The NSAID (non-steroidal anti-inflammatory drug) indomethacin, a cyclo-oxygenase-1 and -2 inhibitor with anti-inflammatory and analgesic properties, is known to possess anticancer activity against CRC (colorectal cancer) and other malignancies in humans; however, the mechanism underlying the anticancer action remains elusive. In the present study we show that indomethacin selectively activates the dsRNA (double-stranded RNA)-dependent protein kinase PKR in a cyclo-oxygenase-independent manner, causing rapid phosphorylation of eIF2α (the α-subunit of eukaryotic translation initiation factor 2) and inhibiting protein synthesis in colorectal carcinoma and other types of cancer cells. The PKR-mediated translational block was followed by inhibition of CRC cell proliferation and apoptosis induction. Indomethacin did not affect the activity of the eIF2α kinases PERK (PKR-like endoplasmic reticulum-resident kinase), GCN2 (general control non-derepressible-2) and HRI (haem-regulated inhibitor kinase), and induced eIF2α phosphorylation in PERK-knockout and GCN2-knockout cells, but not in PKR-knockout cells or in human PKR-silenced CRC cells, identifying PKR as a selective target for indomethacin-induced translational inhibition. The fact that indomethacin induced PKR activity in vitro, an effect reversed by the PKR inhibitor 2-aminopurine, suggests a direct effect of the drug in kinase activation. The results of the present study identify PKR as a novel target of indomethacin, suggesting new scenarios on the molecular mechanisms underlying the pleiotropic activity of this traditional NSAID.
NSAIDs (non-steroidal anti-inflammatory drugs) lower the incidence of colon cancer and decrease mortality in colon cancer patients . Among the different NSAIDs, INDO (indomethacin), a traditional COX (cyclo-oxygenase)-1 and -2 inhibitor widely used in the clinic for its potent anti-inflammatory and analgesic properties , was shown to possess anti-CRC (colorectal cancer) activity, and to have efficacy for treatment of other solid and haematological malignancies in humans [3–5]. The body of evidence that INDO has anti-neoplastic activity in animal models and in humans is actually one of the largest among the different NSAIDs (reviewed in ).
Since increased COX-2 expression is a hallmark of colon cancer and has been involved in promoting cell division, inhibiting apoptosis, altering cell adhesion and enhancing metastasis, inhibition of COX-2 activity has been considered an important aspect of the anticancer potential of INDO in this type of malignancy ; however, there is now large evidence of COX-2-independent NSAID antineoplastic effects . In the case of INDO, this drug was shown to be effective in a variety of transformed cells that do not express COX-2 [5,7]. In addition, it is well established that INDO concentrations above 100 μM, which is much higher than the COX inhibitory concentration, are necessary to inhibit cell proliferation and/or induce apoptosis in cultured human CRC cell lines (reviewed in ). These observations argue against the relevance of COX-2 inhibition as a target in colorectal epithelial cells; however, despite the advances in identifying its multiple targets, the molecular mechanism by which INDO exerts its antineoplastic potential is not completely understood.
We have previously observed that INDO causes a dysregulation of protein synthesis in virus-infected cells . Protein synthesis is a tightly regulated process that provides a highly responsive level of post-transcriptional control . Components of the translational machinery functionally converge with different fundamental signalling pathways, which are often disrupted in cancer cells; in fact, proteins involved in translation may themselves become oncogenic . For this reason drugs designed to inhibit mRNA translation are currently in preclinical and clinical development (reviewed in ). Translational control is primarily mediated at the level of initiation [9,10]. In mammalian cells the majority of mRNA translation is ‘cap-dependent’, and there are two highly regulated steps of the initiation process: (i) the binding of the small ribosomal subunit to the mRNA 5′-end mediated by eIF (eukaryotic translation initiation factor) 4E and its associated factors, and (ii) the binding of the initiator tRNA to the small ribosomal subunit mediated by eIF2. eIF2, a heterotrimeric 126 kDa complex composed of the α-, β- and γ-subunits, together with the initiator Met-tRNAMet and GTP forms the ternary complex required to bring the first methionine residue to the ribosome. Hydrolysis of eIF2-bound GTP triggered by the GTPase-activating protein eIF5 facilitates the release of the initiation factors from the ribosome. The GDP bound to eIF2 is then exchanged for GTP in a reaction catalysed by the guanine-exchange factor eIF2B [9,10]. Phosphorylation of the eIF2 α-subunit at Ser51 is a potent mechanism of translational control, since phosphorylated eIF2α functions as a dominant inhibitor of eIF2B and impedes the recycling of GTP on eIF2 required for ongoing translation [9,10]. eIF2α is the target of four distinct eukaryotic kinases, including PKR [dsRNA (double-stranded RNA)-dependent protein kinase], PERK [PKR-like ER (endoplasmic reticulum)-resident kinase], HRI (haem-regulated inhibitor kinase) and GCN2 (general control non-derepressible-2), all of which phosphorylate eIF2α at Ser51 in response to stress signals [9–11]. In particular, eIF2α phosphorylation is known to mediate apoptosis in response to activation of PKR [12,13].
PKR, previously referred to as p68 kinase or DAI (double-stranded activated inhibitor) is a serine/threonine protein kinase that generally acts as a sensor of virus replication, and modulates the antiviral and antiproliferative action of IFN (interferon) . Human PKR is a protein of 551 amino acid residues, containing an N-terminal dsRBD (dsRNA-binding domain) consisting of two dsRNA-binding motifs of 70 amino acids each, connected by a short 20 amino acid linker that regulates its activity, and a C-terminal catalytic domain shared by the eIF2α kinases PERK, HRI and GCN2 [11–13]. PKR is expressed constitutively in human cells, but its expression can be induced 5–10-fold by IFN treatment. Although it plays an important role in mediating the antiviral action of IFN, PKR is also implicated in the regulation of cell proliferation and survival in uninfected cells [11–13]. It has been clearly demonstrated that PKR mediates apoptosis induced not only by viruses, but also by oncogenes, serum-starvation or exposure to TNFα (tumour necrosis factor α), LPS (lipopolysaccharide) or other stressors (reviewed in [13,14]). Analysis of PKR effectors and events downstream of PKR activation that mediate cell death suggests an intricate pathway. PKR-dependent apoptosis is associated with FADD (Fas-associated death domain)-mediated activation of caspase 8 and up-regulation of Fas and Bax [11–14]. To different degrees, eIF2α and important regulators of cell survival, including NF-κB (nuclear factor κB) and p53, have been implicated in mediating PKR-induced apoptosis [11–14]. However, despite the different mechanisms implied, eIF2α phosphorylation has been shown to be necessary and sufficient for the PKR apoptotic response .
In the present study we demonstrate that INDO is able to activate PKR, rapidly inducing eIF2α phosphorylation and suppressing protein translation in colon cancer cells. The results suggest that PKR activation and the consequent translational block may participate in the events leading to inhibition of proliferation and induction of apoptosis in INDO-treated CRC cells.
Cell culture, transfection and treatments
Human colon carcinoma (HT29, HCT116 and Caco-2), lung adenocarcinoma (A549) and cervical adenocarcinoma (HeLa) cells were obtained from the A.T.C.C. GCN2−/−, PERK−/− and wild-type MEFs (mouse embryonic fibroblasts)  were kindly supplied by Dr Randal Kaufman (University of Michigan Medical Center, Ann Arbor, MI, U.S.A.), and PKR−/− and wild-type cells  were provided by Dr Mariano Esteban (Centro Nacional de Biotecnologia, Cantoblanco, Madrid, Spain). Cells were cultured in RPMI 1640 medium (Invitrogen) (HT29 and A549) or DMEM (Dulbecco's modified Eagle's medium; Invitrogen) (HeLa, HCT116, Caco-2, GCN2+/+, GCN2−/−, PERK+/+, PERK−/−, PKR+/+ and PKR−/−) supplemented with 10% (v/v) FBS (fetal bovine serum), 2 mM glutamine and antibiotics [penicillin (100 units/ml) and streptomycin (100 units/ml)]. INDO, ASA (aspirin), the COX-1 inhibitor SC-560 and the COX-2 inhibitor NS-398 (Sigma) were dissolved in ethanol; control cells received the same amount of vehicle. Cisplatin, PMA, 2-aminopurine and poly(I):poly(C) were purchased from Sigma. Cell numbers were evaluated using standard procedures, and cell viability was determined by Trypan Blue-exclusion assay and by the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide] into formazan conversion assay, as described previously . For PKR silencing, 5×106 HT29 cells were resuspended in 200 μl of Opti-MEM (Invitrogen) containing 6 mM glucose and 0.11 mM 2-mercaptoethanol. A heterogeneous mixture of 21–23 bp PKR-targeting siRNAs (small interfering RNAs) or non-specific siRNA (0.16 nmol, New England BioLabs) was electroporated in a 2 mm cuvette using Gene Pulser Xcell (Bio-Rad) at 160 V and 500 μF pulses. HCT116 cells were transfected using Lipofectamine™ PLUS Reagent (Invitrogen).
Analysis of apoptosis
For apoptosis detection, cytoplasmic histone-bound DNA fragments (mono- and oligo-nucleosomes) were measured by Cell-Death-Detection ELISAPLUS (Roche). Histone-associated DNA fragments were quantified spectrophotometrically using antibodies against DNA and histones in a colorimetric assay . For DNA staining, cells were fixed with methanol and stained with Hoechst 33342 (Invitrogen). Fluorescence microscopy was performed on a Leica DM-IL microscope equipped with UV excitation filters, and images were captured on a Leica DC-300 camera using Leica Image Manager 500 software. Cells exhibiting morphological features of apoptosis were counted in eight randomly selected fields. Approximately 300 nuclei were examined for each sample. As a biochemical marker of apoptosis PARP [poly(ADP-ribose) polymerase] cleavage was analysed by Western blot analysis.
Protein synthesis and Western blot analysis
For protein synthesis determination, cells were labelled with Redivue Pro-Mix [35S]Met/Cys (10–60 μCi/106 cells; GE Healthcare) for different periods of time, and the radioactivity incorporated into proteins was determined as described previously . For experiments requiring short labelling periods (5–60 min), cells were kept for 15 min in methionine-deprived medium before the addition of [35S]Met/Cys, unless otherwise specified. Samples containing the same amount of protein were separated by SDS/PAGE [3% (w/v) stacking gel, 10% (w/v) resolving gel] and processed for autoradiography . For Western blot analysis, whole-cell extracts were prepared in a high-salt extraction buffer [20 mM Hepes (pH 7.9), 0.35 M NaCl, 20% (v/v) glycerol, 1% (v/v) Nonidet P40, 1 mM MgCl2, 1 mM DTT (dithiothreitol), 0.5 mM EDTA, 0.1 mM EGTA, 1 mM PMSF and protease inhibitor cocktail (Complete™ Mini, Roche). The protein concentration of cell lysates was measured using the Bradford assay (Bio-Rad). Equal amounts of protein (25 μg) from whole-cell extracts were separated by SDS/PAGE and blotted on to nitrocellulose; filters were incubated with polyclonal anti-(phospho-Ser51-eIF2α) (p-eIF2α; Calbiochem), anti-(eIF2α-panspecific) (eIF2α; FL-315), anti-HRI (H-165), anti-PERK (H-300), anti-PARP (Santa Cruz Biotechnology) and anti-GCN2 (Cell Signaling Technology) antibodies, or with monoclonal anti-PKR (B-10), anti-COX-2 (Santa Cruz Biotechnology) and α-tubulin (Sigma) antibodies, followed by detection with horseradish peroxidase-labelled anti-(mouse IgG) or anti-(rabbit IgG) antibodies. Protein bands were visualized by an ECL (enhanced chemiluminescence) system (Pierce). Quantitative evaluation of p-eIF2α and total eIF2α protein levels was determined by Typhoon8600 imager [MDP (Molecular Dynamics Phosphor-Imager™) analysis].
Sucrose gradient polysomal profile
Polysome profiles were performed as described previously . Briefly, after the indicated treatments, cells were lysed with PL buffer [10 mM NaCl, 10 mM MgCl2, 10 mM Tris/HCl (pH 7.5), 1% (v/v) Triton X-100, 1% (v/v) sodium deoxycholate, 0.2 unit/μl RNase inhibitor (Promega) and 1 mM DTT]. After incubation on ice for 1 min, extracts were centrifuged for 1 min in a ice-cold centrifuge and supernatants were loaded on to a 5–65% (w/v) linear sucrose gradient containing 30 mM Tris/HCl (pH 7.5), 100 mM NaCl and 10 mM MgCl2, and centrifuged in a Beckman SW41 rotor (Beckman Coulter) for 3 h at 4°C and 37000 rev./min. Profiles were obtained by monitoring the absorbance at 260 nm.
Immunoprecipitation and eIF2α kinase assay
Whole-cell extracts (150 μg) were immunoprecipitated with anti-PKR, anti-PERK, anti-GCN2 or anti-HRI antibodies in the presence of 15 μl of Protein A–Sepharose at 4°C for 12 h . After extensive washing, immune complexes were resuspended in 25 μl of kinase buffer [20 mM Tris/HCl (pH 7.6), 2.5 mM MgCl2, 2.5 mM magnesium acetate, 50 mM ATP and 1 mM DTT]. Endogenous PKR, PERK, GCN2 and HRI activities were determined using recombinant eIF2α (1 μg; Cell Sciences) as the substrate. After 30 min at 30°C, incubations were terminated by the addition of SDS sample buffer. Phosphorylated eIF2α was detected by Western blot analysis using rabbit anti-p-eIF2α antibodies (Calbiochem). Levels of total eIF2α and kinase recovery from immunoprecipitates were determined by immunoblotting the same samples for a loading control. For the in vitro PKR kinase assay, INDO, the PKR inhibitor 2-aminopurine (20 mM) or the PKR inducer poly(I):poly(C) (10 μg/ml) as a positive control, were added to immunoprecipitated PKR in a final volume of 20 μl of kinase reaction buffer, and the kinase activity was determined as described above.
HT29 cells grown on coverslips were fixed with 4% paraformaldehyde (Sigma) and permeabilized with 0.5% Triton X-100 in PBS. After nuclear staining with DAPI (4′,6-diamidino-2-phenylindole; Sigma) or Hoechst 33342, cells were incubated with a polyclonal anti-p-eIF2α antibody or a monoclonal anti-α-tubulin antibody. After washing, immunocomplexes were detected with FITC-conjugated anti-rabbit and rhodamine-conjugated anti-mouse antibodies respectively. Microscopical examination was performed using a Leica DM-IL microscope equipped with UV excitation filters, and images were captured with a Leica DC-300 camera using Leica Image Manager 500 software.
EMSA (electrophoretic mobility-shift assay)
Aliquots of total extracts (12 μg of protein) were incubated with a 32P-labelled κB DNA probe  followed by analysis of DNA-binding activity by EMSA on non-denaturating 4% (w/v) PAGE . Specificity of protein–DNA complexes was verified by supershift with polyclonal antibodies specific for p65/RelA. Quantitative evaluation of NF-κB–DNA complex formation was determined by a Typhoon 8600 imager, and images were acquired using ImageQuant software (GE Healthcare).
Statistical analysis was performed using Student's t test for unpaired data. Data are means±S.D. for at least quadruplicate samples. P<0.05 was considered statistically significant.
INDO inhibits CRC cell growth and protein synthesis
To investigate the effect of INDO on protein synthesis in CRC, human colon adenocarcinoma HT29 cells were treated with different concentrations of INDO, and labelled with [35S]methionine (10 μCi/106 cells) for 24 h. In parallel samples, cell proliferation and viability was determined using the MTT assay at 72 h and by determining viable cell numbers at different times after INDO treatment. Confirming previous observations [5,22], INDO was found to affect HT29 cell proliferation at concentrations higher than 100 μM (Figures 1A and 1B). A single treatment with 400 μM INDO completely prevented HT29 cell proliferation (Figure 1B) and potently induced cell death at 72 h after treatment (Figure 1A). It should be noted that, at this concentration, the NSAID only modestly affected cell viability at 24 h after treatment (control=3.32±0.02 and INDO=10.90±1.31% mortality, n=6). Parallel with cell-growth inhibition, INDO was found to inhibit HT29 protein synthesis dose-dependently at 24 h after treatment (Figure 1C). A dramatic block of protein synthesis was detected using concentrations between 400 and 800 μM INDO under these conditions. As indicated above, at 24 h INDO only modestly affected cell viability at these concentrations (Figure 1C). In a separate experiment, HT29 cells were treated with different concentrations of INDO and labelled with [35S]methionine for the next 8 h. Protein synthesis was found to be markedly inhibited by 400 and 800 μM INDO also at this time (Figure 1D, top panel). Analysis of the electrophoretic profile of [35S]methionine-labelled proteins confirmed an overall inhibition of protein synthesis, whereas it did not reveal the appearance of proteins whose synthesis could be induced by INDO (Figure 1D, bottom panel). A dose-dependent inhibition was also obtained when protein synthesis was determined after labelling with [35S]methionine for short (15 min) periods of time (Supplementary Figure S1 at http://www.BiochemJ.org/bj/443/bj4430379add.htm).
INDO inhibits CRC cell growth and protein synthesis
The effect of the NSAID on protein synthesis was further investigated at different times after INDO treatment. HT29 cells were treated with 400 μM INDO, and labelled with [35S]methionine (15 min pulse) at 1, 2, 4 and 6 h after treatment in RPMI 1640 medium. Cells were not pre-incubated in methionine-deprived medium in this experiment. Samples containing an equal amount of protein were processed for SDS/PAGE and autoradiography. As shown in Figure 1(E), protein synthesis was already inhibited 1 h after treatment; also under these conditions, no major changes in the electrophoretic pattern of newly synthesized proteins could be observed.
Next, HT29 cells were treated with 400 μM INDO and labelled with [35S]methionine for different periods of time (5, 15, 30 and 60 min pulses). As shown in Figure 1(F), inhibition of protein synthesis occurred very rapidly (15–30 min) after treatment, suggesting an effect on mRNA translation. To investigate whether INDO was acting at the translational level, whole-cell extracts of HT29 cells treated with 400 μM INDO or vehicle for 3 h were subjected to polysome profile analysis after sucrose-gradient fractionation. As shown in Figure 1(G), INDO treatment caused a substantial decrease in polysomes, coupled to an enrichment in the 80S ribosomal fraction, which is characteristic of a reduced rate of translation initiation.
INDO induces eIF2α phosphorylation in CRC cells
As described in the Introduction, translational control is primarily mediated at the level of initiation, and phosphorylation at Ser51 of eIF2α is a potent mechanism of translational control [9,10]. Levels of p-eIF2α and total eIF2α were therefore determined in INDO-treated HT29 cells by Western blot analysis, followed by quantitative evaluation by MDP analysis. As shown in Figure 2(A), INDO induced eIF2α phosphorylation dose-dependently, starting at concentrations between 10 and 50 μM. Phosphorylation was a rapid event occurring in the first 15 min after drug administration and attenuating at 8 h after treatment (Figure 2B). High p-eIF2α cytoplasmic levels were also visualized in HT29 cells by immunofluorescence 2 h after INDO treatment (Figure 2C).
INDO induces eIF2α phosphorylation
To investigate whether eIF2α phosphorylation was specific for HT29 cells or could represent a general effect of INDO occurring in different types of human cancer cells, two different colon carcinoma cell lines (Caco-2 and HCT116), as well as cervical (HeLa) and lung (A549) carcinoma cells were treated with 400 μM INDO. After 1 h, levels of p-eIF2α and total eIF2α were determined by Western blot analysis. The results shown in Figure 2(D) demonstrate that INDO induced eIF2α phosphorylation independently of the cell type.
Interestingly the NSAID was also very effective in inducing eIF2α phosphorylation in HCT116 cells which do not express COX-2  (Figure 2D). To further investigate whether lack of COX-2 expression could interfere with INDO-triggered eIF2α phosphorylation, HT29 and HCT116 cells were treated with different concentrations of the drug and, after 1 h, were analysed by immunoblotting using anti-COX-2, anti-p-eIF2α, anti-eIF2α or anti-α-tubulin antibodies. As shown in Figures 3(A) and 3(B), consistent with previous studies , COX-2 was not detected in HCT116 cells; however, INDO induced eIF2α phosphorylation in HCT116 cells at levels comparable with those in HT29 cells. In addition, INDO-triggered eIF2α phosphorylation was not mimicked by the NSAID ASA, or by selective inhibitors of COX-1 (SC-560) and COX-2 (NS-398) at similar or higher concentrations (Figures 3C and 3D), suggesting a COX-independent mechanism.
Effect of ASA, and selective COX-1 and COX-2 inhibitors on eIF2α phosphorylation
INDO selectively induces the activity of PKR
As indicated above, four distinct eukaryotic eIF2α kinases have been identified to date: PKR, PERK, HRI and GCN2, all of which phosphorylate eIF2α at Ser51 in response to stress signals [10,11]. To identify the kinase responsible for INDO-induced eIF2α phosphorylation, HT29 cells were treated with 400 μM INDO for 2 h, and PKR, PERK, HRI and GCN2 kinase activities were analysed by kinase assay using recombinant eIF2α as a substrate. As shown in Figure 4(A), INDO did not affect PERK, GCN2 and HRI activity, whereas it induced PKR activity.
INDO selectively induces PKR activity
The possibility that INDO could directly activate PKR was then tested. After immunoprecipitation of endogenous PKR from unstimulated cells using anti-PKR antibodies, immune complexes were analysed for their ability to phosphorylate eIF2α by an in vitro kinase assay in the presence of different concentrations of INDO and of the PKR inhibitor 2-aminopurine. The PKR inducer poly(I):poly(C)  was analysed in parallel as a positive control. The results shown in Figure 4(B) demonstrate that INDO was able to induce PKR activity in vitro at levels comparable with that of poly(I):poly(C), and that this effect was reversed by 2-aminopurine, suggesting a direct effect of the drug.
PKR is responsible for INDO-induced eIF2α phosphorylation
PKR involvement in INDO-induced eIF2α phosphorylation was therefore investigated using MEFs derived from PKR−/−, PERK−/− and GCN2−/− mice. PKR+/+, PKR−/−, PERK+/+, PERK−/−, GCN2+/+ and GCN2−/− cells were treated with 800 μM INDO or vehicle for 1 h, and levels of p-eIF2α, total eIF2α and α-tubulin were determined by Western blot analysis, followed by quantitative evaluation by MDP analysis. In a parallel experiment, to determine the effect of the drug on protein synthesis, wild-type and knockout cells were labelled with [35S]methionine soon after INDO treatment for the next 3 h. The lack of PERK or GCN2 kinase activity did not influence the ability of INDO to induce eIF2α phosphorylation (Figure 5A); moreover, it did not affect the ability of INDO to inhibit protein synthesis (Supplementary Figure S2 at http://www.BiochemJ.org/bj/443/bj4430379add.htm). On the other hand, the ability of INDO to induce eIF2α phosphorylation was greatly reduced, but not totally abolished, in PKR−/− cells even at high (800 μM) concentrations (Figure 5A). PKR−/− cells were also much less sensitive to the drug-induced block of protein synthesis as compared with wild-type cells (Figure 6). In addition PKR−/− cells were found to be less sensitive to the cytotoxic effects of INDO as compared with wild-type cells (Figure 7A).
PKR is responsible for INDO-induced eIF2α phosphorylation
Involvement of PKR in the INDO-induced translational block
Role of PKR in INDO cytotoxicity
The specificity of PKR-mediated signalling was then examined in HT29 cells by establishing a PKR loss-of-function phenotype. HT29 cells were transfected with PKR siRNA or non-specific siRNA and, after 48 h, were treated with 400 μM INDO. After 2 h, whole-cell extracts were analysed by immunoblotting using anti-PKR, anti-p-eIF2α, anti-eIF2α or anti-α-tubulin antibodies. Using siRNA-mediated silencing, PKR levels were reduced more than 90% in HT29 cells; as a consequence of PKR depletion, INDO-induced eIF2α phosphorylation was impaired (Figures 5B and 5C).
Taken together, these results identify PKR as the major INDO-responsive kinase implicated in eIF2α phosphorylation during treatment with the NSAID.
siRNA-mediated silencing was also utilized to investigate the effect of PKR down-regulation on INDO cytotoxicity in CRC cells. HCT116 cells were transfected with PKR siRNA or non-specific siRNA and, after 24 h, were treated with 400 μM INDO. Cell viability was determined after 24 h using an MTT assay and by determining viable cell numbers. As shown in Figures 7(B) and 7(C), siRNA-mediated PKR silencing resulted in partial protection from INDO-mediated cytotoxicity.
INDO enhances colon cancer cell sensitivity to cisplatin
Cisplatin is used in the treatment of various solid carcinomas; however, colon cancer is often refractory to cisplatin due to the high DNA repair capacity of colon cancer cells . INDO was previously shown to enhance cisplatin cytotoxicity in tumour cells . In order to determine whether INDO could interfere with cisplatin chemotherapy in CRC, HT29 cells were treated with 20 μM cisplatin (a concentration that was previously found not to be cytotoxic in HT29 cells) alone or in the presence of 400 μM INDO. A cytotoxic concentration (80 μM) of cisplatin was used in parallel samples. After 24 h, apoptosis was determined by DNA fragmentation and PARP-cleavage analysis. The number of apoptotic cells was determined by fluorescence microscopy after Hoechst 33342 nuclear staining.
In HT29 cells, cisplatin at a concentration of 80 μM induced a 2.5-fold increase in apoptosis as determined by DNA fragmentation, whereas, as expected, it had no effect at a concentration of 20 μM (Supplementary Figure S3A at http://www.BiochemJ.org/bj/443/bj4430379add.htm). Treatment with 400 μM INDO caused a moderate increase in DNA fragmentation and PARP cleavage after 24 h (Supplementary Figures S3A and S3B), indicating that the increase in cell death previously observed at this time (Figure 1C) was due to apoptosis. Interestingly, INDO was found to sensitize HT29 cells to cisplatin treatment, markedly inducing apoptosis in HT29 cells exposed to low-dose (20 μM) cisplatin, as detected by DNA fragmentation, PARP-cleavage analysis and by determining the number of Hoechst 33342-stained cells exhibiting morphological features of apoptosis, including chromatin condensation and nuclear fragmentation (Supplementary Figures S3A and S3B).
The anticancer activity of INDO has been widely described in vitro and in vivo [3–5]. Several COX-dependent and -independent mechanisms, including PPAR (peroxisome-proliferator-activated receptor) γ and PPARα regulation, Wnt signalling pathway alteration, decrease in β-catenin and cyclin D1 levels, p38 MAPK (mitogen-activated protein kinase) activation and induction of pro-apoptotic NAG-1 (NSAID-activated gene-1) expression, have been suggested [5,25]; however, despite an extensive amount of studies, the mechanism underlying the anticancer activity has remained elusive.
In the present study we show that INDO is a potent inducer of PKR activity, causing rapid eIF2α phosphorylation in CRC and other types of cancer cells. INDO treatment induces dose-dependent phosphorylation of eIF2α at Ser51, starting at concentrations above 10 μM in HT29 cells. This is a rapid event occurring in the first 15 min after the administration of INDO, and high p-eIF2α cytoplasmic levels can be detected in HT29 cells 2 h after treatment. eIF2α phosphorylation is not specific for HT29 cells, but represents a general effect of INDO occurring in different types of human cancer cells, including two different colon carcinoma cell lines (Caco-2 and HCT116), as well as cervical (HeLa) and lung (A549) carcinoma cells. Interestingly, intraperitoneal administration of INDO to mice at concentrations of 10 mg/kg of body weight, which is in the range of the therapeutic dose in humans , also increased eIF2α phosphorylation levels in the lung and spleen at 6 h after treatment (C. Brunelli, G. Belardo, and M.G. Santoro, unpublished work).
As indicated in the Introduction, phosphorylation of eIF2α at Ser51 is a potent mechanism of translational control, since phosphorylated eIF2α functions as a dominant inhibitor of the guanine-exchange factor eIF2B and impedes the recycling of GTP on eIF2 required for ongoing translation [9,10]. Four distinct eukaryotic kinases, including PKR, PERK, HRI and GCN2, phosphorylate eIF2α at Ser51 in response to stress signals [10,11].
We have identified PKR as the kinase responsible for INDO-induced eIF2α phosphorylation, on the basis of the following observations: (i) INDO selectively activated PKR activity, whereas it had no effect on PERK, GCN2 and HRI, as determined by a kinase assay; (ii) by using PERK−/− and GCN2−/− cells, we showed that the lack of PERK or GCN2 kinase activity did not influence the ability of INDO to phosphorylate eIF2α or to inhibit protein synthesis; (iii) the ability of INDO to induce eIF2α phosphorylation was instead greatly reduced in PKR−/− cells or in HT29 cells after PKR silencing by siRNA. Taken together, these results demonstrate that PKR plays a major role in the INDO-triggered eIF2α phosphorylation; however, it cannot be excluded that other eIF2α kinases may partially contribute to this effect. These results also indicate that INDO behaves differently from the NSAID sulindac, which was recently shown to induce eIF2α phosphorylation, mainly via the activation of the ER protein kinase PERK .
In addition, the fact that PKR−/− cells were much less sensitive to the INDO-induced block of protein synthesis as compared with wild-type cells indicates that PKR plays a major role in controlling the drug-induced translational block. It should be pointed out that, at high (800 μM) concentrations, INDO was still able to partially inhibit protein synthesis in PKR−/− cells, suggesting that, in addition to eIF2α phosphorylation, other mechanisms may participate in inhibiting protein translation. In fact it was recently reported that INDO may down-regulate mTOR (mammalian target of rapamycin) signalling, an event that could also result in inhibiting protein synthesis in CRC cells .
The mechanism by which INDO triggers PKR activity remains to be established. As indicated in the Introduction, in addition to the C-terminal kinase domain shared by eIF2α kinases PERK, HRI and GCN2, PKR also contains an N-terminal dsRBD consisting of two dsRNA-binding motifs connected by a short 20 amino acid linker which regulates its activity [11–13]. In uninfected non-stressed cells, PKR is found in a monomeric latent state due to the autoinhibitory effect of its dsRBDs, which normally occlude the kinase domain; upon binding to dsRNA produced as a virus replication intermediate, PKR undergoes homodimerization and autophosphorylation on Thr446 and Thr451, leading to substrate docking, eIF2α phosphorylation and block of protein synthesis [11,28]. Apart from dsRNA, cellular proteins, such as PACT (PKR-associated activator) and its mouse orthologue RAX, upon stress-triggered phosphorylation may bind to and activate PKR . Previously it has been shown that PKR also acts as a substrate of IFN-activated JAK1 (Janus kinase 1) , and its expression can be regulated by p53 . Therefore, in addition to translational control, PKR participates in several signalling pathways regulating transcription and is implicated in the control of cell growth, differentiation and apoptosis, generally with tumour suppression function [11–14]. In particular, PKR was also shown to mediate apoptosis in the absence of viral infection, in response to chemotherapy or genotoxic conditions via an intricate pathway of effectors, which include eIF2α, ATF3 (activating transcription factor 3), ASK1 (apoptosis signal-regulating kinase 1), p53 and TCTP (translationally controlled tumour protein), as well as NF-κB [11,30,31]. NF-κB is a critical stress-regulated factor that is known to play an important role in regulating inflammation, cell proliferation and apoptosis . Since INDO has been suggested to activate the NF-κB pathway in CRC cells , we investigated the possibility that INDO-triggered PKR induction may lead to NF-κB activation in our model; however, by using EMSAs, we could not detect NF-κB DNA-binding activity in HT29 cells treated with 400 μM INDO (results not shown).
INDO-mediated PKR induction appears to be COX-independent, since eIF2α phosphorylation occurs in cells lacking COX-2 activity, and the effect of INDO is not mimicked by ASA, or by selective COX-1 and COX-2 inhibitors, even at high concentrations. Intriguingly, INDO is able to stimulate PKR activity in the absence of the natural dsRNA inducer, and does not require the presence of IFN that induces PKR expression. It should also be pointed out that INDO-mediated PKR activation is JAK1-independent, since JAK1 activity was not affected by the drug in CRC cells (results not shown). The fact that INDO was able to induce PKR activity in vitro at levels comparable with the PKR inducer poly(I):poly(C), and that this effect is reversed by the PKR inhibitor 2-aminopurine, suggests a direct effect of the drug in kinase activation.
In CRC cells, INDO-triggered PKR/eIF2α signalling leads to a translational block, inhibition of cell proliferation and induction of apoptosis. The fact that PKR−/− cells and PKR-silenced CRC cells were found to be less sensitive to the cytotoxic effects of INDO suggests that PKR activation and the consequent eIF2α-mediated translational block may participate in the events leading to CRC proliferation inhibition and induction of apoptosis in INDO-treated cells. Interestingly, INDO was also found to markedly increase apoptosis in HT29 cells exposed to low-dose cisplatin, suggesting that INDO-triggered inhibition of protein synthesis may sensitize CRC cells to cisplatin treatment.
Finally, on the basis of PKR engagement in the regulation of several critical cellular signalling pathways, the identification of PKR as a molecular target for INDO suggests new scenarios on the molecular mechanisms underlying the pleiotropic activity of this traditional NSAID.
eukaryotic translation inititation factor
electrophoretic mobility-shift assay
general control non-derepressible-2
haem-regulated inhibitor kinase
Janus kinase 1
Molecular Dynamics Phosphor-Imager™
mouse embryonic fibroblast
nuclear factor κB
non-steroidal anti-inflammatory drug
dsRNA-dependent protein kinase
PKR-like ER-resident kinase
small interfering RNA
Claudia Brunelli, Carla Amici and Chiara Fracassi performed studies on protein synthesis, cell proliferation and survival, and PKR signalling; Mara Angelini performed analysis of cell toxicity and polysomal profiles; Giuseppe Belardo participated in PKR activity studies; M. Gabriella Santoro participated in the study concept and design, and wrote the paper.
We thank Dr Mariano Esteban (Centro Nacional de Biotecnologia, Cantoblanco, Madrid, Spain) for providing the PKR−/− and wild-type cells, and Dr Randal Kaufman (University of Michigan Medical Center, Ann Arbor, MI, U.S.A.) for providing GCN2−/−, PERK−/− and wild-type MEFs.
This work was supported by the Italian Ministry of Health (ISS project ‘Ricerca Oncologica’ [grant number N. 7OCF/7] and ‘Alleanza contro il Cancro’ [grant number N. ACC12]), the EU-EICOSANOX project, and Regione Lazio Research Project ‘Caratterizzazione di principi attivi, di origine naturale e non, per patologie tumorali, cardiovascolari e infettive’.