Imexon Induces an Oxidative Endoplasmic Reticulum Stress Response in Pancreatic Cancer Cells

Published OnlineFirst January 24, 2012; DOI: 10.1158/1541-7786.MCR-11-0359

Molecular Cancer DNA Damage and Cellular Stress Responses Research

Elena V. Sheveleva1, Terry H. Landowski1, Betty K. Samulitis1, Geoffrey Bartholomeusz2, Garth Powis2, and Robert T. Dorr1

Abstract Oxidative protein folding in the endoplasmic reticulum (ER) requires strict regulation of redox homeostasis. Disruption of the lumenal redox balance induces an integrated ER stress response that is associated with reduced protein translation, increased chaperone activity, and ultimately cell death. Imexon is a small-molecule chemotherapeutic agent that has been shown to bind glutathione (GSH) and induce oxidative stress in tumor cells; however, the mechanism of cytotoxicity is not well understood. In this report, we investigate the effects of imexon on the integrated ER stress response in pancreatic carcinoma cells. Acute exposure to imexon induces an ER stress response characterized by accumulation of the oxidized form of the oxidoreductase Ero1a, phosphorylation of eIF2a, and inhibition of protein synthesis. An RNA interference chemosensitization screen identiﬁed the eukaryotic translation initiation factor eIF2B5 as a target that enhanced imexon-induced growth inhibition of MiaPaCa-2 pancreatic cancer cells, but did not signiﬁcantly augment the effects of imexon on protein synthesis. Concurrent reduction of intracellular thiols with N-acetyl cysteine reversed imexon activity, however cotreatment with superoxide scavengers had no effect, suggesting thiol binding may be a primary component of the oxidative effects of imexon. Moreover, the data suggest that disruption of the redox balance in the ER is a potential therapeutic target. Mol Cancer Res; 10(3); 392–400. 2012 AACR.

Introduction overall protein translation including reduced synthesis of the Imexon is a thiol-reactive small molecule that has shown tumor survival protein hypoxia-inducible factor-1a (9). The preclinical activity in pancreatic cancer (1, 2) and was mechanism for the global inhibition of protein synthesis by evaluated in patients with advanced pancreatic cancer in imexon in pancreatic cancer cells was not established. combination with gemcitabine (3). The mechanism of Whereas the previous mechanistic studies focused largely action is believed to involve binding to reduced sulfhydryls on the effects of imexon-induced ROS on the mitochondria, such as cysteine and glutathione (GSH; ref. 4). In cell line no studies had been carried out on possible oxidative effects models, the biologic consequences of this binding include a in the endoplasmic reticulum (ER), an organelle which is known to be highly dependant on proper redox balance to decrease in GSH levels, the accumulation of reactive oxygen fi species, loss of the mitochondrial membrane potential, allow for proper protein folding (10). Disul de bond for- fi mation in the ER is catalyzed by the transfer of electrons cytochrome C leakage from the mitochondria, and nally, fi activation of caspases-3 and -9 of the intrinsic pathway of from substrate proteins to oxidized protein disul de isom- apoptosis, and caspase-8 of the extrinsic pathway of apo- erase (PDI), resulting in the reduction of conserved thior- – edoxin motifs (CXXC). Oxidized PDI is then regenerated by ptosis (5 8). In human pancreatic cancer cells, similar effects fl were seen involving the accumulation of reactive oxygen the thiol oxidase avoprotein Ero1a. Protein assembly and species (ROS), loss of the mitochondrial membrane poten- folding in the ER requires coordinated oxidation and reduction processes, and we hypothesized that those redox pro- tial, cell-cycle arrest in G2 phase, and activation of caspases-3, -8, and -9 leading to apoptosis (1). Pancreatic cancer cells cesses could be disrupted by the pro-oxidant effects of treated with imexon also showed a marked reduction in imexon. In this study, we provide evidence that imexon interferes with redox balance in the ER, inducing an ER stress response characterized by reduced protein translation Authors' Affiliations: 1Arizona Cancer Center, University of Arizona, Tuc- and inhibition of cell growth. son, Arizona; and 2Department of Experimental Therapeutics, Division of Cancer Medicine, The University of Texas MD Anderson Cancer Center, Houston, Texas Materials and Methods Corresponding Author: Robert T. Dorr, Arizona Cancer Center, University Cell lines of Arizona, 1515 N. Campbell Avenue, Tucson, AZ 85724. Phone: 520-626- Three human pancreatic cancer cell lines were obtained 7892; Fax: 520-626-2751; E-mail: [email protected] from low-passage seed stocks at the American Type Culture doi: 10.1158/1541-7786.MCR-11-0359 Collection. A range of differentiation phenotypes was used, 2012 American Association for Cancer Research. including the poorly differentiated BxPC3 (CRL-1687), the

392 Mol Cancer Res; 10(3) March 2012

Imexon Induces Oxidative ER Stress

relatively gemcitabine-resistant Panc-1 cells (CRL-1469), Protein synthesis assay and the largely undifferentiated MiaPaCa-2 cells (CRL- Total protein synthesis was analyzed by 14C valine incor- 1420). All cell lines were maintained at 37C in humidified poration studies. Cells were plated in 96-well plates, trans- 5% CO2 in RPMI-1640 supplemented with penicillin (100 fected with nontargeting siRNA control (siCT) or si-eIF2B5 U/mL), streptomycin (100 mg/mL), and L-glutamine (2 overnight and exposed to imexon for a total of 24 hours, with mmol/L; Invitrogen), and 10% bovine calf serum 0.5 mCi/mL of 14C valine (Amersham Biosciences) added (Hyclone). Cell line identity was verified by short tandem during the last 6 hours of imexon exposure. Cells with repeat analysis by the University of Arizona Genomics incorporated radioactivity were harvested with a Packard Core (UAGC, Tucson, AZ). Imexon (4-imino-1,3-dia- FilterMate Harvester (Perkin Elmer). Data are expressed as zabicyclo[3.1.0.]-hexan-one) was provided by the Pharma- percent of control in which the control is untreated cells ceutical Resources Branch of the National Cancer Institute transfected with siCT (mean SEM, n ¼ 7). Data were under a Rapid Access to Intervention in Development analyzed by a 2 sample Wilcoxon rank-sum test. Analysis of (RAID) grant to R.T. Dorr. the dose effect within each siRNA treatment group was carried out by ANOVA, with Dunnett's correction for siRNA synthetic lethal screen multiple comparisons against a control. An siRNA human library comprising 4 siRNA duplexes for each of 21,000 target genes was obtained from Dhar- Reversal of oxidative effects of imexon by N-acetyl macon. Transfection efficiency and target gene silencing cysteine were optimized using a panel of 9 noncoding and 4 cell death MiaPaCa-2 cells were pretreated overnight with 10 mmol/ inducing sequences (siTOX, siKIF11, siPLK-1, and L N-acetyl cysteine (NAC), followed by incubation with 0, siCOPB2; ref. 11). For each target, the 4 siRNA duplexes 250, or 500 mmol/L of imexon for a total of 24 hours. provided in the library were combined and arrayed in a 384- During this 24-hour period, NAC was either maintained well format. MiaPaCa-2 cells were transfected with the ("continuous") or washed out and replaced with fresh media siRNA library for 48 hours using Lipofectamine RNAiMax ("pretreated only"). Alternatively, imexon was added 2 hours (Invitrogen), at which time 90 mmol/L imexon (72 hours before the addition of NAC ("post-imexon"). Protein syn- 3 IC10) was added and the plates were incubated an additional thesis was measured by incorporation of 500 mCi/mL of H- 48 hours. Genes identified in the high-throughput screening leucine (Perkin Elmer) for 18 hours. Data are expressed as were validated using Dharmacon ON-TARGET plus percent of control in which the control is untreated cells SMARTpool siRNAs, which were distinct sequences from (mean SEM, n ¼ 20). Statistical significance was evaluated those used in the screening process. Nontargeting siRNA by ANOVA. sequences were also used to ensure no off-target effects. The efficiency of RNA interference (RNAi) was measured by Reversal of oxidative effects of imexon by superoxide immunoblotting. scavengers MiaPaCa-2 cells were pretreated for 2 hours with the Growth inhibition analysis superoxide scavengers, 100 U/mL PEGylated superoxide Cell survival for the siRNA screening experiments were dismutase (PEG-SOD; Sigma Chemical Co.), 200 U/mL calculated by the conversion of resazurin to resorufinby PEGylated catalase (PEG-CAT; Sigma Chemical Co.), or metabolically active cells resulting in a fluorescent product 100 mmol/L 1-oxyl-2,2,6,6-tetramethyl-4-hydroxy-piperi- (em. 490; ref. 12; Promega, Inc.). Confirmatory growth dine (Tempol, Axxora, LLC), followed by incubation with 0, inhibition assays with eIF2B5 silencing were done with the 100, 250, or 500 mmol/L of imexon for a total of 24 hours. MTT assay as previously described (13). Cell growth inhi- Protein synthesis was measured by incorporation of 3H- bition data are expressed as percent survival compared with leucine (Perkin Elmer) for 18 hours. Data are expressed as fi untreated cells. The IC50 is de ned as the drug concentration percent of control in which the control is untreated cells required to produce 50% growth inhibition. (mean SEM, n ¼ 18). Statistical significance was evaluated by ANOVA. Colony-forming cell assay Colony-forming assays were conducted on MiaPaca-2 Measurement of ER stress cells transfected with eIF2B5 or nontargeting siRNA (Dhar- Transcriptional activity of the ER stress element (ERSE) macon ON-TARGET plus SmartPool siRNA), followed by was measured with the Cignal ERSE Reporter Luciferase exposure to imexon for an additional 72 hours. Imexon was Assay (SA Bioscience Qiagen). The ERSE reporter assay is a removed and cells were then plated in MethoCult (StemCell mixture of an ERSE-responsive luciferase construct and a Technologies, Inc.) following the manufacturer's instruc- constitutively expressed Renilla luciferase construct (40:1). tions and incubated for 7 to 10 days, and colony formation Luciferase activity was measured with Promega's Dual (>50 cells) was monitored by manual colony counting using Luciferase assay kit, and the results are expressed as a ratio a scoring grid. Data are expressed as percent of control in of the ERSE firefly reporter luciferase to Renilla luciferase. which the control is mock transfected (no siRNA) cells MiaPaCa-2, Panc-1, or BxPC3 cells were transfected in 96- (mean SEM, n ¼ 3). Statistical significance was evaluated well plates with an ERSE reporter, a negative control, or a by ANOVA, with post-ANOVA Tukey's analysis. positive control for 24 hours. The cells were then exposed for

www.aacrjournals.org Mol Cancer Res; 10(3) March 2012 393

Sheveleva et al.

an additional 8 hours with increasing concentrations of imexon. Positive control ER stress–inducing agents used IMX (μmol/L) 0 250 500 750 1,0001,200Dia DTT included thapsigargin (100 nmol/L), tunicamycin (2 mg/ mL), or hydrogen peroxide (0.2 mmol/L). Data are Ero1α OX1 expressed as mean SEM (n ¼ 6–9), with each experiment OX2 standardized to untreated control cells. β-Actin Immunoblotting Pancreatic cells were washed with ice-cold PBS and lysed with M-PER protein lysis buffer (Thermo Scientific) plus Figure 1. Accumulation of oxidized Ero1a by imexon (IMX)-treated protease inhibitors aprotinin, soybean trypsin inhibitor, MiaPaCa-2 cells. MiaPaCa-2 cells were incubated with the indicated leupeptin, sodium orthovanadate, and phenylmethylsulfo- concentration of imexon for 6 hours and separated on a nonreducing nylfluoride. Samples were centrifuged to remove insoluble acrylamide gel. Diamide (Dia) treatment was used to identify fully oxidized particles, and protein was quantitated by bicinchoninic acid (OX2) Ero1a (lane 7), whereas DTT was used to identify fully reduced (BCA) assay (Bio-Rad). Samples were separated by SDS- (OX1) Ero1a (lane 8). PAGE, transferred to polyvinylidene difluoride membranes (Bio-Rad), and probed with antibodies from Cell Signaling, gene (24 hours) and incubated with the indicated concen- with the exception of b-actin, which was from Sigma. trations of imexon for an additional 8 hours (Fig. 2). The data are expressed as a ratio of ERSE reporter luciferase to Ero-1a redox immunoblotting Renilla luciferase and standardized to untreated cells. We MiaPaCa-2 cells were treated with increasing concentra- identified an approximate 2-fold increase in ERSE reporter tions of imexon for 6 hours and lysed with M-PER, as activity in the 3 cell lines upon exposure to imexon. This described previously. Proteins were separated on a 4% to 20% nonreducing PAGE. Cells were also treated with the thiol reductant, dithiothreitol (DTT), or the oxidant diamide to identify the differences in the electrophoretic A Panc-1 mobility of fully reduced (OX1) versus fully oxidized (OX2) 200 Ero1a species. 150 Results 100 Imexon induces oxidative stress in the ER Oxidative protein folding in the ER relies on a series of 50 direct thiol–disulfide exchange reactions. We postulated that 0 μ imexon could induce ER stress by interfering with the redox 0 250 500 1,000 mol/L Imexon cycling of the key ER oxidoreductases, PDI, and ER oxidase B BxPC3 (Ero1a). Ero1a activity is regulated by intramolecular 200 fi disul de bonds, and the oxidation status can be determined 150 by the electrophoretic mobility of the reduced (OX1) versus the oxidized (OX2) form on a nonreducing acrylamide gel 100 (14). To determine whether the redox status of Ero1a could 50 be altered by imexon treatment, we examined the distribu- 0 tion of oxidized and reduced forms of these proteins in Luciferase activity, % control 0 250 500 1,000 μmol/L Imexon MiaPaCa-2 cells. Electrophoretic mobility on a nonreducing C MiaPaCA-2 gel showed a dose-dependent shift in the migration of Ero1a 350 from a primarily reduced form to a highly oxidized form 300 (Fig. 1). However, there was no significant change in the 250 level of Ero1a expression following incubation with imexon. 200 150 Control treatments of DTT and diamide are used to identify 100 the fully reduced (OX1) and fully oxidized (OX2) species. 50 0 0 250 500 1,000 μmol/L Imexon Imexon induces an integrated stress response Inhibition of protein translation initiation through phosphorylation of eIF2a and reduced activity of the eIF2 Figure 2. Imexon activates an ER stress response. (A) Panc-1, (B) BxPC3, initiation complex is one of the hallmarks of the integrated and (C) MiaPaCa-2 cells were transfected with an ERSE reporter, a ER-stress response (15). To determine whether imexon negative control, or a positive control for 24 hours, at which time, the cells fi were washed and then exposed to increasing concentrations of imexon treatment induces an integrated ER stress response, we rst for an additional 8 hours. Data shown are the mean SEM, n ¼ 9 cis examined activation of the -acting ERSE. Pancreatic cell (for MiaPaCa-2 and Panc-1) and n ¼ 6 (for BxPC3). , P < 0.05; lines were transfected with an ERSE dual-luciferase reporter , P < 0.01.

394 Mol Cancer Res; 10(3) March 2012 Molecular Cancer Research

Imexon Induces Oxidative ER Stress

translation. The most signiﬁcant enhancement of imexon MiaPaCa-2 Panc-1 BxPC3 cytotoxicity was seen with silencing of eIF2B5, a guanine PDI nucleotide exchange factor (GEF), which is a key regulator of Ero1α translation initiation. BiP To validate the effects of eIF2B5 gene silencing on

β-Actin pancreatic cell response to imexon, we examined growth

0 100 250 500 tun H2O2 0 100 250 500 tun H2O2 0 10050 250 500 tun inhibition in 3 pancreatic cell lines by MTT assay. The Imexon Imexon Imexon MiaPaCa-2, Panc-1, and BxPC3 cells were transfected with μ μ μ ( mol/L) ( mol/L) ( mol/L) eIF2B5 siRNA, followed by incubation with increasing concentrations of imexon for 72 hours. The data show that Figure 3. Effect of imexon on ER oxidative response proteins. Pancreatic reduction of eIF2B5 expression significantly inhibits the cell lines were incubated with the indicated concentration of imexon for growth of all 3 pancreatic cell lines, and this effect is 24 hours, and cell lysates were separated by SDS-PAGE and analyzed by enhanced with the addition of imexon (Fig. 4A). We also immunoblotting. PDI, Ero1a, BiP, and b-actin are 57, 60, 78, and 42 kDa, respectively. examined the ability of MiaPaca-2 cells to form colonies after eIF2B5 silencing and subsequent exposure to imexon. As was observed with the MTT assay, there was a 70% reduc- reached statistical significance in the Panc-1 and BxPC3 cell tion in colony formation in cells with silenced eIF2B5 and an lines at concentrations of 500 and 1,000 mmol/L imexon additional 10% to 20% reduction in colony formation with (Fig. 2). These data were validated with 3 well-characterized the addition of imexon (Fig. 4B). However, these latter ER stress–inducing agents; thapsigargin, tunicamycin, and results did not reach statistical significance. Western blot H2O2, which each showed a 3- to 5-fold increase in ERSE analysis of eiF2B5 protein expression in pancreatic cell lines activity (data not shown). following 24-hour transfection with si-eIF2B5 is shown To investigate the effects of imexon on the expression of in Fig. 4C. These results confirm a near total knockdown additional ER-resident stress response proteins, the 3 pan- of eIF2B5 expression. creatic cancer cell lines were incubated with increasing To investigate the role of eIF2B5 in imexon-mediated concentrations of imexon for 24 hours. As shown in Fig. protein synthesis inhibition, we next examined protein 3, all 3 cell lines showed increased expression of the oxido- synthesis in the MiaPaCa-2 cell line with or without silenc- reductase PDI, but as in Fig. 1, no change in the expression ing of eIF2B5. As shown in Fig. 5A, siRNA silencing of of Ero1a. In contrast to the effects of tunicamycin, imexon eIF2B5 inhibits protein synthesis by 70% (P ¼ 0.0017). In did not increase levels of the molecular chaperone BiP the absence of eIF2B5 silencing, comparable inhibition of (Grp78). This suggests that the ER stress response activated translation required exposure to 500 mmol/L imexon for 24 by imexon is not likely a secondary effect of misfolded hours. Moreover, the addition of 250 or 500 mmol/L imexon protein accumulation, but rather, may be directly related to cells with reduced eIF2B5 expression further inhibited to ER protein redox status. protein translation by only 8.5% and 5.2%, respectively. This suggests that the effects of imexon on protein synthesis ER-associated proteins participate in the imexon are mediated within the eIF2 pathway. To address this response question, we examined the direct effects of imexon on the To identify gene products that may contribute to imexon- expression eIF2B5 and other proteins associated with trans- induced ER stress and growth inhibition, we conducted an lation initiation. MiaPaCa-2, Panc-1, and BxPC3 cells were RNAi synthetic lethal screen in MiaPaCa-2 cells. An siRNA incubated with imexon for 24 hours and analyzed by array was used to individually silence 21,000 genes, and cell Western blot. In all 3 cell lines, imexon did not significantly survival was assayed by resorufin fluorescence after 48 hours alter the levels of eIF2B5, however, there was a dose- exposure to imexon. Array plates were incubated with or dependent increase in the phosphorylation of eIF2a,aswell without minimally toxic concentrations of imexon (72 hours as an increase in the levels of GTP exchange protein eIF2B2 fi IC10 of 90 mmol/L). Molecular targets were identi ed as (Fig. 6). It is important to note that protein synthesis those resulting in at least 50% change in response to imexon, inhibition (Fig. 5A) was analyzed following 24 hours of calculated as the ratio of percentage survival in the [siRNA þ drug exposure, which would produce minimal cell growth imexon] treatment population to percentage survival in inhibition, whereas growth-inhibitory effects (Fig. 4A) were [siRNA only] treatment. Genes in which silencing resulted analyzed after 72 hours of incubation with imexon. in greater than 50% cell death in the [siRNA only] treatment Previous studies have shown inhibition of protein syn- group were eliminated from consideration. Of the remaining thesis in cells treated with imexon, but a molecular mech- 18,512 genes silenced, only 20 known genes showed a cell anism for the inhibition was not identified (9, 16). To death ratio less than 50%, signifying an increased contri- determine whether the oxidative activity of imexon is bution to cell growth inhibition by the siRNA þ imexon responsible for the observed inhibition of protein transla- combination (Table 1). Three general classes of gene pro- tion, we examined new protein synthesis under 3 conditions: ducts emerged in this analysis: (i) the serpin family of pretreatment with the reducing agent NAC and continuous extracellular peptidase inhibitors; (ii) G-coupled receptor exposure during imexon incubation period; pretreatment proteins; and (iii) gene products associated with protein with NAC and wash out before imexon incubation; addition

www.aacrjournals.org Mol Cancer Res; 10(3) March 2012 395

Sheveleva et al.

Table 1. Gene products whose silencing resulted in greater than 50% cell death in the presence of 90 mmol/L imexon (IC10)

siRNA siRNA þ imexon Ratio siRNA þ Gene symbol Gene name (% survival) (% survival) imexon siRNA

EIF2B5 eIF 2B, subunit 5 78.2 27.7 35.4 SLC9A3R2 Solute carrier family 9 53.7 19.4 36.0 (sodium/hydrogen exchanger), member 3 regulator 2 ARHGAP26 Rho GTPase activating protein 26 73.1 31.0 42.5 SERPINC1 Serpin peptidase inhibitor, 85.9 36.6 42.6 clade C, member 1 OR2AE1 Olfactory receptor, family 2, 101.7 43.4 42.7 subfamily AE, member 1 ARHGEF5 Rho GEF 5 89.2 38.8 43.8 KDR Kinase insert domain receptor 87.5 38.1 43.5 SERPINE1 Serpin peptidase inhibitor, 66.1 28.8 43.5 clade E, member 1 SET SET nuclear oncogene 82.4 36.2 43.9 OR1L1 Olfactory receptor, family 1, 105.7 47.2 44.7 subfamily L, member 1 SERPINE2 Serpin peptidase inhibitor, 55.1 25.2 45.7 clade E, member 2 FBX08 F-box protein 8 65.4 30.1 46.0 ZNF200 Zinc ﬁnger protein 200 102.2 47.2 46.2 SP7 Sp7 transcription factor 36.5 17.0 46.7 ZFP36L1 Zinc ﬁnger protein 36, C3H 76.8 36.0 46.9 type-like 1 OR51G2 Olfactory receptor, family 51, 85.1 40.4 47.5 subfamily G, member 2 WASH1 WAS protein family homolog 1 90.7 43.7 48.2 MANF Mesencephalic astrocyte-derived 60.0 29.4 49.0 neurotrophic factor CYTH1 Cytohesin 1 55.1 27.3 49.6 MRLP10 Mitochondrial ribosomal 65.4 32.5 49.8 protein L10

of NAC 2 hours after imexon treatment. Imexon-mediated Protein synthesis is mediated by a series of eukaryotic inhibition of protein synthesis was reversed only when NAC translation initiation factors (eIF) that are responsive to a was present during the imexon incubation period (Fig. 5B). variety of physiologic conditions and cell stresses. One of the Furthermore, adding NAC 2 hours post-imexon resulted in primary regulators of protein synthesis is the interaction a partial reversal of imexon's oxidative effects showing that between the classic G-protein, eIF2a, and the GEF eIF2B. NAC is not simply binding to imexon and preventing it's Phosphorylation of eIF2a in response to ER stress increases uptake into the cell. These data were statistically significant its affinity for eIF2B thereby reducing the nucleotide only at the highest concentrations of imexon. exchange function and resulting in decreased translation We also examined the effect of the ROS scavengers, PEG- initiation (17). These results show that imexon induces the SOD, PEG-CAT, and tempol on imexon-induced inhibi- phosphorylation of eIF2a and reduces protein synthesis to tion of protein synthesis (Fig. 5C). Although there was a approximately the same extent as the silencing of the catalytic trend toward reversal of protein synthesis inhibition by subunit of the eIF2B GEF, eIF2B5. These effects were not PEG-CAT, especially at the 250 mmol/L imexon dose, the synergistic, suggesting a common pathway for the inhibitory difference was not statistically significant between treatment activity. groups (P ¼ 0.26846, 0.38651, and 0.74059 at 100, 250, The ER is the primary site of folding and posttranslational and 500 mmol/L imexon, respectively). Thus, none of these modification of newly synthesized proteins. Protein folding ROS scavengers, except NAC, significantly reversed protein in the ER requires the formation of disulfide bonds in a synthesis inhibition caused by imexon. relatively oxidizing environment. This is accomplished via

396 Mol Cancer Res; 10(3) March 2012 Molecular Cancer Research

Imexon Induces Oxidative ER Stress

AB 120 140 siCT 120 100 si-elF2B5 100 80 80 60 60 40 40

Survival (% of control) 20 20 0 0 IMX (% of control) formation Colony IMX 0 100 250 0 100 250 0 100 250 μ 0 100 250 0 100 250 0 100 250 ( mol/L) (μmol/L) MiaPaCa-2 Panc-1 BxPC3 Control siCT si-elF2B5

C elF2B5

β-Actin

–––++ + si-elF2B5 MiaPaCa-2 Panc-1 BxPC3

Figure 4. Growth-inhibitory effects of imexon in pancreatic cell lines with eIF2B5 silencing. Pancreatic cells were transfected with siRNA to eIF2B5 or a nontargeting sequence (siCT) followed by 72-hour incubation with imexon (IMX). A, growth inhibition was analyzed by MTT dye metabolism. Data are presented as percent of control, where 100% is siCT with no imexon (mean SD, n ¼ 8). Effect of IMX in siCT: , P < 0.05; , P < 0.01; effect of IMX in si-eIF2B5: ^, P < 0.05; ^^, P < 0.01; the effect of si-eIF2B5 versus siCT is significant at all concentrations of IMX (P < 0.01). B, growth inhibition was analyzed by colony-forming assay. Data are presented as percent of control, where control cells are mock transfected with transfection reagent but no siRNA (mean SEM, n ¼ 3). Effect of IMX: , P < 0.05. C, Western blot analysis of eIF2B5 protein expression in pancreatic cell lines following 24-hour transfection with siRNA. eIF2B5 and b-actin are 85 and 42 kDa, respectively. thiol–disulfide exchange reactions catalyzed by thiol– g-glutamyl cysteine synthetase increased the rate of pro- disulfide oxidoreductases, including PDIs and the ER tein oxidation and lead to an accumulation of oxidized oxidase 1 (Ero1) family (18, 19). The rate-limiting step PDI and Ero1a. We cannot speculate as to whether an in oxidative protein folding is generally considered to be imexon–GSH conjugate is formed within the ER lumen, the terminal FAD-dependent electron transfer by Ero1a or if cytosolic GSH depletion by imexon is sufficient to and regeneration of the fully oxidized PDI enzyme (19). reduce the transport of GSH into the ER. Nonetheless, Tocatalyzetheseoxidativereactions,theactivesitedis- these data suggest that oxidative stress within the ER is ulfides in these enzymes need to be reduced, highlighting induced by exposure to imexon. the intricate balance of oxidation and reduction that must The widespread introduction of RNAi technology has be maintained for proper protein folding. Our data show provided an opportunity to systematically identify targets for that imexon maintains Ero1a in the oxidized, and there- drug response and resistance (24). Several recent studies have by, inactive state, suggesting that disruption of the redox highlighted the use of RNAi chemosensitization screens to status of the ER may be part of imexon's mechanism of identify novel targets with chemotherapeutic agents (25– action. Since previous studies have shown that imexon 27). For example, Giroux and colleagues used a human readily reacts with both cysteine and GSH in vitro (4), it is kinome library to identify kinases that support the survival of possible that imexon is directly binding intracellular GSH pancreatic cell lines and a subset that contribute to gemci- leading to a loss of reducing equivalents in the ER and tabine resistance (27). In a similar study, Azorsa and col- interfering with redox homeostasis. leagues used siRNAs to identify the checkpoint kinase 1 Despite the relatively low ratio of GSH:GSSG in the (CHK1) as a sensitizing target in pancreatic cancer cells lumenoftheER(reportedtorangefrom1:1to1:3),GSH exposed to gemcitabine (26). Using a 21,000 member has been proposed as the primary source of reducing genome-wide siRNA library, we identified molecular path- equivalents for ER redox homeostasis (20, 21). Previous ways associated with protein synthesis as potential mediators work by Molteni and colleagues (22) and Chakravarthi of imexon-induced growth inhibition in human pancreatic and Bulleid (23) showed the critical role of GSH as an carcinoma. We initially hypothesized that this screen might antagonist of Ero1a. In these studies, depletion of GSH highlight antioxidant response pathways. In contrast, the either via cell membrane permeabilization or inhibition of major class of gene products whose inhibition significantly

www.aacrjournals.org Mol Cancer Res; 10(3) March 2012 397

Sheveleva et al.

Figure 5. A, effects of imexon and A 120 eIF2B5 silencing on protein expression. MiaPaCa-2 cells were 100 transfected with eIF2B5 siRNA or a 80 nontargeting sequence (siCT) and siCT examined for protein synthesis. 60 si-elF2B5 Data are expressed as the percent 40 of control, where the control is

(% of control) untreated cells transfected with a 20 nontargeting siRNA (mean SD, C valine incorporation n ¼ ﬁ

14 8). Imexon signi cantly 0 inhibited protein synthesis in both 0 250 500 IMX (μmol/L) siCT and si-eIF2B5–transfected B cells (, P 0.05). Knockdown of 125 eiF2B5 also inhibits protein synthesis and is augmented by 250 Imexon dose mmol/L IMX (^, P ¼ 0.0017). There 100 is no signiﬁcant augmentation of 0 μmol/L synthesis inhibition by 500 mmol/L 62.5 μmol/L P ¼ 75 IMX ( 0.75, si-eIF2B5 vs. siCT). 125 μmol/L B, effect of NAC supplementation on protein synthesis inhibition by 250 μmol/L 50 imexon. MiaPaCa-2 cells were μ (% of control) 500 mol/L pretreated with 10 mmol/L NAC 1,000 μmol/L overnight. NAC was either

H-Leucine incorporation 25

3 maintained ("continuous") or the NAC removed ("pretreated only"), 0 in the presence of varying None Continuous Pretreated only Post-imexon concentrations of imexon for 24 hours. Alternatively, the addition of NAC treatment NAC was delayed until 2 hours after imexon exposure ("post-imexon"). C 120 Protein synthesis was measured by 3H-leucine incorporation. Data are 100 expressed as percent of control, where the control is untreated cells (mean SEM, n ¼ 20). , P < 0.05; P 80 , < 0.01 C, effect of ROS scavengers on protein synthesis inhibition by imexon. MiaPaCa-2 μ 60 0 mol/L Imexon cells were pretreated for 2 hours 100 μmol/L with the 100 U/mL of PEGylated (% of control) 250 μmol/L superoxide dismutase (PEG-SOD), 40 μ 200 U/mL of PEGylated catalase H-Leucine incorporation 500 mol/L 3 (PEG-CAT), or 100 mmol/L of 1- oxyl-2,2,6,6-tetramethyl-4- 20 hydroxy-piperidine (tempol) before the addition of imexon for a total of 24 hours. Protein synthesis was 0 measured by 3H-leucine None PEG-SOD PEG-CAT Tempol incorporation. Data are expressed Superoxide scavenger as percent of control, where the control is untreated cells (mean SEM, n ¼ 18).

enhanced the cytotoxic activity of imexon comprised genes typically inhibits translation by the phosphorylation of involved in protein synthesis, and particularly eIF2B5. eIF2a (29). However, none of the enzymatic inhibitors of fi Previous studies have established imexon as a pro-oxidant H2O2 signi cantly blocked the effect of imexon on protein that can deplete intracellular thiols and induce oxidative synthesis. This indirectly suggests that the oxidative stress stress (4, 28). In addition, we have previously shown that induced by imexon may be mediated by cellular oxidants pancreatic cancer cell lines incubated with imexon display a other than H2O2 and superoxide, but still causing phos- marked inhibition of protein translation and degradation phorylation of eIF2a. before commitment to cell death (9, 16). The new findings In summary, the current findings extend the oxidizing show that the addition of NAC blocks the effects of imexon effects of imexon from the mitochondria to the ER com- on protein synthesis, suggesting that oxidative effects of the partment. The data further suggest that the alteration in ER drug may mediate the reduction in protein synthesis. The redox status may mediate the inhibitory effects of imexon on effect of imexon on eIF2a is identical to that of H2O2, which protein synthesis. However, the degree that ER protein

398 Mol Cancer Res; 10(3) March 2012 Molecular Cancer Research

Imexon Induces Oxidative ER Stress

A range of concentrations that can be achieved in patients with cancer (30). For example, at the maximally tolerated phase I MiaPaCa-2 Panc-1 BxPC3 dose of 875 mg/m2/d, imexon peak plasma levels of 715 elF2B5 mmol/L were routinely obtained. elF2B5 fl β-Actin Disclosure of Potential Con icts of Interest R.T. Dorr is a corporate officer and unpaid consultant to and holds stock in 0 10050 250 500 050 100 250 500 0 10050250 500 AmpliMed Corporation. He is also an inventor who has established University of Imexon Imexon Imexon Arizona patents for imexon and other cyanoaziridine analogs that are licensed to AmpliMed Corporation. T.H. Landowski has received research laboratory funding (μmol/L) (μmol/L) (μmol/L) from AmpliMed Corporation. No potential conflicts of interest were disclosed by B other authors. MiaPaCa-2 Panc-1 BxPC3 α Authors' Contributions p-elF2 Conception and design: E. Sheveleva, T.H. Landowski, B.K. Samulitis elF2α Development of methodology: E. Sheveleva, T.H. Landowski Acquisition of data (provided animals, acquired and managed patients, provided β-Actin facilities, etc.): B.K. Samulitis Analysis and interpretation of data (e.g., statistical analysis, biostatistics, compu- 0 100 250 500 0 100 250 500 010050 250 500 tational analysis): E. Sheveleva, T.H. Landowski, B.K. Samulitis Imexon Imexon Imexon Writing, review, and/or revision of the manuscript: E. Sheveleva, T.H. Landowski, (μmol/L) (μmol/L) (μmol/L) B.K. Samulitis, G. Bartholomeusz, G. Powis, R. T. Dorr Study supervision: E. Sheveleva, T.H. Landowski, B.K. Samulitis

Figure 6. A and B, effect of imexon on proteins associated with ER Acknowledgments The authors thank Geoffrey V. Grandjean for technical expertise in carrying out translational control. The pancreatic cell lines MiaPaCa-2, Panc-1, or siRNA gene silencing studies. BxPC3 were incubated with the indicated concentration of imexon for 24 hours. Cell lysates were separated by SDS-PAGE and analyzed by immunoblotting. Data shown are representative of 3 independent Grant Support experiments. eiF2B5, eiF2B2, p-EIF2a, eIF2Ba, and b-actin are 85, 39, This work was supported by NIH CA017094 (R.T. Dorr), NIH CA023074 (D. 38, 38, and 42 kDa, respectively. Alberts), NIH CA077204 (G. Powis), and NIH CA120613 (G. Powis). 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 oxidation and reduced protein synthesis contribute to overall 18 U.S.C. Section 1734 solely to indicate this fact. imexon-induced tumor cell growth inhibition is not known. Importantly, the concentrations of imexon used to inhibit Received July 27, 2011; revised January 18, 2012; accepted January 19, 2012; protein translation in the current experiments are within the published OnlineFirst January 24, 2012.

References 1. Dorr RT, Raymond MA, Landowski TH, Roman NO, Fukushima S. 10. Sevier CS. New insights into oxidative folding. J Cell Biol 2010;188: Induction of apoptosis and cell cycle arrest by imexon in human 757–8. pancreatic cancer cell lines. Int J Gastrointest Cancer 2005;36:15–28. 11. Bartholomeusz G, Cherukuri P, Kingston J, Cognet L, Lemos R, Leeuw 2. Roman NO, Samulitis BK, Wisner L, Landowski TH, Dorr RT. Imexon TK , et al. In vivo therapeutic silencing of hypoxia-inducible factor 1 enhances gemcitabine cytotoxicity by inhibition of ribonucleotide alpha (HIF-1a) using single-walled carbon nanotubes noncovalently reductase. Cancer Chemother Pharm 2011;67:183–92. coated with siRNA. Nano Res 2009;2:279–91. 3. Cohen SJ, Zalupski MM, Modiano MR, Conkling P, Patt YZ, Davis P, 12. O'Brien J, Wilson I, Orton T, Pognan F. Investigation of the Alamar Blue et al. A phase I study of imexon plus gemcitabine as ﬁrst-line therapy (resazurin) ﬂuorescent dye for the assessment of mammalian cell for advanced pancreatic cancer. Cancer Chemother Pharmacol cytotoxicity. Eur J Biochem 2000;267:5421–6. 2010;66:287–94. 13. Mosmann T. Rapid colorimetric assay for cellular growth and survival: 4. Iyengar BS, Dorr RT, Remers WA. Chemical basis for the biological application to proliferation and cytotoxicity assays. J Immunol Meth- activity of imexon and related cyanoaziridines. J Med Chem 2004;47: ods 1983;65:55–63. 218–23. 14. Benham AM, Cabibbo A, Fassio A, Bulleid N, Sitia R, Braakman I. The 5. Dvorakova K, Payne CM, Tome ME, Briehl MM, McClure T, Dorr RT. CXXCXXC motif determines the folding, structure and stability of Induction of oxidative stress and apoptosis in myeloma cells by the human Ero1-Lalpha. EMBO J 2000;19:4493–502. aziridine-containing agent imexon. Biochem Pharmacol 2000;60: 15. HardingHP,ZhangY,ZengH,NovoaI,LuPD,CalfonM,etal. 749–58. An integrated stress response regulates amino acid metabolism 6. Dvorakova K, Waltmire CN, Payne CM, Tome ME, Briehl MM, Dorr RT. and resistance to oxidative stress. Mol Cell 2003;11:619– Induction of mitochondrial changes in myeloma cells by imexon. Blood 33. 2001;97:3544–51. 16. Hersh EM, Gschwind CR, Taylor CW, Dorr RT, Taetle R, Salmon SE. 7. Dvorakova K, Payne CM, Landowski TH, Tome ME, Halperin DS, Dorr Antiproliferative and antitumor activity of the 2-cyanoaziridine com- RT. Imexon activates an intrinsic apoptosis pathway in RPMI8226 pound imexon on tumor cell lines and fresh tumor cells in vitro. J Natl myeloma cells. Anticancer Drugs 2002;13:1031–42. Cancer Inst 1992;84:1238–44. 8. Evens AM, Prachand S, Shi B, Paniaqua M, Gordon LI, Gartenhaus RB. 17. Boyce M, Bryant KF, Jousse C, Long K, Harding HP, Scheuner D, et al. Imexon-induced apoptosis in multiple myeloma tumor cells is cas- A selective inhibitor of eIF2alpha dephosphorylation protects cells pase-8 dependent. Clin Cancer Res 2004;10:1481–91. from ER stress. Science 2005;307:935–9. 9. Samulitis BK, Landowski TH, Dorr RT. Inhibition of protein synthesis by 18. Ellgaard L, Ruddock LW. The human protein disulphide isomerase imexon reduces HIF-1alpha expression in normoxic and hypoxic family: substrate interactions and functional properties. EMBO Rep pancreatic cancer cells. Invest New Drugs 2009;27:89–98. 2005;6:28–32.

www.aacrjournals.org Mol Cancer Res; 10(3) March 2012 399

Sheveleva et al.

19. Hatahet F, Ruddock LW. Protein disulfide isomerase: a critical eval- 25. Whitehurst AW, Bodemann BO, Cardenas J, Ferguson D, Girard L, uation of its function in disulfide bond formation. Antioxid Redox Signal Peyton M, et al. Synthetic lethal screen identification of chemosensi- 2009;11:2807–50. tizer loci in cancer cells. Nature 2007;446:815–9. 20. Hwang C, Sinskey AJ, Lodish HF. Oxidized redox state of glutathione 26. Azorsa DO, Gonzales IM, Basu GD, Choudhary A, Arora S, Bisanz KM, in the endoplasmic reticulum. Science 1992;257:1496–502. et al. Synthetic lethal RNAi screening identifies sensitizing targets for 21. Bass R, Ruddock LW, Klappa P, Freedman RB. A major fraction of gemcitabine therapy in pancreatic cancer. J Transl Med 2009;7:43. endoplasmic reticulum-located glutathione is present as mixed dis- 27. Giroux V, Iovanna J, Dagorn JC. Probing the human kinome for kinases ulfides with protein. J Biol Chem 2004;279:5257–62. involved in pancreatic cancer cell survival and gemcitabine resistance. 22. Molteni SN, Fassio A, Ciriolo MR, Filomeni G, Pasqualetto E, Fagioli C, FASEB J 2006;20:1982–91. et al. Glutathione limits Ero1-dependent oxidation in the endoplasmic 28. Samulitis BK, Landowski TH, Dorr RT. Correlates of imexon sensitivity reticulum. J Biol Chem 2004;279:32667–73. in human multiple myeloma cell lines. Leuk Lymphoma 2006;47: 23. Chakravarthi S, Bulleid NJ. Glutathione is required to regulate the 97–109. formation of native disulfide bonds within proteins entering the secre- 29. Grant CM. Regulation of translation by hydrogen peroxide. Antioxid tory pathway. J Biol Chem 2004;279:39872–9. Redox Signal 2011;15:191–203 24. Iorns E, Lord CJ, Turner N, Ashworth A. Utilizing RNA interference to 30. Dragovich T, Gordon M, Mendelson D, Wong L, Modiano M, Chow HH, enhance cancer drug discovery. Nat Rev Drug Discov 2007;6: et al. Phase I trial of imexon in patients with advanced malignancy. J 556–68. Clin Oncol 2007;25:1779–84.

400 Mol Cancer Res; 10(3) March 2012 Molecular Cancer Research

Imexon Induces an Oxidative Endoplasmic Reticulum Stress Response in Pancreatic Cancer Cells

Elena V. Sheveleva, Terry H. Landowski, Betty K. Samulitis, et al.

Mol Cancer Res 2012;10:392-400. Published OnlineFirst January 24, 2012.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-11-0359

Cited articles This article cites 30 articles, 11 of which you can access for free at: http://mcr.aacrjournals.org/content/10/3/392.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/10/3/392.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/10/3/392. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.