form of eIF2␣ acts as a dominant-negative protein to prevent the phosphorylation of wild-type eIF2␣ (24). As shown in Figure 7, A, CAN induced eIF2␣ phosphorylation in U251-neo cells, that is, cells transfected with an empty vector containing the neomycin resistance gene. In contrast, both basal and CANinduced levels of eIF2␣ phosphorylation were dramatically reduced in cells that expressed the mutated form of eIF2␣ (U251-S51A). To test the involvement of eIF2␣ in the inhibition of protein synthesis by CAN, we measured [35S]-Met/Cys incorporation following treatment of U251-neo and U251-S/A cells with CAN. As in previous experiments, CAN caused a dose-dependent reduction of protein synthesis, as determined by [35S]-Met/Cys incorporation into TCA-precipitated protein in U251-neo cells (Figure 7, B). The presence of S51A-mutated eIF2␣ only partially abrogated the inhibition of protein synthesis by 10 or 50 nM CAN in U251-S/A cells, with a 71.2% (95% CI = 63.8% to 78.8%) decrease in [35S]Met/Cys incorporation still detectable in cells treated with CAN at 50 nM, despite the lack of eIF2␣ phosphorylation. These results suggest that eIF2␣ phosphorylation contributed to, but was not sufficient for, the inhibition of protein synthesis observed in cells treated with CAN and are consistent with the possibility that other factors, for example, eEF2 phosphorylation, may be involved. They also confirmed our earlier finding that low and high doses of CAN may have distinct impacts on inhibition of protein synthesis. We next addressed the role of eIF2␣ phosphorylation in the induction of apoptosis by CAN. Apoptosis was measured by the annexin V assay in U251-neo and U251-S51A cells treated with varying amounts of CAN for 24 hours (Figure 7, C). As in previous jnci.oxfordjournals.org

experiments, CAN induced apoptosis in a dose-dependent fashion in U251-neo cells. In contrast, apoptosis was substantially decreased in U251-S51A cells (Figure 7, C). Indeed, CAN induced 21% and 33% of the U251-neo cells to undergo programmed cell death at 10 and 50 nM concentrations, respectively. In contrast, it induced only 8% and 14% of the U251-S51A cells to undergo programmed cell death at 10 and 50 nM concentrations, respectively (Figure 7, C). Camptothecin, a DNA damaging agent that causes programmed cell death by a topoisomerase I–dependent mechanism, induced similar levels of apoptosis in both U251-neo and U251S51A cells (data not shown). Taken together, these results demonstrate that phosphorylation of eIF2␣ was required for the induction of apoptosis by CAN, but not by a classic DNA damaging agent. In addition, they suggest that eIF2␣ phosphorylation was essential for the induction of apoptosis by CAN but was only a contributing factor for the inhibition of protein synthesis. Mechanism of CAN-Induced eIF2␣ Phosphorylation: PKR We next sought to determine which kinase(s) mediate phosphorylation of eIF2␣ induced by CAN. eIF2␣ can be phosphorylated by a number of kinases, including PERK, PKR, HRI, and GCN2, which are activated during distinct stress conditions (14,24,30–33). Because eIF2␣ phosphorylation has been implicated in the induction of apoptosis in response to PKR activation or overexpression (34–37), we determined whether CAN increased the levels of the active, phosphorylated form of PKR, again using a phosphospecific antibody on western blots. As shown in Figure 8, A, CAN induced a dose-dependent increase in PKR phosphorylation without affecting the total levels of PKR. Notably, pretreatment of U251 cells with 2-AP, an inhibitor of PKR, completely abrogated CAN-induced eIF2␣ phosphorylation in U251 cells (Figure 8, B). The involvement of PKR in the induction of eIF2␣ phosphorylation by CAN was further supported by experiments using siRNA targeting PKR. As shown in Figure 8, C, PKR levels were substantially reduced in cells transfected with PKRspecific siRNA (siPKR) relative to cells transfected with a negative control siRNA (siNC). More importantly, the induction of eIF2␣ phosphorylation induced by CAN in U251 cells JNCI

|

Articles 1241

Figure 7. Effect of eukaryotic initiation factor-2 (eIF2)-␣ phosphorylation on induction of apoptosis by candidaspongiolide (CAN). A) Effect of a dominant-negative eIF2␣ construct. U251 cells stably transfected with either a control plasmid (U251-neo) or a plasmid containing a mutant allele of eIF2␣ (U251-S51A) were treated in the absence or presence of the indicated concentrations of CAN for 16 hours. Levels of total and Ser51-phosphorylated eIF2␣ were then measured in total cell lysates by western blot. B) Protein synthesis in U251 cells with dominant-negative eIF2␣. U251-neo (solid bars) and U251-S51A (open bars) cells were treated with increasing concentrations of CAN for 6 hours. One hour before harvesting lysates, [35S]-Met/Cys was added to the media, and the relative amount of [35S]-Met/Cys incorporation was measured by

trichloroacetic acid precipitation of cell lysates. Results are expressed as percent of [35S]-Met/Cys incorporation relative to that of untreated control cells (equal to 100%). Data are means of three experiments with 95% confidence intervals (P = .01 for 10nM CAN and P = .02 for 50 nM CAN) for U251-S51A relative to U251-neo. C) Apoptosis in U251 cells with dominant-negative eIF2␣. U251-neo and U251-S51A cells were treated with 10 or 50 nM of CAN for 24 hours, and the percentage of apoptotic cells was measured by flow cytometric analysis of annexin V– vs propidium iodide (PI)–positive cells. Percentage of annexin V– positive cells (propidium iodide–negative, bottom; propidium iodide– positive, top) from one representative experiment is shown out of three that were performed.

transfected with siNC was not detectable in cells transfected with siPKR, further demonstrating an essential role for PKR in the induction of eIF2␣ phosphorylation by CAN (Figure 8, C). The results shown in Figure 7, C had indicated that eIF2␣ is essential for the induction of apoptosis by CAN. Therefore, we reasoned that the inhibition of PKR activation might not only be required for eIF2␣ phosphorylation by CAN but also affect the ability of CAN to induce apoptosis. To address this question, U251 cells were treated with increasing concentrations of CAN in the absence or presence of 5 mM 2-AP. As shown in Figure 8, D, 5 mM 2-AP caused a slight increase in apoptosis relative to levels in untreated cells (12% vs 7%). However, addition of 2-AP substantially decreased the amount of apoptosis induced by 50 nM CAN, and the

percentage of apoptotic cells induced by treatment with 50 nM CAN dropped from 33% in the absence of 2-AP to 15% in the presence of 2-AP. The finding further supports the involvement of the PKR kinase in the signal transduction pathways induced by CAN (Figure 8, D).

1242 Articles

|

JNCI

Differences in CAN-Mediated Activities in Normal Cells vs Cancer Cells Our original hypothesis that led us to further investigate the mechanism of action of CAN was that cancer cells might be more sensitive to inhibitors of protein synthesis than normal cells. To further test this hypothesis, we investigated the effects of CAN on protein synthesis and apoptosis in normal human fibroblasts. Vol. 100, Issue 17

|

September 3, 2008

Figure 8. RNA-dependent protein kinase (PKR) is required for eukaryotic initiation factor-2 (eIF2)-␣ phosphorylation induced by candidaspongiolide (CAN). A) Western blot analysis of levels of phosphorylated (Thr451) PKR and total PKR expression in U251 cells treated with increasing concentrations of CAN for 16 hours. ␤-Actin levels are shown as loading control. B) Effect of a PKR inhibitor on eIF2␣ phosphorylation. U251 cells were treated with 50 nM CAN in the absence or presence of the PKR inhibitor 2-aminopurine (2-AP, 5 mM). Total and phosphorylated levels of eIF2␣ were then measured by western blot. C) Effect of siRNA to PKR on eIF2␣ phosphorylation. U251 cells were transiently transfected with negative control siRNA (siNC) or siRNA targeting PKR (siPKR). Twenty-four hours following transfection, cells were treated with increasing concentrations of CAN for 16 hours. Levels of total PKR, and total and phosphorylated eIF2␣ (Ser51) were then measured by western blot on total cell lysates. ␤-Actin levels are shown as loading control. D) Effect of a PKR inhibitor on CAN-induced apoptosis. U251 cells were treated with increasing concentration of CAN (open bars) in the absence or presence of 5 mM 2-AP (solid bars) for 24 hours. Percentage of annexin V–positive cells was then measured by flow

cytometry. Total percentage of annexin V–positive cells (propidium iodide–negative and propidium iodide–positive combined) is shown as the mean of three independent experiments with upper 95% confidence interval.

We first tested the effects of CAN on eIF2␣ phosphorylation in two normal human fibroblast cell lines, AG06858 and AG068103. Both normal human fibroblast cell lines showed constitutive phosphorylation of eIF2␣ that was not substantially increased by treatment with up to 100 nM CAN (Figure 9, A). However, in [35S]-Met/Cys incorporation experiments, CAN retained its ability to inhibit protein synthesis in a dose-dependent fashion (Figure 9, B) to a similar extent to what we observed in cancer cell lines. To address whether CAN induced apoptosis in

AG06858 and AG068103 normal human fibroblasts, cells were treated in the absence or presence of 100 nM CAN for 24 hours, and the percentage of apoptotic cells was assessed by annexin V binding. Interestingly, CAN induced little or no increase in the percentage of apoptotic cells relative to untreated controls (9% with CAN vs 6% without CAN in AG06858 and 5% with CAN vs 3% without CAN in AG068103) (Figure 9, C). These results further support the link between eIF2␣ phosphorylation and induction of apoptosis by CAN, and they suggest that, even

Figure 9. Effect of candidaspongiolide (CAN) on protein synthesis, eukaryotic initiation factor-2 (eIF2)-␣ phosphorylation, and apoptosis, in normal human fibroblasts. A) Time course of eIF2␣ phosphorylation in normal human fibroblasts. AG06858 and AG068103 normal human fibroblasts were incubated in the absence or presence of increasing concentrations of CAN for 16 hours. Levels of total and Ser51phosphorylated eIF2␣ were then measured by western blot. B) Protein synthesis in CAN-treated normal cells. AG06858 (solid bars) and AG068103 (open bars) human fibroblasts were incubated in the absence or presence of increasing concentrations of CAN for 6 hours. One hour before harvesting [35S]-Met/Cys was added to the media, and the relative amount of [35S]Met/Cys incorporation was measured by trichloroacetic acid precipitation. Results represent percent of [35S]-Met/Cys incorporation relative to that of untreated control cells (equal to 100%) are expressed as means of three experiments with upper 95% confidence interval (P < .05) for samples treated with 10 and 50 nM of CAN, relative to control. C) Apoptosis in CAN-treated normal cells. AG06858 and AG068103 human fibroblasts were cultured in the absence or presence of 100 nM CAN for 24 hours, and percentage of annexin V–positive cells was measured by flow cytometry analysis. Percentage of annexin V– positive cells (propidium iodide–negative, bottom; propidium iodide–positive, top) from one representative experiment is shown.

jnci.oxfordjournals.org

JNCI

|

Articles 1243

though CAN inhibits protein synthesis in normal human fibroblasts, these cells are much less sensitive than cancer cell lines to eIF2␣ phosphorylation and apoptosis induced by CAN.

Discussion Natural products are an important source of novel, complex chemicals, many of which exhibit potent cytotoxic activities and are currently being used for cancer therapy. In this study we have examined the mechanism of action of CAN, a novel natural product isolated from the marine sponge Candidaspongia sp. We observed two main effects in cells treated with CAN: inhibition of protein synthesis and induction of apoptosis. Also, we have provided evidence that the main signaling pathway required for the induction of cell death by CAN is the PKR-dependent phosphorylation of eIF2␣ and not a pathway downstream of inhibition of protein synthesis. We demonstrated that 1) in cells stably transfected with a dominant-negative allele of eIF2␣, CAN no longer caused eIF2␣ phosphorylation and induction of apoptosis was substantially impaired, yet protein synthesis was effectively inhibited; 2) inhibition of PKR by siRNA or 2-AP prevented phosphorylation of eIF2␣ and induction of apoptosis by CAN; 3) CAN efficiently inhibited protein synthesis in normal human fibroblasts, yet did not induce phosphorylation of eIF2␣ or apoptosis even at high concentrations; and 4) activation of p38 and JNK was transient and not required for the induction of cell death by CAN. These results are consistent with at least two possibilities. The first possibility is that CAN retains some activities of other members of the tedanolide family of macrolides, such as induction of stress-activated kinases JNK/p38 and inhibition of protein synthesis, yet has a unique propensity to induce phosphorylation of eIF2␣ and apoptosis. The second possibility is that inhibition of protein synthesis, although possibly contributing to the overall cytotoxic activity of CAN, is necessary but not sufficient to trigger an apoptotic cascade. Recently, the discovery of inhibitors of protein translation has attracted considerable interest for their potential use in cancer therapy, and high-throughput screens have been successfully implemented to identify small-molecule modulators of translation initiation (38–41). Selective inhibition of protein translation by most of these agents has been attributed to their binding to the structured conformation of the 5′ untranslated region (5′-UTR) of mRNAs. Indeed, it has been postulated that mRNAs that have a short unstructured 5′UTR, for example, ␤-actin, would be less sensitive to inhibitors of translation initiation than those mRNAs with long, highly structured 5′-UTRs, for example, those of oncogenes (41). However, it is also conceivable that global protein synthesis inhibition may preferentially affect short-lived proteins, many of which are the products of mRNAs with long, highly structured 5′-UTRs. Indeed, our results indicate that CAN predominantly inhibited proteins with a short half-life, for example, HIF-1␣, whereas it had little effect on proteins with a long half-life, including HIF-1␤, ␤-actin, and all the kinases tested in this study, even after prolonged treatment at high concentration. Whether this result reflects an intrinsic property of CAN to preferentially target short-lived proteins or is a consequence of the extent and duration of global inhibition of translation remains to be determined. Likewise, the extent to which modulation of translation may impact the therapeutic potential of CAN is not yet known. 1244 Articles

|

JNCI

The finding that CAN inhibited protein synthesis in normal human fibroblasts yet did not induce eIF2␣ phosphorylation and apoptosis is consistent with our conclusion that eIF2␣ phosphorylation induced by CAN is required for the induction of apoptosis, and it suggests that metabolically active cancer cells, in which key translational factors are frequently dysregulated, may be more sensitive than normal cells to inhibition of protein synthesis. Indeed, cancer cells readily adapt to stress conditions, including hypoxia and nutrient starvation, and are less responsive than normal cells to physiological control mechanisms that regulate protein synthesis (42). Interestingly, normal human fibroblasts showed constitutive phosphorylation of eIF2␣ at Ser51 that was not affected by CAN. Similar results have been reported in MCF10A human breast epithelial cells, which, unlike breast cancer cell lines, express constitutive, but not inducible, eIF2␣ phosphorylation at Ser51, consistent with distinct regulation of eIF2␣ in normal and cancer cell lines (42). Inhibitory phosphorylation of eIF2␣ at serine 51 is observed under diverse stress conditions, including ER stress and the UPR (20,21). Although CAN induced consistent and sustained eIF2␣ phosphorylation at serine 51, it did not activate ATF4, CHOP, or PERK, all of which are involved in the classic UPR. Several kinases have been implicated in the phosphorylation of eIF2␣ under distinct stress conditions. Our results clearly implicate PKR as the upstream kinase that is involved in CAN-dependent eIF2␣ phosphorylation and apoptosis. Indeed, inhibition of PKR, by either specific siRNA or 2-AP, completely abrogated eIF2␣ phosphorylation and apoptosis induced by CAN. PKR is a well-characterized kinase that is activated by viral double-stranded RNA (dsRNA) to phosphorylate eIF2␣ and inhibit protein synthesis. Activation of a PKR-dependent pathway has been associated with induction of apoptosis in response to different cellular stress conditions, including the presence of dsRNA, serum deprivation, and TNF␣ treatment (22,35,37,43,44). CAN appeared to induce a unique cell death cascade involving activation of a caspase 12–dependent apoptotic pathway. Caspase 12 is an initiator caspase that regulates ER stress–mediated apoptosis (26). Thapsigargin, a specific inhibitor of the ER-associated Ca++ ATPase, induces processing of caspase 12 and apoptotic cell death (26). The role of caspase 12 in ER stress–induced apoptosis is well established in rodents (26) yet still debated in humans, although evidence of a human caspase 12–like protein has been provided (45–47). Our findings are consistent with the involvement of a human caspase 12–like protein in the induction of apoptosis by CAN. Indeed, apoptotic cell death induced by CAN was almost completely abrogated by Z-ATAD, an inhibitor of caspase 12, or by Z-VAD, a global caspase inhibitor. By contrast, inhibitors of caspase 8 or caspase 9 had no effect on the induction of apoptosis by CAN, suggesting activation of a pathway distinct from the intrinsic and extrinsic apoptotic pathways. Involvement of caspase 12 was further suggested by dose-dependent cleavage of procaspase 12 by CAN and by induction of caspase 12 activity, as assessed by a fluorescent enzymatic assay, to a similar extent as by thapsigargin or tunicamycin, inducers of ER stress. Overall, our results are consistent with the activation of an ER stress–like pathway by CAN, with induction of PKR-dependent phosphorylation of eIF2␣ and activation of a caspase 12–dependent apoptotic pathway. Further studies will be required to fully elucidate the triggering signals leading to activation of caspase 12 and its association with eIF2␣ phosphorylation. Vol. 100, Issue 17

|

September 3, 2008

CAN is a novel compound with potential antitumor activity. Members of the tedanolide family of compounds exert potent cytotoxic activity, which is associated with inhibition of protein synthesis. Agents affecting global metabolic pathways, such as protein translation, face intense scrutiny for clinical development. However, recent successful examples of anticancer agents that affect protein degradation (proteasome inhibitors), chaperone function (Hsp90), mRNA synthesis (flavopiridol), or protein acetylation (HDAC) have lessened the skepticism surrounding these types of compounds. Potential for further development of CAN as an anticancer agent will ultimately be determined by availability of this compound, either from the primary source or through synthetic chemistry, and the magnitude of its “therapeutic window.” Thus far, our work is limited by the fact we have demonstrated the propensity of CAN to kill cancer cells more readily than normal cells only through in vitro experiments in cell lines. The evidence provided by this study, which also identifies a biomarker, eIF2␣ phosphorylation, that is associated with induction of cell death by CAN, may form the basis for further preclinical studies to test the efficacy and safety of this drug on mouse tumor models and may facilitate the assessment of its potential for therapeutic applications in humans. References 1. Bergquist PR, Sorokin S, Karuso P. Pushing the boundaries: a new genus and species of Dictyoceratida. Mem Queensl Mus. 1999;44(1):57–62. 2. Meragelman TL, Willis RH, Woldemichael GM, et al. Candidaspongiolides, distinctive analogues of tedanolide from sponges of the genus Candidaspongia. J Nat Prod. 2007;70(7):1133–1138. 3. Schmitz FJ, Gunasekera SP, Yalamanchili G, Hussain MB, van der Helm D. Tedanolide, a potent cytotoxic macrolide from the Caribbean sponge Tedania ignis. J Am Chem Soc. 1984;106(23):7251–7252. 4. Fusetani N, Sagawara T, Matsunaga S. Cytotoxic metabolites of the marine sponge Mycale adhaerens Lambe. J Org Chem. 1991;56(16): 4971–4974. 5. Chevallier C, Bugni TS, Feng X, Harper MK, Orendt AM, Ireland CM. Tedanolide C: a potent new 18-membered-ring cytotoxic macrolide isolated from the Papua New Guinea marine sponge Ircinia sp. J Org Chem. 2006;71(6):2510–2513. 6. Lee KH, Nishimura S, Matsunaga S, et al. Induction of a ribotoxic stress response that stimulates stress-activated protein kinases by 13deoxytedanolide, an antitumor marine macrolide. Biosci Biotechnol Biochem. 2006;70(1):161–171. 7. Nishimura S, Matsunaga S, Yoshida M, Hirota H, Yokoyama S, Fusetani N. 13-Deoxytedanolide, a marine sponge-derived antitumor macrolide, binds to the 60S large ribosomal subunit. Bioorg Med Chem. 2005;13(2): 449–454. 8. Schroeder SJ, Blaha G, Tirado-Rives J, Steitz TA, Moore PB. The structures of antibiotics bound to the E site region of the 50 S ribosomal subunit of Haloarcula marismortui: 13-deoxytedanolide and girodazole. J Mol Biol. 2007;367(5):1471–1479. 9. Smith AB, Lee D. Total synthesis of (+)-tedanolide. J Am Chem Soc. 2007;129(35):10957–10962. 10. Ruggero D, Pandolfi PP. Does the ribosome translate cancer? Nat Rev Cancer. 2003;3(3):179–192. 11. Meric F, Hunt KK. Translation initiation in cancer: a novel target for therapy. Mol Cancer Ther. 2002;1(11):971–979. 12. Kleijn M, Scheper GC, Voorma HO, Thomas AA. Regulation of translation initiation factors by signal transduction. Eur J Biochem. 1998;253(3): 531–544. 13. Proud CG. eIF2 and the control of cell physiology. Semin Cell Dev Biol. 2005;16(1):3–12. 14. Kimball SR. Eukaryotic initiation factor eIF2. Int J Biochem Cell Biol. 1999;31(1):25–29. jnci.oxfordjournals.org

15. Pause A, Belsham GJ, Gingras AC, et al. Insulin-dependent stimulation of protein synthesis by phosphorylation of a regulator of 5′-cap function. Nature. 1994;371(6500):762–767. 16. Haghighat A, Mader S, Pause A, Sonenberg N. Repression of capdependent translation by 4E-binding protein 1: competition with p220 for binding to eukaryotic initiation factor-4E. EMBO J. 1995;14(22): 5701–5709. 17. Richter JD, Sonenberg N. Regulation of cap-dependent translation by eIF4E inhibitory proteins. Nature. 2005;433(7025):477–480. 18. Bjornsti MA, Houghton PJ. The TOR pathway: a target for cancer therapy. Nat Rev Cancer. 2004;4(5):335–348. 19. Elmore S. Apoptosis: a review of programmed cell death. Toxicol Pathol. 2007;35(4):495–516. 20. Koumenis C, Bi M, Ye J, Feldman D, Koong AC. Hypoxia and the unfolded protein response. Methods Enzymol. 2007;435:275–293. 21. Feldman DE, Chauhan V, Koong AC. The unfolded protein response: a novel component of the hypoxic stress response in tumors. Mol Cancer Res. 2005;3(11):597–605. 22. Jagus R, Joshi B, Barber GN. PKR, apoptosis and cancer. Int J Biochem Cell Biol. 1999;31(1):123–138. 23. Garcia A, Gil J, Ventoso I, et al. Impact of protein kinase PKR in cancer biology: from antiviral to antiproliferative action. Microbiol Mol Biol Rev. 2006;70(4):1032–1060. 24. Koumenis C, Naczki C, Koritzinsky M, et al. Regulation of protein synthesis by hypoxia via activation of the endoplasmic reticulum kinase PERK and phosphorylation of the translation initiation factor eIF2alpha. Mol Cell Biol. 2002;22(21):7405–7416. 25. Del Bufalo D, Trisciuoglio D, Scarsella M, Zangemeister-Wittke U, Zupi G. Treatment of melanoma cells with a bcl-2/bcl-xL antisense oligonucleotide induces antiangiogenic activity. Oncogene. 2003;22(52): 8441–8447. 26 Szegezdi E, Fitzgerald U, Samali A. Caspase-12 and ER-stress-mediated apoptosis: the story so far. Ann N Y Acad Sci. 2003;1010:186–194. 27. Laskin JD, Heck DE, Laskin DL. The ribotoxic stress response as a potential mechanism for MAP kinase activation in xenobiotic toxicity. Toxicol Sci. 2002;69(2):289–291. 28. See http://www.sanger.ac.uk/genetics/CGP/NCI60/. 29. van den BT, Koritzinsky M, Wouters BG. Translational control of gene expression during hypoxia. Cancer Biol Ther. 2006;5(7):749–755. 30. Boyce M, Bryant KF, Jousse C, et al. A selective inhibitor of eIF2alpha dephosphorylation protects cells from ER stress. Science. 2005;307(5711): 935–939. 31. Bi M, Naczki C, Koritzinsky M, et al. ER stress-regulated translation increases tolerance to extreme hypoxia and promotes tumor growth. EMBO J. 2005;24(19):3470–3481. 32. Lu PD, Jousse C, Marciniak SJ, et al. Cytoprotection by pre-emptive conditional phosphorylation of translation initiation factor 2. EMBO J. 2004; 23(1):169–179. 33. Zhang P, McGrath BC, Reinert J, et al. The GCN2 eIF2 alpha kinase is required for adaptation to amino acid deprivation in mice. Mol Cell Biol. 2002;22(19):6681–6688. 34. Donze O, Deng J, Curran J, Sladek R, Picard D, Sonenberg N. The protein kinase PKR: a molecular clock that sequentially activates survival and death programs. EMBO J. 2004;23(3):564–571. 35. Gil J, Alcami J, Esteban M. Induction of apoptosis by double-strandedRNA-dependent protein kinase (PKR) involves the alpha subunit of eukaryotic translation initiation factor 2 and NF-kappaB. Mol Cell Biol. 1999;19(7):4653–4663. 36. Hsu LC, Park JM, Zhang K, et al. The protein kinase PKR is required for macrophage apoptosis after activation of Toll-like receptor 4. Nature. 2004;428(6980):341–345. 37. Srivastava SP, Kumar KU, Kaufman RJ. Phosphorylation of eukaryotic translation initiation factor 2 mediates apoptosis in response to activation of the double-stranded RNA-dependent protein kinase. J Biol Chem. 1998;273(4):2416–2423. 38. Low WK, Dang Y, Schneider-Poetsch T, et al. Inhibition of eukaryotic translation initiation by the marine natural product pateamine A. Mol Cell. 2005;20(5):709–722. JNCI

|

Articles 1245

39. Bordeleau ME, Matthews J, Wojnar JM, et al. Stimulation of mammalian translation initiation factor eIF4A activity by a small molecule inhibitor of eukaryotic translation. Proc Natl Acad Sci USA. 2005;102(30): 10460–10465. 40. Bordeleau ME, Mori A, Oberer M, et al. Functional characterization of IRESes by an inhibitor of the RNA helicase eIF4A. Nat Chem Biol. 2006;2(4):213–220. 41. Moerke NJ, Aktas H, Chen H, et al. Small-molecule inhibition of the interaction between the translation initiation factors eIF4E and eIF4G. Cell. 2007;128(2):257–267. 42. Connolly E, Braunstein S, Formenti S, Schneider RJ. Hypoxia inhibits protein synthesis through a 4E-BP1 and elongation factor 2 kinase pathway controlled by mTOR and uncoupled in breast cancer cells. Mol Cell Biol. 2006;26(10):3955–3965. 43. Der SD, Yang YL, Weissmann C, Williams BR. A double-stranded RNAactivated protein kinase-dependent pathway mediating stress-induced apoptosis. Proc Natl Acad Sci USA. 1997;94(7):3279–3283. 44. Balachandran S, Kim CN, Yeh WC, Mak TW, Bhalla K, Barber GN. Activation of the dsRNA-dependent protein kinase, PKR, induces apoptosis through FADD-mediated death signaling. EMBO J. 1998;17(23):6888–6902. 45. Nakagawa T, Zhu H, Morishima N, et al. Caspase-12 mediates endoplasmic-reticulum-specific apoptosis and cytotoxicity by amyloidbeta. Nature. 2000;403(6765):98–103.

1246 Articles

|

JNCI

46. Rao RV, Hermel E, Castro-Obregon S, et al. Coupling endoplasmic reticulum stress to the cell death program. Mechanism of caspase activation. J Biol Chem. 2001;276(36):33869–33874. 47. Smith WW, Jiang H, Pei Z, et al. Endoplasmic reticulum stress and mitochondrial cell death pathways mediate A53T mutant alpha-synucleininduced toxicity. Hum Mol Genet. 2005;14(24):3801–3811.

Funding Dr Daniela Trisciuoglio is a recipient of a fellowship from the Italian Foundation for Cancer Research (FIRC). This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract N01-CO-12400. This research was supported (in part) by the Developmental Therapeutics Program in the Division of Cancer Treatment and Diagnosis of the National Cancer Institute.

Notes The authors take sole responsibility for the design, execution, analysis, and interpretation of this work. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government. Manuscript received November 5, 2007; revised June 5, 2008; accepted June 13, 2008.

Vol. 100, Issue 17

|

September 3, 2008