Lovastatin‐induced apoptosis is mediated by activating transcription factor 3 and enhanced in combination with salubrinal

Abstract

We have previously demonstrated the ability of lovastatin, a potent inhibitor of mevalonate synthesis, to induce tumor‐specific apoptosis. The apoptotic effects of lovastatin were regulated in part by the integrated stress response (ISR) that regulates cellular responses to a wide variety of stress inducers. A key regulator of the ISR apoptotic response is activating transcription factor 3 (ATF3) and its target gene CHOP/GADD153. In our study, we demonstrate that in multiple lovastatin‐resistant clones of the squamous cell carcinoma (SCC) cell line SCC9, lovastatin treatment (1‐25 μM, 24 hr) in contrast to the parental line failed to significantly induce ATF3 expression. Furthermore, the SCC‐derived cell lines SCC25 and HeLa that are sensitive to lovastatin‐induced apoptosis also preferentially induce ATF3 expression compared to resistant breast (MCF‐7) and prostate carcinoma (PC3)‐derived cell lines. In HeLa cells shRNA targeting ATF3 expression as well as in ATF3‐deficient murine embryonic fibroblasts, lovastatin‐induced cytotoxicity and apoptosis were attenuated. In ex vivo HNSCC tumors, lovastatin also induced ATF3 mRNA expression in two of four tumors evaluated. Salubrinal, an agent that can sustain the activity of a key regulator of the ISR eIF2α, further increased the expression of ATF3 and demonstrated synergistic cytotoxicity in combination with lovastatin in SCC cells. Taken together, our results demonstrate preferential induction of ATF3 in lovastatin‐sensitive tumor‐derived cell lines that regulate lovastatin‐induced apoptosis. Importantly, combining lovastatin with salubrinal enhanced ATF3 expression and induced synergistic cytotoxicity in SCC cells.

Abbreviations

ATF

activating transcription factor

CI

combination index

eIF

eukaryotic initiation factor

ER

endoplasmic reticulum

HMG‐CoA

3‐hydroxy‐3‐methyl glutaryl coenzyme A

ISR

integrated stress response

LC

lethal concentration

MEFs

murine embryonic fibroblasts

MMP

mitochondrial membrane potential

MTT assay

3‐(4,5‐dimethylthiazol‐2‐yl)‐2,5‐diphenyltetrazolium bromide assay

SCC

squamous cell carcinoma

TG

thapsigargin

wt

wild‐type

Deregulated or elevated activity of 3‐hydroxy‐3‐methyl glutaryl coenzyme A (HMG‐CoA) reductase, the initial and rate‐limiting enzyme of the mevalonate pathway, has been shown in a range of different tumors.1, 2 The statin family of drugs are potent inhibitors of HMG‐CoA reductase3 and a number of studies have demonstrated that they can directly inhibit tumor cell growth, invasion and metastatic potential both in vitro and in vivo.2, 4-6 However, as single agents, statins failed to demonstrate significant clinical activity in mainly breast, prostate and central nervous system cancer patients.7 Renewed interest in statins has revolved around our and other recent work demonstrating that lovastatin can induce tumor‐specific apoptosis, especially in squamous cell carcinomas (SCCs).2, 8, 9 This work led to a Phase I study by our group evaluating lovastatin in recurrent metastatic SCC patients where 23% of enrolled patients showed disease stabilization.10 The molecular mechanism underlying the cancer‐specific cytotoxic effect of statins is required for their future clinical development.

We have recently demonstrated that the integrated stress response (ISR) is a key mediator of lovastatin‐induced apoptosis.11 Activation of ISR by a variety of cellular stresses induces phosphorylation of eukaryotic initiation factor 2α (eIF2α) ultimately leading to inhibition of global protein translation to adapt to and cope with the stress conditions or alternatively promote apoptosis.12-14 Phosphorylation of eIF2α is mediated by a family of four kinases that respond to specific stresses, a modification that prevents the assembly of the preinitiation complex and halts global protein translation.14 Under these conditions, however, the selective translation of activating transcription factor (ATF) 4 is enhanced and plays a significant role in modulating cellular responses to stress.15 Enhanced expression of the transcription factors activating transcription factor (ATF) 3, ATF4, CHOP and their target genes can mediate ISR‐induced apoptosis.13, 14, 16 In particular, ATF3 is a member of the CREB/CAAT‐binding family of transcription factors that is usually expressed at basal levels in normal cells, is rapidly induced by the ISR and can regulate its apoptotic response.17-21 Moreover, overexpression of ATF3 alone is sufficient to trigger apoptosis in HeLa cells.22 Understanding the role of ATF3 in regulating cellular stress‐induced cytotoxicity may uncover novel therapeutic approaches in the treatment of cancer.

In our study, we investigated the role of ATF3 in regulating the cytotoxic and apoptotic responses of sensitive SCC cells to lovastatin treatment. Furthermore, we assessed the potential of the addition of another stress inducer, salubrinal, to potentiate lovastatin‐induced tumor cell cytotoxicity.

Material and Methods

Tissue culture

MCF‐7 (breast adenocarcinoma), PC3 (prostate carcinoma), SCC9, SCC25 and HeLa (SCCs) were obtained from the American Type Culture Collection (Rockville, MD). The murine embryonic fibroblasts (MEFs) ATF3−/− deficient in ATF3 expression through gene knockout and their wild‐type counterparts ATF3+/+ were kindly provided by Dr. T. Hai (Ohio State University, Columbus, OH). All cell lines were maintained in DMEM (HyClone, Logan, UT) supplemented with 10% fetal bovine serum (FBS; Medicorp, Montreal, QC, Canada). The cell lines used in our study were exposed to solvent control, lovastatin (Apotex, Mississauga, ON, Canada), mevalonate (Sigma, St. Louis, MO), thapsigargin (Calbiochem, San Diego, CA) or salubrinal (ChemBridge, San Diego, CA). The SCC9 lovastatin‐resistant clones designated lovaR1–R4 were derived from continuous treatment of the parental cell line with 20 μM lovastatin for 25 days and surviving single cells were isolated and propagated in control media.

3‐(4,5‐Dimethylthiazol‐2‐yl)‐2,5‐diphenyltetrazolium bromide assay (MTT assay)

Cells were seeded in 96‐well flat‐bottomed plates (Fisher, Mississauga, ON, Canada), with ∼7,500 cells/150 µl of cell suspension and incubated overnight. After treatment with lovastatin, the MTT cell viability assay was performed as previously described.23 The plates were analyzed on a Synergy Mx Monochromator‐Based Multi‐Mode Microplate Reader using Gen5 software, both from Biotek Instruments (Winooski, VT), at 570 nm to determine the absorbance of the samples. Treatments were performed in replicates of six and the means expressed as the percent viability relative to the untreated control (100% viable). The combination effect of lovastatin with salubrinal was evaluated by the Chou–Talalay method using CalcuSyn computer software (Biosoft, Cambridge, GBR, UK). Combination index (CI) values were obtained, where a CI < 1 is a synergistic interaction, CI = 1 is additive and CI > 1 is antagonistic.

Flow cytometry

Approximately 3.5 × 10⁵ cells were seeded on 10‐cm plates (Fisher), which were incubated overnight to allow for cell attachment and recovery. The cell lines were treated with solvent control or lovastatin for 48 hr and processed as previously described.23 A total of 10,000 cells were evaluated using the Beckman Coulter Epics XL Flow Cytometer and the percentage of cells in pre‐G₁ phase was determined using the ModFit LT program (Verity Software House, Topsham, ME).

Real time RT‐PCR

Total RNA was extracted from cell samples using the RNAeasy kit (Qiagen, Frederick, MD). RNA concentrations were quantified using a NanoDrop ND‐1000 spectrophotometer (Wilmington, DE). cDNAs were prepared from total RNA extracted from cells using High Capacity Transcription kit according to the manufacturer's protocol (Applied Biosystems, Foster City, CA). The Applied Biosystems AB 7500 Real‐Time PCR system (Applied Biosystems) was used to detect amplification with Taq Man Gene Expression Assays (Applied Biosystems, ATF3, HS00231069; CHOP/ GADD153/DDIT3, HS00358796 and human GAPDH, HS4333764‐F). Three independent experiments were performed to determine the average gene expression and standard deviation.

Western blot analysis

Total cellular protein was extracted and Western blotting was performed as previously described.11 The antibodies used were specific for ATF3, CHOP, LKB1, phospho‐LKB1 (Ser428), AMPK, phospho‐AMPK (Thr172), cyclinD1, rhoA and Mcl‐1 (Santa Cruz Biotechnologies, Santa Cruz, CA); PARP, p‐eIF2α (ser51), eIF2α (Cell Signaling Technology, Danvers, MA) and actin (Sigma). Peroxidase‐conjugated AffiniPure goat anti‐mouse/rabbit IgG (Jackson ImmunoResearch, West Grove, PA) secondary antibodies were applied. The blots were then processed for detection with the Supersignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL) and imaged using Gene GNOME Imager and Genesnap Imaging Software (Syngene, Frederick, MD).

Generation of HeLa shRNAs‐expressing cells

The constructs expressing the 19 mer targeting two independent sequences within ATF3 mRNA or GFP (control shRNA) mRNAs were previously described.21 Briefly, packaging cells were transfected with ATF3‐shRNA plasmids#1, #2 or gfp‐shRNA, using FuGENE® HD Transfection Reagent (Roche, Missassauga, ON, Canada). After generation of stable clones and determination of viral titer, HeLa cells were infected with viral supernatant using 4 μg/ml polybrene. Infected cells stably expressing shRNAs were selected using 3 μg/ml puromycin (Sigma).

Mitochondrial membrane permeability assay

The MitoProbe JC‐1 Assay Kit (Molecular Probes, Eugene, OR) was used following the manufacturer's protocol to evaluate changes in mitochondrial membrane potential (MMP). MCF‐7 and HeLa cells were untreated, treated with CCCP (carbonyl cyanide 3‐chlorophenylhydrazone; 50 μM, 15 min), an MMP disrupter or lovastatin (10 μM, 24 hr), followed by incubation with a 2 μM JC‐1 (5′,6,6′‐tetrachloro‐1,1′,3,3′‐tetraethylbenzimidazolylcarbocyanine iodide) solution for 15 min at 37°C. Cells were pelleted and resuspended in PBS for flow cytometric analysis. Loss of MMP is visualized as a decrease in the red/green fluorescence intensity ratio.24

Generation of stable eIF2α or S51A‐expressing MEFs

Both the wt and S51A eIF2α‐containing plasmids (kindly provided by Dr. D. Ron, Skirball Institute, New York, NY) were digested with SmaI/BstBI to remove the CD2 region, which was then replaced by the SmaI/BstB1 Neo/G418 cassette from PCDNA3.1. Proper subcloning was confirmed by sequencing. Stable clones were selected using 800 µg/ml neomycin/G418 and then assayed for p‐eIF2α expression after TG treatment.

ATF3 promoter activity

HeLa cells were grown in DMEM culture medium to 70–80% confluency in a 96‐well plate (3 × 10⁴ cells) before transfection. Plasmid DNA (2 μg for the renilla ATF3‐luciferase chimeras and 100 ng for firefly‐luciferase as transfection efficiency control) was transfected using HD FUGEN transfection reagent (Roche) according to the manufacturer's instructions. Twenty‐four hours posttransfection cells were treated with drugs for another 24 hr and extracts were obtained using the Promega Dual Luciferase Assay kit (Promega, Madison, WI). Lysis buffer and the supernatants were assayed for firefly and luciferase activity and data were normalized to untreated controls.

Ex vivo tumor analysis

Tumor tissue from four patients with HNSCC was collected upon resection in our study that was approved by the Ottawa Hospital Research Ethics Board (Protocol# 20120559‐01H). Areas containing tumor were identified by routine histological examinations. Approximately 2‐mm cores were obtained using a sterile biopsy punch that were further sliced with a scalpel to obtain ∼2 mm × 1 mm tumor slices. The slices were randomized, two slices were placed into each well of 24‐well plate and cultured in DMEM (HyClone) supplemented with 10% heat‐inactivated FBS (Medicorp) and 100 U/ml antibiotic/antimycotic solution (Sigma). After 48 hr drug treatments, the tumor slices were processed for RNA extraction and RT‐PCR analysis of ATF3 mRNA levels as described above. Initial tumor cell viability was assessed by addition of a 10% Alamar Blue (Invitrogen) solution and measured in Synergy Mx Monochromator‐Based Multi‐Mode Microplate Reader using Gen5 software, both from Biotek Instruments, at 560‐nm excitation wavelength and 590‐nm emission wavelength.

Caspase 3 activity

Cells were analyzed with the fluorometric Caspase 3 assay kit (Enzo Life Sciences, Ann Arbor, MI) following the manufacturer's protocol and lysates in the 96‐well plate were read on a Synergy MX Plate Reader (BioTek).

Results

Lovastatin‐resistant SCC9 cells fail to induce ATF3 expression

In our study, we developed a series of lovastatin‐resistant clones from the SCC9 cell line (designated lovaR1–4). The lethal concentration (LC) 50 (MTT activity reduced by 50%) in lovastatin‐treated cells for 48 hr was 5 µM for the SCC9 parental line and greater than 25 μM for each of the four resistant sublines (lovaR1–4) (Fig. 1a). To establish that lovastatin‐induced cytotoxicity was the result of specifically effecting HMG‐CoA reductase activity, we coadministered mevalonate. This product of the HMG‐CoA reductase reaction reversed lovastatin‐induced cytotoxicity in SCC9 and its resistant clones indicating specificity to targeting of HMG‐CoA reductase (Fig. 1b). This differential sensitivity to lovastatin of the parental and resistant clones of SCC9 was specific as AT101 treatments, a pan Bcl‐2 family inhibitor that targets several antiapoptotic proteins,25 showed no significant difference in response between SCC9 and its lovaR1–4 sublines (Fig. 1c). No significant morphological differences were observed between SCC9 and its resistant clones as well (Fig. 1d). Thus, we have generated a number of SCC9‐derived sublines with specific resistance to lovastatin.

Figure 1

Characterization of SCC9‐resistant sublines. (a) MTT viability assay comparing the response of parental SCC9 cells to its resistant variants lovaR1‐4 cell lines to lovastatin treatment. Cells were treated with 0–50 µM lovastatin for 48 hr, and cell viability was assessed with the activity of untreated cells taken to be 100%. (b) MTT analysis and treatments as above with the coadministration of 100 μM mevalonate. (c) MTT viability assay comparing the response of parental SCC9 cells to its resistant variants lovaR1‐4 cell lines to AT101 treatment. Cells were treated with 0–5 µM AT101 for 48 hr. (d) Phase contrast microscopy demonstrating similar morphological characteristics of the SCC9 cells and its resistant lovaR sublines.

Besides the ISR, we also identified the ability of statins to inhibit the epidermal growth factor receptor (EGFR)23, 26 signaling and induce the LKB1/AMPK stress pathway that is activated upon ATP depletion during metabolic stress.27 In our study, we evaluated the ability of lovastatin to target each of these pathways in SCC9 and its four resistant sublines. Lovastatin targets EGFR activity through inhibiting rhoA function by inhibiting its posttranslational modification by the lipid mevalonate metabolite geranylgeranyl that properly anchors it to membranes critical to its function.26 The lack of this modification was detectable by an upward shift in molecular weight by Western blot analysis28 with lovastatin treatment (0–25 μM, 24 hr) in SCC9 and all four of its resistant clones (Fig. 2a). Inhibition of EGFR leads to growth inhibition that is associated with decreased expression of the cell cycle regulator cyclinD1. Lovastatin treatment was consistently associated with decreasing cyclinD1 levels in the SCC9 parental and its resistant sublines (Fig. 2a). Similar results were also demonstrated for the LKB1/AMPK pathway as lovastatin treatments induced the activation through specific phosphorylation of both LKB1 and AMPK proteins that were similarly observed in the SCC9 parental and its resistant sublines (Fig. 2b). The parental SCC9 cell line showed a dose‐dependent induction of ATF3 in response to lovastatin treatment (0–25 μM, 24 hr) that was absent (lovaR1, R2 and R4) or attenuated (lovaR3) in the SCC9‐resistant clones (Fig. 2c). Upregulation of the antiapoptotic Bcl‐2 family member Mcl‐1 has recently been identified as a key inhibitor of ER stress‐induced apoptosis.29 In three of four of the SCC9‐resistant clones, Mcl‐1 was overexpressed compared to the parental line suggesting the potential of induced Mcl‐1 to allow for SCC9 cell survival in the presence of lovastatin.

Figure 2

Lack of ATF3 induction after lovastatin treatment in lovastatin‐resistant SCC9 sublines (a) Western blot analysis of SCC9 cells and lovaR1–4 treated for 24 hr showing no difference in response to lovastatin regulation of downstream targets of the EGFR including rhoA and the cell cycle regulator cyclinD1. (b) Western blot analysis of activated phosphorylated LKB1 (ser428) and AMPK (thr172) along with total LKB1 and AMPK as loading controls demonstrated a dose‐dependent induction of activated LKB1 and AMPK in SCC9 and all of its resistant sublines. (c) Similar Western blot analysis shows differential induction of ATF3 in parental SCC9 cells following lovastatin treatment with elevated Mcl‐1 levels in its resistant clones and actin as a loading control.

Differential induction of ATF3 by lovastatin

The MCF‐7 breast adenocarcinoma and the PC3 prostate carcinoma cell lines are resistant, whereas the SCC‐derived cell lines SCC25 and HeLa are sensitive to lovastatin treatment‐induced cytotoxicity.9 The MTT assay results determined the LC50 in lovastatin‐treated cells for 48 hr to be less than 10 µM lovastatin for the SCC25 and HeLa cell lines, whereas for MCF‐7 and PC3 the LC50 was greater than 40 µM (Fig. 3a). Using flow cytometry to measure cellular DNA content, we demonstrated that MCF‐7 cells treated with 10 and 25 µM lovastatin for 48 hr did not display a significant apoptotic response, as a characteristic pre‐G1 apoptotic peak was not evident.30 However, in SCC25 cells with identical treatments, a pronounced pre‐G1 apoptotic peak was observed at both the 10 and 25 μM lovastatin concentrations, whereas PC3 cells displayed an intermediate response (Fig. 3b).

Figure 3

Lovastatin selectively induces ISR in sensitive cell lines. (a) MTT viability assay comparing the response of HeLa, SCC25, PC‐3 and MCF‐7 cells to lovastatin treatment. Cells were treated with 0–50 µM lovastatin for 48 hr, and cell viability was assessed with the activity of untreated cells taken to be 100%. (b) Flow cytometric analysis of subG1 apoptotic fraction as determined by propidium iodide staining of cellular DNA content. MCF‐7, PC3 and SCC25 cells were treated with 10 and 25 µM lovastatin for 48 hr, fixed, then stained with propidium iodide and analyzed by FACS. (c) Levels of ATF3, ATF4 and CHOP mRNA were analyzed by quantitative RT‐PCR after lovastatin treatment in MCF‐7 and HeLa cells. mRNA extracted from 0 to 50 µM lovastatin treatments for 24 hr. Fold changes were calculated after normalization to gapdH levels (ΔΔCt) and expressed as means (±SD) (n = 3). (d) Western blot analysis of ATF3 protein levels in MCF‐7, PC‐3, HeLa and SCC25 cells after lovastatin treatment (0–50 µM, 24 hr) and 1 µM thapsigargin (TG). Expression levels of actin were assayed as the loading control. (e) JC‐1 flow cytometric analysis of mitochondrial membrane potential (MMP) of untreated cells (10 μM, 24 hr), lovastatin‐treated (50 μM 15min) or CCCP‐treated MCF‐7 and HeLa cells. A shift from red to green fluorescence indicates loss of MMP in treated cells.

In our study, we evaluated the induction of these ISR mediators by lovastatin in these sensitive and resistant tumor cell lines described above. Using quantitative RT‐PCR, 24‐hr lovastatin treatments from 1 to 50 µM did not induce ATF3, ATF4 or CHOP mRNA expression in MCF‐7 cells (Fig. 3c). In contrast, treatment of HeLa cells readily induced ATF3, ATF4 and CHOP levels in a dose‐dependent manner (Fig. 3c). We next examined the protein expression levels of ATF3 after lovastatin treatment in the above four cell lines to confirm our mRNA induction results. In our study, 1 µM thapsigargin (TG) treatment was used as a control for the induction of the ISR.31 In all four cell lines under study, TG‐induced ATF3 indicated that the ISR is indeed functional. In the SCC‐derived cell lines, ATF3 was significantly induced in a dose‐dependent manner with levels slightly lower than 1 μM TG; however, in MCF‐7 24‐hr lovastatin treatment (1–25 µM) failed to induce ATF3 expression, whereas in PC3 cells only a weak induction was seen at the 50 µM dose (Fig. 3d).

In our study, we evaluated the role of mitochondrial dysfunction as a regulator of lovastatin‐induced apoptosis. Using JC‐1 flow cytometric analysis that is an established measure of MMP,24 we demonstrated a loss of MMP in sensitive HeLa cell but not in resistant MCF‐7 cells. Loss of MMP is visualized as a shift from the red to the green fluorescence spectrum of this dye resulting from a loss of MMP. Untreated cells and cells treated with CCCP, a MMP disruptor, were used as control treatments (Fig. 3e).

ATF3 regulates lovastatin‐induced cytotoxicity

To assess the role of ATF3 as a regulator of lovastatin‐induced cytotoxicity, we first evaluated the role of eIF2α. MEFs stably expressing either wt eIF2α or a nonphosphorylatable allele (S51A) on lovastatin‐induced cytotoxicity were evaluated. The S51A MEFs showed attenuation of lovastatin‐induced cytotoxicity compared to wt eIF2α MEFs (Fig. 4a). To confirm the role of ATF3 as a regulator of lovastatin‐induced apoptosis, we similarly evaluated MEFs deficient in ATF3 as well as HeLa cells expressing shRNA targeting ATF3 expression. The induction of ATF3 by lovastatin in ATF3+/+ MEFs was associated with the cleavage of PARP, where the 85‐kDa cleavage product was visualized and is a characteristic of apoptotic cell death.32 ATF3−/− MEFs showed no significant cleavage of PARP when treated with up to 25 µM lovastatin for 24 hr and MTT assay analysis demonstrated significant resistance to lovastatin‐induced toxicity after 48 hr relative to ATF3+/+ MEFs (Fig. 4b). To further characterize the role of ATF3 in lovastatin‐induced cytotoxicity, HeLa cells expressing two independent sequence shRNAs targeting ATF3, or green fluorescent protein (GFP) for control, were treated with 0–50 µM lovastatin for 48 hr. shRNA‐mediated knockdown of ATF3 protected cells from lovastatin‐induced cytotoxicity compared to cells expressing control GFP shRNA (Fig. 4c). As ATF3 is not readily detected under normal cell culture conditions,33, 34 the shRNAs targeting ATF3 were selected based on their ability to inhibit TG (1 µM, 24 hr)‐induced ATF3 expression (Fig. 4c).

Figure 4

ATF3 has a proapoptotic role in lovastatin‐induced cell death. (a) Immunoblot analysis of p‐eIF2α protein levels after TG treatment (1 μM, 3 hr) that was used for selection of stably transfected clones expressing either wt‐eIF2α or S51A mutant allele. MTT viability assays comparing the responses of wild‐type eIF2α and S51A mutant MEFs treated with 0–50 µM lovastatin for 48 hr. (b) MTT viability assays comparing the responses of ATF3+/+ and ATF3^–/– MEFS treated with 0–50 µM lovastatin for 48 hr, with activity of untreated cells taken to be 100%. Western blot analysis of ATF3 protein and cleaved PARP levels after lovastatin treatment of ATF3^+/+ and ATF3^–/– MEFs. Expression levels of actin were assayed as the loading control. (c) MTT viability assays comparing the responses of HeLa cells stably expressing small short hairpin RNA (shRNA) to two different sequences against ATF3 (sh‐ATF3‐1 and sh‐ATF3‐2) as well as GFP (sh‐GFP) as control. Cells were treated with 0–50 µM lovastatin for 48 hr. Immunoblot analysis was used to assess the efficiency of sh‐ATF3 clones for knockdown of ATF3 expression. Cells were treated with 1 µM TG for 24 hr to assess inhibition of ATF3 induction by these shRNAs. Expression levels of actin were assayed as the loading control.

Salubrinal enhances lovastatin‐induced ATF3 expression

Next we tested whether salubrinal, an agent that can induce and sustain p‐eIF2α and ATF3 expression,35 can enhance lovastatin‐induced ATF3 expression and cytotoxicity. SCC25 cells were treated with 10 μM lovastatin or 10, 25 and 75 μM salubrinal for 24 hr either alone or in combination and evaluated for ATF3 expression by Western blot analysis (Fig. 5a, left panel). ATF3 protein levels were induced by lovastatin and salubrinal (dose dependent) treatments and significantly enhanced in the combination treatments. In SCC25 cells both 10 μM lovastatin and 10 μM salubrinal treatments for 24 hr resulted in elevated p‐eIF2α ser51 phosphorylation levels (Fig. 5a, right panel). Similarly, in SCC25 cells treated with 10 μM lovastatin or 10, 25 or 75 μM salubrinal or their combination, ATF3 mRNA (Fig. 5b) and CHOP mRNA (Fig. 5c) levels were induced at significantly higher levels in the lovastatin and salubrinal combination treatments than with either agent alone. To confirm that this induction was driven by enhanced transcription of the ATF3 gene, we transfected ATF3 promoter constructs driving luciferase expression into HeLa cells. The constructs included the −1850 full‐length ATF3 promoter, the −1850 promoter with a mutated ATF/CRE site that plays a significant role in the ISR induction of ATF336, 37 and a promoter‐less construct (Fig. 5d). Induction of ATF3 was most robust with the lovastatin/salubrinal combination treatment (10 μM treatments, 24 hr) and was dependent upon the presence of the ATF/CRE site (Fig. 5d).

Figure 5

Salubrinal potentiates lovastatin‐induced ATF3 expression. (a) Left panel, Western blot analysis of ATF3 protein levels after solvent control, 10 μM lovastatin 10, 25 and 75 μM salubrinal and their combination treatments in SCC25 cells. Right panel, Western blot analysis of p‐eIF2α and eIF2α protein levels after lovastatin and salubrinal treatments of SCC25 cells. Lysates from 10 µM lovastatin, 10 µM salubrinal or their combination treated cells for 24 hr were evaluated. Expression levels of actin were assayed as the loading control. (b) Levels of ATF3 mRNA and (c) CHOP mRNA were analyzed by quantitative RT‐PCR after lovastatin, salubrinal and their combination treatments for 24 hr as above in SCC25 cells. Fold changes were calculated after normalization to gapdH levels (ΔΔCt) and expressed as means (±SD) (n = 4). (d) Schematic of the 5′‐deletion constructs of the human ATF3 gene‐firefly luciferase fusions. HeLa cells in 96‐well plates were transfected for 24 hr with each plasmid along with constitutively active renilla luciferase plasmid followed by treatment for 24 hr with 10 μM lovastatin, 10 μM salubrinal or both. Promoter activity was expressed as ratios of firefly/renilla RFU. Fold induction was normalized to untreated control cells. Replicate data were represented as means (±SD) (n = 3). (e) Schematic representation of ex vivo tumor tissue processing and evaluation. Levels of ATF3 mRNA were analyzed by real‐time quantitative RT‐PCR after solvent control, 10 and 25 μM lovastatin (f) or 10, 25 and 75 μM salubrinal (g) in four HNSCC ex vivo tissues. Fold changes were calculated after normalization to gapdH levels (ΔΔCt) and expressed as means (±SD) (n = 3).

To assess the potential of lovastatin and salubrinal to induce ATF3 mRNA expression in a more clinically relevant model, we assessed ex vivo HNSCC tumor tissue in culture. Tissue samples were processed after surgical excision into 2 mm × 1 mm slices as outlined (Fig. 5e). After 48‐hr treatments with either 10 and 25 μM lovastatin or 10, 25 and 75 μM salubrinal, changes in ATF3 mRNA expression were evaluated by quantitative RT‐PCR. In four HNSCC tumors evaluated, lovastatin (Fig. 5f) and salubrinal (Fig. 5g) induced ATF3 mRNA expression in tumors A and C with no change in tumors B and D.

Salubrinal enhances lovastatin‐induced cytotoxicity

We then evaluated the potential of salubrinal treatment to enhance the cytotoxicity of lovastatin in SCC cells. In the SCC25 and MCF‐7 cell lines, as a single agent, salubrinal cytotoxicity at doses of 10, 25 and 75 μM was dose dependent (Figs. 6a and 6b). In the SCC25 cell line, the addition of either 10, 25 or 75 μM salubrinal enhanced lovastatin‐induced cytotoxicity (Fig. 6a). In contrast, in MCF‐7 cells, the MTT activity observed in all the combination treatments showed no significant differences compared to the salubrinal treatments alone (Fig. 6b). Determining the CI in SCC25 cells showed that combination of either 10, 25 or 75 μM salubrinal with therapeutically relevant concentrations of lovastatin (1, 5 and 10 μM) consistently induced synergistic cytotoxicity (Fig. 6c). CI values less than 1 were consistently observed in SCC25 cells demonstrating synergistic cytotoxicity. Consistent with the increased toxicity of this combined treatment, caspase 3 activity, a marker of apoptosis induction, was increased in SCC25 cells treated with 10 μM lovastatin and 10 μM salubrinal for 24 hr and as expected this activity was suppressed by a caspase 3 inhibitor (Fig. 6d).

Figure 6

Lovastatin in combination with salubrinal induce synergistic cytotoxicity in SCC25 cells. (a and b) MTT analysis of the effects of 10, 25 and 75 μM salubrinal on 0–10 μM lovastatin treatment in SCC25 and MCF‐7 cells. (c) Calcusyn (BioSoft) software was used to calculate the dose‐effect CI values for lovastatin–salubrinal combination treatments using the Chou–Talalay method. CI values less than 1 demonstrate synergistic cytotoxicity. (d) Caspase 3 activity (U/ml) measured in untreated controls and cells treated for 24 hr with 10 μM lovastatin, 10 μM salubrinal, their combination as well as this combination with a caspase 3 inhibitor (C3in). Activity levels were determined by standard curve and expressed as means (±SD). (e) MTT analysis of the effects of 20 μM salubrinal on 0–25 μM lovastatin treatment in SCC9 and its four lovastatin‐resistant lovaR sublines. (f) Western blot analysis of SCC9 cells and lovaR1–4 treated for 24 hr with solvent control, 10 μM lovastatin, 20 μM salubrinal and their combination evaluating rhoA and cyclinD1 expression; activated phosphorylated LKB1 (ser428) and AMPK (thr172) as well as ATF3 and actin as a loading control.

The combined activity of lovastatin and salubrinal in SCC25 cells is likely regulated by their ability to induce elevated and sustained ATF3 expression. Thus, in the SCC9 sublines that are resistant to lovastatin and do not significantly induce ATF3 when treated with this agent, the combined cytotoxic effects of lovastatin and salubrinal should be attenuated. In the SCC9 parental cell line, the addition of 20 μM salubrinal enhanced lovastatin‐induced cytotoxicity; however, this combined effect was attenuated in all four resistant sublines as determined by the MTT cell viability assay after 48 hr of treatments (Fig. 6e). Lovastatin treatment (10 μM) alone and in combination with salubrinal (20 μM) induced the expression of the unmodified rhoA variant with decreased cyclinD1 levels in the SCC9 parental and the lovaR1 subline where salubrinal treatment alone had no effect after 24‐hr treatments (Fig. 6f). For the LKB1/AMPK pathway, lovastatin treatment induced the activation through specific phosphorylation of both LKB1 and AMPK proteins that were similarly observed in the SCC9 parental and the lovaR1 subline; however, salubrinal treatments inhibited their activation (Fig. 6f). ATF3 again was differentially expressed in the lovastatin‐resistant subline and enhanced expression with salubrinal was only readily detected in the parental SCC9 cell line (Fig. 6f).

Discussion

In our study, we demonstrate that lovastatin significantly induces the ISR and ATF3 expression in the SCC cell lines SCC25 and HeLa that are sensitive to its cytotoxic and apoptotic effects. In contrast, resistant breast (MCF‐7) and prostate (PC3) cancer cells under similar conditions, lovastatin treatments failed to induce ISR and ATF3 expression. The ISR pathway was functional in these tumor‐derived cell line models. This differential induction of ATF3 in lovastatin‐sensitive versus ‐resistant cell lines was recapitulated in parental SCC9 cells compared to four derived lovastatin‐resistant sublines. The proapoptotic role of ATF3 in regulating lovastatin‐induced cytotoxicity was confirmed through evaluating ATF3‐depleted MEFs and shRNA targeting ATF3 expression in HeLa cells that demonstrated attenuation of its cytotoxic effects. Similarly, ATF3−/− MEFs and shRNA knockdowns of ATF3 in HeLa cells demonstrated a role for ATF3 in regulating UV‐induced cytotoxicity and apoptosis.22 Similar differential expression of the ATF3 target gene CHOP, a proapoptotic transcription factor,38-40 in response to lovastatin treatment was also observed. Altogether, these results show that differential activation of the ISR and ATF3 expression by lovastatin plays significant roles in regulating lovastatin‐induced SCC cytotoxicity. However, as complete rescue of lovastatin‐induced cytotoxicity was not observed in eiF2α S51A mutant expressing MEFs or cells with depleted ATF3 expression, other cellular pathways likely play a role in regulating this response.

Of clinical significance, agents that can induce ISR through alternative mechanisms may enhance the tumor cell cytotoxic effects of statins. To address this potential, we evaluated the ability of salubrinal, an inhibitor of eIF2α dephosphorylation that can enhance and prolong the duration of the ISR,35 to potentiate the cytotoxic effects of lovastatin. Furthermore, we observed that the eIF2α phosphorylation mutant S51A transfected into MEFs inhibits lovastatin‐induced cytotoxicity. Combinations of lovastatin and salubrinal showed enhanced ATF3 induction and importantly induced synergistic cytotoxicity in SCC25 cells. We have also shown that ATF3 mRNA expression is also induced in a subset of HNSCC ex vivo tumor slices in culture that more closely mimics the clinical setting by lovastatin and salubrinal treatments. Recently, two reports have similarly demonstrated synergistic cytotoxicity of salubrinal in combination with the proteasome inhibitor bortezomib, which also induces the ISR and ATF3, in multiple myeloma leukemic cells.41, 42 Our recent work demonstrated that ATF3 is induced with cytotoxic doses of cisplatin21 and that HDAC inhibitors and disulfiram that also induce ATF3 enhanced cisplatin‐induced ATF3 expression and both exhibit synergistic cytotoxicity in combination with cisplatin.43, 44 This suggests that induction of elevated ATF3 levels above a threshold that drives tumor cell apoptotic response may represent a novel therapeutic approach.

Coadministration of mevalonate inhibits the induction of ATF3 and cytotoxicity associated with lovastatin treatment in SCC25 cells9 and SCC9 cells and its derived resistant clones. Thus, ATF3 induction by lovastatin is a specific consequence of the modulation of the mevalonate pathway. Furthermore, the coadministration of the more downstream mevalonate metabolite geranylgeranyl reverses lovastatin‐induced ATF3 expression and cytotoxicity11 and as such warrants further study. For example, Rab and rho proteins, which are posttranslationally modified by geranylgeranyl that acts a membrane anchors critical for their proper cellular localization and function, regulate actin cytoskeleton architecture and cellular trafficking.45, 46

Impairing the dephosphorylation of eIF2α can prolong and/or enhance stress‐induced ISR activation.47 This is likely the mechanism where combination treatments of salubrinal and lovastatin induce synergistic cytotoxicity in SCC cells. Combinations of salubrinal and lovastatin also induced high levels of ATF3 at both mRNA and protein levels as well as its target gene, CHOP. These results provide the mechanistic rationale for combining lovastatin and salubrinal that may represent a novel therapeutic approach for SCC.

Acknowledgements

The authors thank Drs T. Hai, D. Ron and S. Kitajima as well as Apotex Canada for generously providing reagents used in our study. Research support from the Canadian Institute of Health Research (J.D.) and the Joan Sealy Trust is greatly appreciated.