Inhibition of IκB kinase by a new class of retinoid-related anticancer agents that induce apoptosis

MOLECULAR AND CELLULAR BIOLOGY, Feb. 2003, p. 1061–1074 Vol. 23, No. 3 0270-7306/03/$08.00⫹0 DOI: 10.1128/MCB.23.3.1061–1074.2003 Copyright © 2003, American Society for Microbiology. All Rights Reserved. Inhibition of I␬B Kinase by a New Class of Retinoid-Related Anticancer Agents That Induce Apoptosis Yolanda Bayon, 1 Maria A. Ortiz, 1 Francisco J. Lopez-Hernandez, 1 Feng Gao, 1 Michael Karin, 2 Magnus Pfahl, 1,3 and F. Javier Piedrafita 1 * Sidney Kimmel Cancer Center, 1 Maxia Pharmaceuticals, 3 and Department of Pharmacology, University of California–San Diego School of Medicine, 2 San Diego, California Received 1 May 2002/Returned for modification 11 June 2002/Accepted 29 October 2002 The transcription factor NF-␬B is overexpressed or constitutively activated in many cancer cells, where it induces expression of antiapoptotic genes correlating with resistance to anticancer therapies. Small molecules that inhibit the NF-␬B signaling pathway could therefore be used to induce apoptosis in NF-␬B-overexpressing tumors and potentially serve as anticancer agents. We found that retinoid antagonist MX781 inhibited the activation of NF-␬B-dependent transcriptional activity in different tumor cell lines. MX781 was able to completely inhibit tumor necrosis factor alpha-mediated activation of I␬B kinase (IKK), the upstream regulator of NF-␬B. Inhibition of IKK activity resulted from direct binding of MX781 to the kinase, as demonstrated by in vitro inhibition studies. Two other molecules, MX3350-1 and CD2325, which are retinoic acid receptor gamma-selective agonists, were capable of inhibiting IKK in vitro, although they exerted variable inhibition of IKK and NF-␬B activities in intact cells in a cell type-specific manner. However, N-(4-hydroxyphenyl)-retinamide, another apoptosis-inducing retinoid, and retinoic acid as well as other nonapoptotic retinoids did not inhibit IKK. Inhibition of IKK by the retinoid-related compounds and other small molecules correlated with reduced cell proliferation and increased apoptosis. Reduced cell viability was also observed after overexpression of an IKK␤kinase-dead mutant or the I␬B␣superrepressor. The induction of apoptosis by the retinoid-related molecules that inhibited IKK was dependent on caspase activity but independent of the retinoid receptors. Thus, the presence of an excess of retinoic acid or a retinoid antagonist did not prevent the inhibition of IKK activation by MX781 and CD2325, indicating a retinoid receptor-independent mechanism of action. Aberrant expression of NF-␬B has been associated with oncogenesis and carcinogenesis (35, 46), and constitutive high levels of NF-␬B activity have been detected in tumor cells. Particularly high activity was found in breast cancer cells, and this activity increased in more advanced tumors (51, 62), although other studies have shown that in most human breast cancer cells, NF-␬B is activated independently of the hormonal status (14). NF-␬B is also activated in human melanoma, gastric carcinoma, colon carcinoma, and pancreatic carcinoma cells (3, 22, 41, 59, 74, 77). NF-␬B activation in cancer cells correlates with tumor resistance to induction of apoptosis by tumor necrosis factor (TNF) or irradiation, which is likely due to the activation of NF-␬Bresponsive genes that control survival pathways. For instance, NF-␬B induces expression of TRAF-1 and TRAF-2, as well as inhibitor-of-apoptosis proteins 1 and 2 (IAP-1 and -2) and X-linked IAP (13, 63, 72), which inhibit caspase activity and prevent apoptosis (15). Upregulation of the caspase 8 inhibitor cFLIP by NF-␬B also prevents apoptosis induced by Fas ligand (47). Furthermore, the Bcl-2 homolog Bfl-1/A1 was found to be induced by inflammatory cytokines and is a direct target of NF-␬B (24, 70, 81). Consistent with this, inhibition of NF-␬B by means of chemical inhibitors or microinjection of I␬B protein or an anti-c-Rel antibody was shown to induce apoptosis in B cells (75), while NF-␬B expression prevented TNF-␣-induced apoptosis (7, 42, 68, 71). More recently, it has been shown that inhibition of NF-␬B enhances the antitumor activity of TNF-␣and a camptothecin analog in nude mice (69). In addition to upregulating the expression of antiapoptotic genes, NF-␬B also activates target genes involved in cell proliferation (cyclin D1) (25), angiogenesis (interleukin-8, vascular endothelial growth factor) (32, 36), and metastasis (matrix metalloproteinases) (11, 30, 31, 73). Therefore, compounds that block NF-␬B activation pathways could serve as novel anticancer agents and, as recently suggested, might be particularly useful for the treatment of estrogen receptor-negative breast cancer (9, 10). NF-␬B is activated in response to inflammation, viral and bacterial infection, and various stress stimuli. In addition to apoptosis, NF-␬B also participates in cellular transformation and oncogenesis. Activation of NF-␬B usually requires phosphorylation of I␬B at two specific serine residues at the N terminus of I␬B␣(Ser32/36) or I␬B␤(Ser19/23), which triggers the polyubiquitination and subsequent proteasome-dependent degradation of the inhibitor (12, 16). This leaves free NF-␬B that translocates into the nucleus and binds DNA to activate transcription of responsive genes (reviewed in reference 4). Cytokine-responsive phosphorylation of I␬B␣and I␬B␤is achieved by the I␬B kinase (IKK) (17), which is a complex formed by at least three different protein kinases, IKK␣, IKK␤, and IKK␥. Both IKK␣and IKK␤are catalytic subunits, which * Corresponding author. Mailing address: 10835 Altman Row, San Diego, CA 92121. Phone: (858) 410-4188. Fax: (858) 450-3251. E-mail: [email protected]. 1061 Downloaded from https://journals.asm.org/journal/mcb on 18 July 2021 by 157.88.19.157. are brought together through interactions with the regulatory subunit IKK␥(reviewed in references 34 and 80). Genetic studies have demonstrated that IKK␤is essential for cytokine-dependent activation of the IKK complex, while IKK␣is required for the proliferation and differentiation of epidermal keratinocytes and skeletal patterning (29, 38–40). IKK therefore represents a major regulator of the NF-␬B pathway, and because of the central role of NF-␬B activity in important cellular functions, inhibitors of IKK show promise for the treatment of different human diseases. The anticancer activity of natural retinoids has been connected to their ability to induce differentiation and/or to their ability to inhibit cell proliferation by antagonizing the transcription factor AP-1 in a retinoid receptor-dependent manner (20). In the search for more potent compounds, synthetic retinoid-related molecules (RRMs) that induce apoptosis in a variety of cancer cell lines in vitro and in vivo were discovered (21, 43, 53, 56, 61, 67). Most of the apoptotic RRMs are analogs of CD437, a retinoic acid receptor ␥(RAR␥)-selective agonist (8) that induces apoptosis by targeting the mitochondrial pathway (45). We found that this type of novel RRM causes a strong and sustained activation of c-Jun N-terminal kinase (JNK) and p38 kinase in Jurkat cells, leading to the release of cytochrome cinto the cytosol and subsequent activation of caspases (54). This indicated that molecules with retinoid activity could also target nonretinoid signaling pathways. Another type of apoptotic RRM structurally not related to CD437 is the retinoid antagonist MX781, which showed significant anti-breast cancer activity in vitro and in vivo (21). Here we investigated the effect of the antagonist MX781 and RAR␥-selective RRMs on the NF-␬B survival pathway. We observed that MX781 inhibited TNF-␣-induced activation of NF-␬B DNA binding and transcriptional activities in various cancer cells in which it induced apoptosis. We found that the inhibition of NF-␬B activity was mediated by direct inhibition of IKK and that MX781 reversibly bound IKK and blocked kinase activity in vitro. In addition to the antagonist, several RAR␥-selective RRMs that induced apoptosis also inhibited IKK and NF-␬B activity in vitro but were overall weaker inhibitors in intact cells and exerted only partial effects in certain cancer cell lines. In contrast, retinoids that did not induce apoptosis also had no effect on IKK activity. Other inhibitors of IKK not related to retinoids prevented cell proliferation and induced apoptosis in cancer cells. Moreover, nonpharmacological inhibition of NF-␬B activity achieved by overexpression of a dominant negative mutant of IKK␤or a nonphosphorylatable form of I␬B␣significantly reduced cell viability, demonstrating that interference with the IKK/NF-␬B pathway may be sufficient to activate the apoptotic process. Synthetic peptides that inhibit caspase activity prevented the induction of apoptosis by selective RRMs, suggesting a caspase-dependent mechanism. However, the induction of caspase activity and the inhibition of IKK by the apoptotic RRMs were not affected by the presence of an excess of alltrans-retinoic acid (RA) or a retinoid antagonist, which indicates a RAR-independent mechanism of action. Thus, our data point to new backbone structures that could be useful for the development of novel IKK inhibitors as anticancer drugs. MATERIALS AND METHODS Cell lines, reagents, and retinoids.Prostatic carcinoma cell lines (DU145 and PC3) were grown in RPMI with 10% fetal bovine serum. HeLa (cervical cancer) and A549 (non-small cell lung carcinoma) cells were maintained in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum. Glutamine, penicillin G (100 U/ml), and streptomycin sulfate (100 ␮g/ml) were added to the media. TNF-␣was purchased from R&D. RA and 15-deoxy-⌬ 12-14 -prostaglandin J 2 (15dPGJ 2 ) were obtained from Biomol. Other chemicals were purchased from Sigma. Synthetic RRMs were obtained from Maxia Pharmaceuticals Inc.; 10 mM stock solutions were made in dimethyl sulfoxide and properly diluted in culture medium prior to use. Treatment with retinoids was performed in the presence of 0.5% fetal bovine serum, keeping the concentration of dimethyl sulfoxide under 0.2% (vol/vol). Prior to RRM exposure, cells were grown overnight (16 to 20 h) in medium containing 0.5% fetal bovine serum. Z-Val-Ala-Asp-fluoromethyl ketone (Z-VAD-FMK) was purchased from Enzyme System Products. Cells were incubated for 1 h with 20 ␮M Z-VAD-FMK prior to RRM treatment. Fresh Z-VAD-FMK was added to the culture medium every 12 to 16 h during prolonged experiments. Electrophoretic mobility shift assay.Nuclear extracts were prepared as previously described (18). Double-stranded oligonucleotides containing a consensus NF-␬B (GGGGACTTTCCC) or Oct-1 (ATGCAAAT) binding site were labeled with T4 polynucleotide kinase and [␥- 32 P]ATP, and 5 to 10 ␮g of protein extract was mixed with the labeled DNA (⬇100,000 cpm) in the presence of 1 ␮gof poly(dI-dC) in 20 ␮l of binding buffer (20 mM HEPES [pH 7.9], 50 mM KCl, 1 mM dithiothreitol, 2.5 mM MgCl 2 , 10% glycerol). After 30 min at 4°C (NF-␬B) or room temperature (Oct-1), protein-DNA complexes were separated in a 5% polyacrylamide gel under nondenaturing conditions. Gels were subsequently dried and analyzed by autoradiography. Transfections and luciferase assays.Typically, 5 ⫻10 4 HeLa or PC3 cells were seeded in 24-well plates the day before transfection. Cells were cotransfected with 100 ng of NF-␬B–luciferase reporter and 10 ng of NF-␬B-inducing kinase or constitutively active MEKK1 (MEKK1⌬70) expression vectors following a calcium phosphate-DNA precipitation procedure for 16 h (HeLa) or SuperFect transfection reagent for 2 h (PC3). Control cells were cotransfected with empty vector (pcDNA3). Carrier DNA (pBluescript) was added up to 1 ␮g/well. After transfection, cells were washed with phosphate-buffered saline and treated with retinoids or vehicle in Dulbecco’s modified Eagle’s medium supplemented with 0.5% charcoal-treated fetal bovine serum for 6 h, when cell extracts were prepared and assayed for firefly luciferase activity. Alternatively, cells were transfected with the NF-␬B–luciferase reporter and carrier DNA, washed after transfection, and incubated with retinoids for 2 h prior to stimulation with 200 ng of tetradecanoyl phorbol acetate per ml for an additional 4 h. Protein concentration was measured to normalize the luciferase activity. Kinase activity.Total cell extracts (150 ␮g) were incubated with 0.5 ␮gof anti-IKK␣antibody (PharminGen) in lysis buffer (50 mM Tris [pH 8], 1 mM EDTA, 0.1% NP-40, 250 mM NaCl, 10% glycerol, 1 mM Na 3 VO 4 ,20mM ␤-glycerophosphate, 2 mM phenylmethylsulfonyl fluoride, 2 ␮g of leupeptin per ml) for 30 min at 4°C and then isolated with protein A-Sepharose beads (Amersham Pharmacia Biotech) after3hofincubation. The immunoprecipitates were washed three times with lysis buffer and once with kinase buffer (20 mM HEPES [pH 7.5], 10 mM MgCl 2 ,20mM␤-glycerophosphate, 1 mM Na 3 VO 4 ). The washed beads were incubated with 20 ␮l of kinase buffer containing 2 ␮gof glutathione S-transferase (GST)-I␬B␣(1-54), 20 ␮M ATP, 2 mM dithiothreitol, and 5 ␮Ci of [␥- 32 P]ATP for 30 min at 30°C. The reactions were stopped by addition of 5⫻sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDSPAGE) sample buffer, and proteins were resolved by SDS–15% PAGE. The gels were dried and autoradiographed. Quantification of the radioactive bands was performed with a Storm phosphorimager (Molecular Dynamics). JNK, p38, and CDK2 activities were measured with immune complex kinase assays slightly modified from those described in the literature with whole-cell extracts and GST-c-Jun, GST-ATF2, and histone H1 as the substrate, respectively (52, 65). Western blots.A total of 50 ␮g of total protein extract was separated by SDS-PAGE and transferred onto Immobilon-P (Millipore) by standard procedures. Endogenous IKK␣and I␬B␣were detected with anti-IKK␣(PharminGen) or anti-I␬B␣(Santa Cruz Biotechnology) used at a 1:1,000 dilution in Blotto (5% nonfat milk in Tris-buffered saline containing 0.5% Tween 20). Ectopic expression of the different IKK constructs transfected was detected with antihemagglutinin (HA; Covance) diluted 1:1,000 in Blotto. To examine the activation of caspases, cytosol extracts were prepared in CE buffer (25 mM PIPES [piperazine-N,N⬘-bis(2-ethanesulfonic acid), pH 7], 25 mM KCl, 5 mM EGTA, 1 mM dithiothreitol, 10 ␮M cytochalasin B, 0.5% NP-40, and a mixture 1062 BAYON ET AL. MOL.CELL.BIOL. Downloaded from https://journals.asm.org/journal/mcb on 18 July 2021 by 157.88.19.157. of protease inhibitors consisting of 1 mM phenylmethylsulfonyl fluoride, 1 ␮gof leupeptin per ml, and 1 ␮g of aprotinin per ml). From 25 to 50 ␮g of protein extract was analyzed by immunoblot with antibodies that recognize active caspases 3 and 9 (Cell Signaling Technologies) following the manufacturer’s recommendations. Incubation with secondary antibodies, horseradish peroxidase-coupled donkey anti-rabbit or sheep anti-mouse immunoglobulin, was subsequently carried out for1hatroom temperature. Proteins were detected by chemiluminescence with the ECL kit (Amersham Pharmacia). Cell viability and apoptosis assays. PC3 and A549 cells were seeded in 96-well plates (3,000 cells per well) and allowed to attach. Cells were treated with retinoids or other inhibitors in medium containing 0.5% fetal bovine serum for the indicated periods of time, when cell proliferation was measured with a colorimetric assay with 3-(4,5-dimethylthiazol-2-yl)2,5-diphenyl-tetrazolium bromide (MTT) as the substrate. The induction of DNA fragmentation was determined with a cell death detection enzyme-linked immunosorbent assay (ELISA) (Roche) following the manufacturer’s instructions. A total of 10,000 cells in 96-well plates were treated in medium containing 0.5% fetal bovine serum with the compounds for 24 h and then lysed, and 20 ␮l of lysate was used for the ELISA. The induction of apoptosis was calculated with untreated cells as a control. Individual cell death was analyzed by labeling fragmented DNA with terminal deoxynucleotide transferase with a terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick-end labeling (TUNEL) labeling assay in situ cell death detection kit (fluorescein or tetramethylrhodamine red) (Roche) and subsequent flow cytometry analysis. Labeled cells were further cytospun onto glass slides and examined by fluorescence microscopy. Caspase activity was measured with 10 ␮g of cytosol extract prepared in CE buffer diluted in a total volume of 100 ␮l of caspase buffer (50 mM HEPES [pH 7.4], 100 mM NaCl, 1 mM EDTA, 5 mM dithiothreitol, 0.1% CHAPS [3-[(3cholamidopropyl)-dimethylammonio]-1-propanesulfonate], and 10% sucrose), which contained 100 ␮M acetyl-Asp-Glu-Val-Asp-7-amino-trifluoromethyl coumarin (Ac-DEVD-AFC; Enzyme Systems Products). The release of AFC was measured every 3 min as emission at 510 nm upon excitation at 390 nm with a Victor2 microplate reader (Perkin Elmer). Caspase activity (picomoles of AFC per milligram per minute) was calculated from the initial slope. A standard curve with known concentrations of free AFC (Enzyme Systems Products) was performed to calculate the amount of cleaved AFC. To detect apoptosis in transfected cells, PC3 and A549 cells (3 ⫻10 5 per well) were cotransfected in six-well plates with 1 ␮gof␤-galactosidase expression vector and 4 ␮gofIKK␤K44M, I␬B␣S32/36A, or empty vector with Superfect (Qiagen) (PC3 cells) or Lipofectin (Invitrogen) (A549 cells) following the manufacturer’s instructions. At 24 h after transfection, cells were stained basically as described before (48). Briefly, cells were washed with phosphate saline buffer and stained with a 0.4% (wt/vol) solution of trypan blue in phosphate saline buffer for 10 min. After fixation, cells were stained with Red-Gal (Research Organics) until a pink color of the desired intensity was developed. Apoptotic transfected cells appeared rounder with dark pink-purple coloration due to the double staining with Red-Gal and trypan blue. At least eight different fields containing a total of 400 to 600 transfected cells were counted to quantitate the percentage of apoptotic cells. RESULTS MX781 inhibits activation of NF-␬B.When analyzing the effect of a series of synthetic retinoid derivatives on the DNA binding activity of various transcription factors, we found that certain compounds that induced apoptosis also inhibited the binding of NF-␬B to DNA without affecting the levels of expression of NF-␬B proteins (data not shown). NF-␬B activity is strongly induced upon stimulation with cytokines such as TNF-␣. We therefore investigated the effect of several of these FIG. 1. Selective RRMs inhibit the induction of NF-␬B activity by TNF-␣. (A) Inhibition of NF-␬B DNA binding activity. DU145 cells (2 ⫻10 6 in 60-mm dishes) were grown for 16 h in medium containing 0.5% fetal bovine serum prior to exposure to solvent (none) or a 6 ␮M concentration of the indicated RRMs for 4 h. Cells were then stimulated with 20 ng of TNF-␣per ml for 10 min. Control cells were not stimulated with TNF. Nuclear extracts were prepared and examined for NF-␬B and Oct-1 DNA binding activities. Data from one representative experiment are shown. (B) Inhibition of NF-␬B transcriptional activity by retinoids. HeLa or PC3 cells were cotransfected with 100 ng of NF-␬B–luciferase reporter together with 10 ng of NF-␬Binducing kinase (dark gray), MEKK1⌬70 (light gray), or empty vector (control). After transfection, cells were incubated with 1 ␮M RA, MX781, CD2325, or MX3350-1 for6hinmedium containing 0.5% charcoal-treated fetal bovine serum. Alternatively, cells were incubated for 2 h with the indicated RRMs and then stimulated with tetradecanoyl phorbol acetate (black columns) or not (control) for 4 h. Cells were harvested and analyzed for luciferase activity and protein concentration. The percentage of luciferase activity relative to that in cells stimulated in the absence of RRMs (no RRM) is indicated. The experiment was performed at least two times in triplicate, and the results of one representative experiment are shown. VOL. 23, 2003 APOPTOTIC RETINOID-RELATED MOLECULES INHIBIT IKK 1063 Downloaded from https://journals.asm.org/journal/mcb on 18 July 2021 by 157.88.19.157. RRMs on the DNA binding activity of NF-␬B induced by TNF-␣. Strong NF-␬B DNA binding activity was observed in nuclear extracts prepared from DU145 cells stimulated with 20 ng of TNF-␣per ml (Fig. 1A). This increased NF-␬B activity was completely inhibited when cells were incubated with 6 ␮M MX781 prior to TNF-␣stimulation. The effect of this retinoid antagonist was dependent on the concentration and the time of incubation, and only those concentrations that were effective in the induction of apoptosis inhibited the DNA binding activity of NF-␬B (data not shown). Interestingly, RAR␥-selective RRMs such as MX2870-1 had a marginal effect on TNF-mediated NF-␬B-DNA binding activity, although they were potent inducers of apoptosis in different cancer cell lines (54, 56) (data not shown). Another apoptotic retinoid, N-(4-hydroxyphenyl)-retinamide (4-HPR), as well as the natural RAR agonist RA had no significant effects on NF-␬B DNA binding activity. The RRMs, however, did not affect the DNA binding activity of the POU family member Oct-1, indicating specific inhibition of NF-␬B binding activity (Fig. 1A). Similar results were observed in other cancer cell lines, including PC3, LNCaP, Jurkat, and HeLa cells (not shown). We next examined the inhibition of NF-␬B mediated transcriptional activity by selected RRMs in HeLa and PC3 cells. The antagonist strongly inhibited (by 40 to 60%) the induction of NF-␬B activity by tetradecanoyl phorbol acetate, MEKK1⌬70, and NF-␬B-inducing kinase in both cell lines (Fig. 1B). The RAR␥-selective compounds CD2325 and MX3350-1 also inhibited luciferase activity induced by all stimuli, while RA exerted a weaker but still significant effect (10 to 40% inhibition). This inhibition of NF-␬B transcriptional activity by RA, MX3350-1, and CD2325 contrasted with the small effect observed on NF-␬B DNA binding activity (see Fig. 1A) and could possibly be caused by negative cross talk mediated by liganded RARs, as previously observed with retinoid X receptors (49). The inhibition of TNF-␣-mediated NF-␬B transcriptional activation by MX781 further correlated with decreased expression of IAP-2 mRNA levels in a timeand RRM concentration-dependent manner, with no effect on other TNF-␣-induced genes (TRAF-1 and A1). In contrast, the RAR␥-selective compounds and RA showed no inhibition of the TNF-␣-upregulated genes (data not shown). MX781 inhibits TNF-mediated activation of IKK.Since IKK is an upstream activator of NF-␬B, we examined if the observed inhibition of NF-␬B activation was due to inhibition of IKK. IKK activity was stimulated by TNF-␣in DU145 cells after exposure for various periods of time to increasing concentrations of MX781. At 6 ␮M, MX781 significantly inhibited TNF-mediated activation of IKK when cells were incubated for 2 h with the compound (Fig. 2A). Activation of IKK correlated with degradation of I␬B␣, which was prevented by increasing the time of incubation with MX781. IKK␣protein levels were, however, not affected, indicating interference with IKK activity or activation. The effect of MX781 on TNF-␣-induced IKK activity and I␬B␣degradation was dependent on the concentration of the compound (Fig. 2B). At least 6 ␮M MX781 was necessary to completely inhibit IKK, which correlated with the optimal concentration required to inhibit NF-␬B DNA binding activity (data not shown). A very similar timeand dose-dependent response was observed in the other cancer cell lines investigated, including PC3 prostate carcinoma cells (data not shown). The effect of other apoptosis-inducing RRMs on TNF-induced IKK activity was next investigated. IKK activity was stimulated with TNF-␣in DU145 cells after incubation with several retinoids, including the antagonist MX781 and four RAR␥-selective RRMs (MX2870-1, MX3350-1, CD2325, and 4-HPR) at concentrations known to induce apoptosis (not shown). Interestingly, only MX781 completely inhibited TNFinduced IKK activity (Fig. 2C). Although MX2870-1, MX3350-1, and CD2325 were strong inducers of apoptosis in DU145 cells, they elicited only a partial inhibition of IKK activation, which correlated with a partial inhibition of TNF- ␣-mediated activation of NF-␬B DNA binding activity (see Fig. 1A). In contrast, IKK activity was not inhibited at all after incubation with RA or 4-HPR, which is another apoptosisinducing retinoid. While the RAR␥-selective RRMs elicited a partial effect on DU145 cells, they exerted a complete inhibition of TNF-␣-mediated activation of IKK in other cell lines, including PC3 and U937 cells (see below and data not shown). The induction of p38 mitogen-activated protein kinase and caspase activities has been shown to participate in the inhibition of NF-␬B activity in certain cell models (1, 57). Since both MX781 and the RAR␥-selective RRMs lead to strong activation of p38 mitogen-activated protein kinase and caspase activities in Jurkat cells (54) as well as in other cancer cell lines (see below), we examined the effect of caspase inhibitors (ZDEVD-FMK and Z-VAD-FMK) and p38 kinase inhibitors (SB203580 and PD169316) on RRM-mediated inhibition of IKK and NF-␬B activities. Our results indicated that the inhibition of TNF-␣-mediated IKK activation was independent of caspase and p38 kinase activities, since it was not prevented by any of these inhibitors (data not shown). Inhibition of exogenous IKK activity.To determine whether the RRMs analyzed here can inhibit the activity of overexpressed IKK and to examine if there is any selectivity towards a particular form of the kinase, we transfected cells with expression vectors containing epitope-tagged IKK␣or IKK␤.In contrast to aspirin, which inhibits IKK␤but not IKK␣(78), CD2325 and the antagonist MX781 were capable of inhibiting both IKK isoforms with similar efficacy in PC3 cells (Fig. 3) as well as in DU145 cells (not shown). These RRMs were also able to inhibit a constitutively active IKK␤kinase (S177E/ S181E) (Fig. 3), indicating that MX781 and CD2325 might function directly at the level of IKK without interfering with any upstream steps that lead to the phosphorylation and activation of IKKs. 15dPGJ 2 and arsenite have been shown to covalently bind to a critical cysteine residue (C179) located within the activation loop, blocking kinase activity in an irreversible manner (33, 58). Since RRMs could potentially form covalent bonds to sulfhydryl groups through their carboxyl group, we examined if the inhibition of IKK activity mediated by these compounds was dependent on the Cys-179 residue. A C179A mutant of IKK␤was transfected into PC3 cells, and its activation by TNF-␣was analyzed in the absence and in the presence of selective RRMs. Figure 3 shows that both antagonists MX781 and CD2325 inhibited the activation of the mutant kinase as well as the wild type, indicating that these compounds do not 1064 BAYON ET AL. MOL.CELL.BIOL. Downloaded from https://journals.asm.org/journal/mcb on 18 July 2021 by 157.88.19.157. inhibit IKK through binding to Cys-179. Furthermore, 2,3dimercaptopropane-1-sulfonic acid did not prevent RRM-mediated inhibition of IKK while interfering with arsenite action (data not shown), suggesting that interaction with a Cys residue is not responsible for RRM-mediated inhibition of IKK. MX781 and CD2325 inhibit IKK in vitro. The data presented above suggested that some of the compounds investigated here could function directly on IKK. We therefore examined the effects in vitro of RRMs on active IKK purified by immunoprecipitation. Epitope-tagged IKK␣and IKK␤were isolated from HeLa cells stimulated with 20 ng of TNF-␣per ml and incubated in vitro with different RRMs. Figure 4A shows that the antagonist MX781 effectively inhibited IKK activity. Interestingly, the RAR␥-selective compound CD2325 completely inhibited both IKK␣and IKK␤activities, while other related molecules (MX2870-1 and MX3350-1) exerted a partial but considerable inhibitory effect. This contrasted with the results obtained in intact cells, in which these RAR␥- selective RRMs elicited a comparable partial activity (see Fig. 2C and 5A below). In contrast, 4-HPR had no significant effect on IKK activity in vitro, confirming our findings in intact cells. Both MX781 and CD2325 inhibited IKK in a concentrationdependent manner, and lower concentrations of CD2325 resulted in significant inhibition of IKK activity, with a 50% inhibitory concentration of approximately 15 ␮M. Higher concentrations of MX781 were required to effectively inhibit IKK in vitro, reflecting differences in the chemical structure of the two RRMs (Fig. 4B). The inhibition of IKK activity in vitro by these two RRMs was specific for IKK, since no substantial inhibition of JNK, p38 mitogen-activated protein kinase, or CDK2 activities was observed (Fig. 4C). We investigated further whether inhibition of IKK activity was due to reversible or covalent binding. Immunopurified IKK was incubated in vitro with MX781, CD2325, or 15dPGJ 2 , and the IKK-containing beads were washed or not three times with kinase buffer prior to measuring the remaining kinase FIG. 2. MX781 inhibits TNF-␣-induced IKK activity. (A) DU145 cells were incubated with 6 ␮M MX781 for the indicated periods of time prior to stimulation with TNF-␣(20 ng/ml for 10 min). IKK activity was subsequently measured by immune complex kinase assay (KA), and the levels of IKK␣and I␬B␣proteins were analyzed by Western blot (WB). The fold induction with respect to untreated (control) cells is indicated under the KA panel. (B) DU145 cells were incubated in the presence of 0.5% fetal bovine serum with the indicated concentrations of MX781 for 4 h prior to TNF-␣stimulation for an additional 10 min where indicated. Cell extracts were obtained and analyzed as above. (C) Inhibition of TNF-␣-mediated IKK activation by RRMs. DU145 cells were incubated with the indicated RRMs at 6 ␮M or with solvent (none) for 4 h and then stimulated with 20 ng of TNF-␣per ml for 10 min. Control cells were left unstimulated. Wholecell extracts were prepared and assayed for IKK activity (KA) and IKK␣protein levels (WB). The fold induction of IKK activity is shown under the KA panel. FIG. 3. Retinoid-related molecules inhibit the activity of overexpressed IKKs. PC3 cells were transiently transfected with expression vectors bearing HA-tagged IKK␣, IKK␤, IKK␤(SS/EE), and IKK␤ (C179A), as indicated. After 16 h, cells were incubated with 6 ␮M MX781, CD2325, or solvent (none) for 4 h prior to stimulation with 20 ng of TNF-␣per ml for 10 min. Kinase activity (KA) was measured in HA immunoprecipitates with GST-I␬B␣(1-54) as the substrate. Control cells were not treated with TNF-␣. Expression levels of the different transfected IKKs were measured by immunoblot (WB) with anti-HA antibody. VOL. 23, 2003 APOPTOTIC RETINOID-RELATED MOLECULES INHIBIT IKK 1065 Downloaded from https://journals.asm.org/journal/mcb on 18 July 2021 by 157.88.19.157. activity. Figure 4D shows that the RRM-mediated inhibitory effect was reversed after the washing steps, indicating that these compounds interact with IKK in a noncovalent manner. In contrast, the inhibition of IKK by 15dPGJ 2 was not reversed, confirming the covalent nature of this interaction (58). Confirming previous observations with immunoprecipitated endogenous IKK, MX781 elicited a partial inhibition in vitro (70 to 75%), while CD2325 was a more potent inhibitor when used at the same concentrations. This result contrasted with the observations in intact cells, in which the antagonist MX781 was a more potent inhibitor of IKK in all the cell lines analyzed. Inhibition of IKK correlates with increased apoptosis. To confirm that the inhibition of IKK is relevant for the anticancer activity of MX781, we investigated the effect on cell proliferation and apoptosis of several related RRMs and other nonretinoid structures that inhibit IKK in PC3 and A549 cells. Cells were also treated with RA, MX3350-1, or CD2325 for comparison. First, the effect on TNF-␣-mediated IKK activation was analyzed. As previously observed, MX781 effectively inhibited IKK activation by TNF-␣in both cancer cell lines. MX3350-1 and CD2325 inhibited the induction of IKK activity to basal levels in PC3 cells but not in A549 cells, although the remaining IKK activity was similar in both cell lines (note that FIG. 4. Inhibition of IKK activity in vitro. (A) Retinoid-related molecules inhibit activated IKK isolated from cancer cells. HeLa cells were transfected with HA-IKK␣or HA-IKK␤expression vectors and, at 16 h posttransfection, were stimulated with TNF-␣for 10 min, after which the cells were lysed and HA-IKK␣and HA-IKK␤were immunoprecipitated with an anti-HA antibody. IKK␣- and IKK␤-containing Sepharose beads were incubated with the indicated compounds at 60 ␮M for 60 min on ice and then subjected to a kinase assay (KA) with GST-I␬B␣(1-54) as the substrate. Levels of IKK␣and IKK␤were analyzed by Western blot (WB) with an anti-HA antibody. (B) Endogenous IKK complex was immunoprecipitated with an anti-IKK␣antibody from HeLa (left) or DU145 (right) cells that were stimulated with 20 ng of TNF-␣per ml for 10 min. Isolated IKK was incubated in vitro with solvent (lanes 0) or the indicated concentration of CD2325 or MX781, and kinase activity was subsequently measured. The percentage of kinase activity remaining after incubation with the RRM is indicated under the KA panel. The chemical structures of the RRMs are shown below for comparison. (C) Jurkat cells were irradiated with UV at 100 J/m 2 and incubated for 30 min to activate JNK. To induce p38 activity, cells were treated with 0.5 M sorbitol for 20 min, and nonstimulated cells were used to isolate CDK2. Kinases were immunoprecipitated with appropriate antibodies, and kinase-containing Sepharose beads were incubated with solvent (none) or the RRMs at 60 ␮M for 60 min on ice and subsequently used in a kinase assay with the corresponding substrate. The percentage of kinase activity remaining is indicated below. (D) Inhibition of IKK activity by RRMs is reversible. HeLa cells were stimulated or not (control) with TNF-␣for 10 min, and endogenous IKK complex was isolated by immunoprecipitation with anti-IKK␣antibody. Isolated IKK was then incubated in vitro in kinase buffer with the indicated RRMs or 15dPGJ 2 at 60 ␮M in duplicate tubes. One set of immunoprecipitates was washed three times with kinase buffer (wash, right), while the second set was kept on ice (no wash, left). The remaining IKK activity was subsequently measured (KA) and quantitated with a phosphorimager. The IKK protein levels were normalized by immunoblot (WB). 1066 BAYON ET AL. MOL.CELL.BIOL. Downloaded from https://journals.asm.org/journal/mcb on 18 July 2021 by 157.88.19.157. IKK was constitutively activated in PC3 cells but not in A549 cells) (Fig. 5A). Other RRMs, CD2701, CD2665, and CD2366, as well as RA showed no effect on the activation of IKK in either PC3 or A549 cells. On the other hand, 15dPGJ 2 and salicylate inhibited IKK activation in both cell lines, while arsenite was very effective in PC3 but not in A549 cells. The effect of these molecules on cell growth was investigated after 24 h (PC3) or 72 h (A549) of incubation in the presence FIG. 5. Inhibition of IKK by small molecules correlates with the induction of cell death. (A) Effect of selective RRMs and other small molecules on IKK activation. PC3 and A549 cells were incubated with 6 ␮M RA or the indicated RRMs, 15 ␮M 15dPGJ2, 100 ␮M sodium arsenite, or 20 mM salicylate for 4 h prior to stimulation with 20 ng of TNF-␣per ml for 10 min. Subsequently, whole-cell extracts were prepared and assayed for IKK activity with an immune complex kinase assay. Kinase activity was quantitated by phosphorimager, and the fold induction with respect to nonstimulated cells (control) is indicated. (B) Inhibition of cell growth and induction of apoptosis by IKK inhibitors. PC3 (black columns) and A549 (white columns) cells were incubated with the same concentrations of the indicated molecules as described above in medium containing 0.5% fetal bovine serum. Top: 24 h (PC3) or 72 h (A549) after treatment, cell proliferation was measured with an MTT assay. Untreated cells (none) served as the control (100%). The experiment was performed twice in triplicate, and the average ⫾standard deviation is shown. Bottom: PC3 (black columns) and A549 (white columns) cells were incubated with the indicated compounds as above for 24 h, when cell extracts were obtained and the amount of DNA fragmentation in the cytosol was quantitated by ELISA. The fold induction of apoptosis relative to untreated cells is indicated. The experiment was performed twice in duplicate, and data from one representative experiment are shown. (C) Inhibition of the IKK/NF-␬B pathway reduces cell viability. PC3 and A549 cells were transiently transfected with empty vector (control) or mutants of IKK␤(K44M, KM) and I␬B␣(S32/36A, AA) together with a ␤-galactosidase expression vector. At 24 h after transfection, cells were double-stained with trypan blue and Red-Gal. Transfected cells (positive for ␤-galactosidase) and apoptotic cells (indicated by arrows) were counted in at least eight different fields (over 400 transfected cells), and the percentage of transfected apoptotic cells is indicated below each panel. The experiment was repeated at least four times with very similar results, and data from one representative experiment are shown. VOL. 23, 2003 APOPTOTIC RETINOID-RELATED MOLECULES INHIBIT IKK 1067 Downloaded from https://journals.asm.org/journal/mcb on 18 July 2021 by 157.88.19.157. of the IKK inhibitors. Only the antagonist MX781 and the RAR␥-selective RRMs MX3350-1 and CD2325, but not the other retinoids, exerted a substantial inhibition of cell proliferation (Fig. 5B). Similarly, the other IKK inhibitors exerted a strong antiproliferative activity that correlated with their effect on IKK. While 15dPGJ 2 and salicylate inhibited the growth of both PC3 and A549 cells, arsenite was only effective in PC3 cells, not in A549 cells. We next investigated if the inhibition of cell proliferation correlated with induction of apoptosis. The appearance of histone-associated DNA fragments in the cytosol was quantitated in PC3 and A549 cells treated for 24 h with RRMs or the IKK inhibitors. Only those compounds that inhibited IKK and cell growth elicited a strong induction of DNA fragmentation as a measure of cell death in either PC3 or A549 cells (Fig. 5B). RA and the selective RRMs CD2701, CD2366, and CD2665 had no effect on DNA fragmentation, while MX3350-1, CD2325, and MX781 as well as 15dPGJ2 and salicylate elicited a strong induction of apoptosis in both cell lines. As expected, arsenite induced apoptosis in PC3 cells (7-fold) but not in A549 cells (1.6-fold), which correlated with its ability to inhibit IKK and cell proliferation only in the prostate carcinoma cell line. The lower apoptosis induction observed in A549 cells compared to PC3 cells with the RRMs correlates with the higher resistance to the killing activity of the molecules (note that 3 days of treatment were required in A549 cells to observe antiproliferative activity similar to that in PC3 cells, which were treated for only 24 h). Alternatively, a nonpharmacological approach was carried out to investigate the effect of IKK/NF-␬B inhibition in apoptosis signaling. NF-␬B activity was inhibited specifically by overexpression of a dominant negative mutant of IKK (IKK␤ K44M) or a nonphosphorylatable form of I␬B␣(I␬B␣S32/ 36A). Analysis of apoptosis in ␤-galactosidase-positive transfected cells indicated that transient expression of either of these mutants elicited a strong induction of cell death in both PC3 and A549 cells (Fig. 5C), as evidenced by changes in cell morphology (smaller and rounder cells) together with a purple coloration that appeared as a consequence of the double staining with Red-Gal (transfected cells) and trypan blue (nonviable cells). Induction of apoptosis by selective RRMs is dependent on caspase activity. We have demonstrated that the RAR␥-selective RRMs induced apoptosis in Jurkat cells in a caspasedependent manner (54, 56). We further characterized the induction of apoptosis by the novel RRMs MX3350-1 and CD2325 as well as the antagonist MX781 in PC3 and A549 cells. The effect of different concentrations of RRM on the induction of caspase activity was first investigated in PC3 cells. The induction of DEVDase activity was significantly enhanced in the presence of increasing amounts of the RRMs (Fig. 6A). As expected, incubation with RA had no effect on caspase activity. The generation of DEVDase activity was completely abolished when cells were incubated in the presence of 20 ␮M Z-VAD-FMK, a specific paninhibitor of caspases. A concentration of at least 4 ␮M of the antagonist MX781 was necessary to induce DEVDase activity, which correlated with the concentration required to inhibiting IKK (see Fig. 2B and data not shown) and suggested a link between IKK inhibition and induction of apoptosis by this compound. In contrast, lower concentrations of MX3350-1 and CD2325 (0.6 ␮M) were sufficient to induce substantial DEVDase activity, which did not compare with the concentration of RRM necessary to inhibit IKK (not shown). The stimulation of DEVDase activity by MX3350-1, CD2325, and MX781 could be further associated with a significant increase in the amount of cytosolic histone-associated DNA fragments (Fig. 6B) and DNA fragmentation, as determined by TUNEL labeling (Fig. 6C), which were completely prevented in the presence of the pancaspase inhibitor Z-VADFMK. Moreover, active caspases 3 and 9 were detected in protein extracts prepared from PC3 cells treated with the RRMs in the absence of Z-VAD-FMK, but not in extracts obtained from RA-treated cells (Fig. 6D). Incubation with Z-VAD-FMK abolished the cleavage of caspase 9 and the generation of the mature caspase 3 p17 subunit, but not the accumulation of the intermediary fragment of caspase 3, p19. The induction of apoptosis in A549 cells was analyzed next. As expected, the selective RRMs but not RA induced a significant increase of cytosolic DEVDase-like activity, which was completely abolished by Z-VAD-FMK (Fig. 7A). As observed in PC3 cells, MX781, MX3350-1, and CD2325 induced activation of caspases 3 and 9 (Fig. 7B). Although the amount of active caspase 9 was similar in cells treated with the three RRMs, the amount of mature caspase 3 (p17 fragment) was substantially lower in MX781-treated cells, which correlated with reduced DEVDase activity and DNA fragmentation (Fig. 7A and data not shown). This could reflect a delayed activation of the apoptotic program by the retinoid antagonist. In support of this observation, we found that prolonged periods of time were necessary for the antagonist to induce levels of apoptosis similar to those obtained with MX3350-1 in A549 cells, as determined by TUNEL (Fig. 7C and data not shown). MX3350-1 and CD2325 induced apoptosis in 24.5% ⫾1% and 26% ⫾10% of cells, respectively, after 48 h of incubation. In the presence of Z-VAD-FMK, cell death was reduced to 10.4% and 2.5%, respectively, while only 0.3% of the control cells were stained positive. However, only 18% apoptotic cells were observed after 72 h of exposure to 6 ␮M MX781 (not shown). Inhibition of IKK and induction of apoptosis are not mediated by retinoid receptors. Retinoids bind and activate the nuclear retinoid receptors (RARs and retinoid X receptors) to produce a biological response. Although retinoid-mediated cell differentiation is dependent on RARs, the role of the receptors in the induction of apoptosis by the RAR␥-selective molecules is controversial. The induction of apoptosis by CD437 has been suggested to proceed independently of the RARs (28, 61, 66), although RARs could modulate CD437-induced apoptosis in ovarian carcinoma cells (27). Therefore, we decided to examine whether the RARs are involved in the apoptosis induced by the RRMS investigated here. PC3 cells were incubated with CD2325 or MX781 in the absence or presence of a threefold molar excess of RA or the retinoid antagonist CD2366, which showed no anti-IKK or antiproliferative activity (see Fig. 5A and B). The induction of DEVDase activity by the apoptotic RRMs was not affected by saturation of the RARs with the natural ligand RA or the RAR antagonist CD2366 (Fig. 8A), suggesting that RRM1068 BAYON ET AL. MOL.CELL.BIOL. Downloaded from https://journals.asm.org/journal/mcb on 18 July 2021 by 157.88.19.157. FIG. 6. Selective RRMs induce apoptosis in PC3 cells. (A) Induction of DEVDase activity is dependent on RRM concentration. PC3 cells were incubated with the following concentrations of RRMs: 0.3, 0.6, 2, or 6 ␮M MX3350-1 or CD2325; 1, 2, 3, 4, 5, or 6 ␮M MX781; or 6 ␮M RA. After 16 h, cytosol extracts were prepared and assayed for DEVDase activity. Control cells were incubated in the presence of an equivalent volume of solvent. Cells incubated with the different retinoids at 6 ␮M (black columns) were also incubated with 20 ␮M Z-VAD-FMK (white columns) for 1 h prior to retinoid treatment. The fold induction of two independent experiments performed in triplicate is shown. (B) Induction of DNA fragmentation by RRMs. PC3 cells were incubated (white columns) or not (black columns) with Z-VAD-FMK, and subsequently the indicated RRMs were added at 6 ␮M. At 16 h (MX3350-1 and CD2325) or 24 h (RA and MX781) after retinoid treatment, cell extracts were prepared and assayed for DNA fragmentation (expressed as fold induction) with a cell death detection ELISA. (C) A total of 10 6 cells were treated with the indicated RRMs at 6 ␮M in the absence (thick line) or in the presence (thin line) of 20 ␮M Z-VAD-FMK. Control cells (dotted line) were left untreated. At 36 h after RRM exposure, cells were harvested, labeled with a fluorescein TUNEL kit, and analyzed by flow cytometry. The percentage of apoptotic cells is indicated within each box. The experiment was performed at least five times with similar outcomes, and a representative result is shown. (D) Activation of caspases 3 and 9 by RRMs is inhibited in the presence of Z-VAD-FMK. PC3 cells were incubated or not with 20 ␮M Z-VAD-FMK and then treated for 16 h with the retinoids as indicated at 6 ␮M. Cytosol extracts were prepared and analyzed by immunoblot with antibodies recognizing active caspases 3 and 9. VOL. 23, 2003 APOPTOTIC RETINOID-RELATED MOLECULES INHIBIT IKK 1069 Downloaded from https://journals.asm.org/journal/mcb on 18 July 2021 by 157.88.19.157.