The Cyclin-Dependent Kinase 2 Inhibitor Down-regulates Interleukin-1β-Mediated Induction of Cyclooxygenase-2 Expression in Human Lung Carcinoma Cells Free

The cell cycle is an exquisitely controlled and ordered sequence of events, culminating in cell growth and division. Perturbation of cell cycle regulation is a key factor in most human neoplasias (1). The regulatory proteins that play key roles in controlling cell cycle progression are the cyclins, cyclin-dependent kinases (CDK), their substrate proteins, the CDK inhibitors, and the tumor suppressor gene products, p53 and pRb. Several CDK inhibitors, such as flavopiridol, UCN-01, CYC202, and BMS-387032, are undergoing clinical evaluation and showing promising results in early clinical trials (2, 3). Flavopiridol inhibits multiple CDK activities, including CDK2, CDK4, CDK6, and CDK7. A new class of CDK inhibitor N-acyl-2-aminothiazoles with nonaromatic acyl side chains (BMS-387032) is reported to be very selective and acts as a competitive small-molecule inhibitor of CDK2/cyclin E complex by interacting with the ATP-binding site of CDK2, blocking cell cycle progression and tumor cell proliferation and inducing tumor cell apoptosis (4). The IC₅₀ of BMS-387032 toward CDK2/cyclin E is 48 nmol/L, whereas the IC₅₀ of BMS-387032 toward CDK4/cyclin D and CDK1/cyclin B is 925 and 480 nmol/L, respectively (4). In A2780 ovarian carcinoma cell line, BMS-387032 inhibits CDK2 phosphorylation and is shown to inhibit the phosphorylation of the downstream targets of CDK2, including pRb, histone H1, and DNA polymerase α (1, 4).

Prostaglandins, the primary metabolites of cyclooxygenase-2 (COX-2) catalyzed oxygenation of arachidonic acid, are a diverse group of autocrine and paracrine hormones. They mediate many cellular and physiologic processes, such as cell proliferation, inflammatory and immune responses, bone development, wound healing, hemostasis, reproductive function, glomerular filtration, and production of extracellular matrix proteins (5–9). There is also a strong positive correlation between overexpression of COX-2 and tumorigenesis. COX-2 is overexpressed in a variety of different tumors, including colon, lung, pancreatic, prostate, and head and neck cancers, and correlates with poor prognosis and shortened survival (10–13). This increased expression of COX-2 and production of prostaglandins seem to provide a survival advantage to transformed cells through the inhibition of apoptosis, increased attachment to extracellular matrix, increased invasiveness, and stimulation of angiogenesis. Given the multifarious roles of COX-2 in normal physiology, it is critically important to identify agents that inhibit COX-2 expression in tumor cells without affecting COX-2 expression and activity in normal cells. Because actively proliferating tumor cells have elevated CDK activities and several tumors also express COX-2, we investigated whether cell cycle regulatory kinases may have any role in the induction of COX-2 gene expression.

Chronic inflammation is linked to the development of cancer in several organs (14). Interleukin-1β (IL-1β), an inflammatory cytokine, induces COX-2 gene expression in several cancer cell lines through mechanisms that involve p38 and p44/p42 mitogen-activated protein kinase (MAPK) signaling pathways (15, 16). Induction of COX-2 expression by IL-1β operates at the transcriptional level involving transcription factors, such as nuclear factor-κB (NF-κB) and CCAAT/enhancer-binding protein (C/EBP), as well as at the level of mRNA stabilization in a extracellular signal-regulated kinase 1/2 (ERK1/2)–dependent process (17, 18). Interestingly, IL-1β also enhances the activity of the CDK2 (19). We therefore investigated whether IL-1β-induced COX-2 expression involves CDK2.

Here, we report for the first time that a potent CDK2 inhibitor, BMS-387032, down-regulates IL-1β-induced expression of COX-2 in non–small cell lung carcinoma (NSCLC) cell lines (H358 and A549) without inhibiting the basal COX-2 level. We show that IL-1β induces CDK2 activity in H358 cells and that IL-1β-induced CDK2 activity may be critical for induction of COX-2 through a pathway that involves the C/EBP group of transcription factors. Our findings suggest a nexus between cell cycle regulatory events and COX-2 induction that could potentially represent an attractive target for tumor-specific blockage of COX-2.

Materials. CDK inhibitor BMS-387032 (molecular weight 416.99) was obtained from Bristol-Myers Squibb (New York, NY) and dissolved in double-distilled water (10 mmol/L) and the stock solution was stored at −20°C. Penicillin, streptomycin, RPMI 1640, fetal bovine serum (FBS), l-glutamine, LipofectAMINE, and Oligofectamine were obtained from Invitrogen (Grand Island, NY). COX-2 monoclonal and COX-1 polyclonal antibodies were purchased from Cayman Chemical (Ann Arbor, MI). Anti-actin antibodies were purchased from Sigma-Aldrich Co. (St. Louis, MO). CDK2 monoclonal antibodies were purchased from Cell Signaling Biotechnology (Beverly, MA). Antibodies against C/EBP subunits were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). RNase protection assay kit II (RPA II) was obtained from Ambion (Austin, TX). Oligodeoxynucleotides for electrophoretic mobility shift assay (EMSA) were obtained from Invitrogen. CDK2 small interfering RNA (siRNA) was obtained from Dharmacon (Lafayette, CO). [α-³²P]UTP and [γ-³²P]ATP were purchased from ICN Biochemicals (Irvine, CA).

Cell lines. H358, A549, and HCC3255 NSCLC cells (kindly provided by Dr. John D. Minna, University of Texas Southwestern Medical Center, Dallas, TX) were maintained in RPMI 1640 supplemented with 5% FBS, 50 units/mL penicillin, 50 μg/mL streptomycin, 2 mmol/L l-glutamine, and 25 mmol/L HEPES. For all experiments, cells were grown to 75% confluency and then treated as indicated.

Immunoblot analysis. Cells were preincubated with the specific kinase inhibitors for 1 hour and then exposed to IL-1β for 8 hours. Cells were harvested, lysed, and subjected to SDS-PAGE as mentioned earlier (20). Membranes were incubated overnight with COX-1, COX-2, and CDK2 antibodies. The same membrane was probed with anti-β-actin antibodies and provided as loading control. Membranes were developed by the enhanced chemiluminescence system (Amersham Pharmacia Biotech, Piscataway, NJ) and exposed to Hyperfilm (Amersham Pharmacia Biotech). In some experiments, the effect of specific kinase inhibitors, such as PD98059 [MAPK/ERK kinase (MEK) inhibitor], SB203580 (p38 MAPK inhibitor), and LY294002 [phosphatidylinositol 3-kinase (PI3K inhibitor)], in comparison with BMS-387032 was quantitatively assessed by densitometric scanning of immunoblots. Blots were digitally scanned using Adobe Photoshop version 7.0 and intensity of bands corresponding to COX-2 was measured in the green channel from inverted images. Background values were subtracted from values obtained from the COX-2 band. Values from samples with increasing concentration of the inhibitor were converted to percent COX-2 expression relative to the value obtained from the COX-2 band from samples with no inhibitor. Percent COX-2 expression was plotted as a function of log inhibitor concentration (mol/L) and EC₅₀s were calculated from sigmoid dose curves using GraphPad Prism version 4 software.

Immunoprecipitation and immune complex kinase assays. H358 cells were preincubated with BMS-387032 for 1 hour followed by the addition of IL-1β for 2 hours. Cell lysates were prepared with ice-cold lysis buffer [50 mmol/L HEPES (pH 7.4), 150 mmol/L NaCl, 10% glycerol, 1% Triton X-100, 1 mmol/L EDTA, 2.5 mmol/L EGTA, 1 mmol/L DTT, 25 mmol/L β-glycerophosphate, 1 mmol/L NaF, 1 mmol/L Na₃VO₄, 10 μg/mL protease inhibitor mixture] and sonicated at 4°C and lysates were clarified by centrifugation at 10,000 × g for 5 minutes. Whole-cell lysates (500 μg) were preincubated with 1.0 μg anti-CDK2 antibodies (Upstate Biotechnology, Lake Placid, NY) for 2 hours and the immune complex was precipitated with protein A-Sepharose beads at 4°C overnight. The immunoprecipitated proteins on the beads were washed thrice with 1 mL lysis buffer and twice with a kinase buffer (containing 20 mmol/L HEPES, 2 mmol/L MgCl₂, 2 mmol/L MnCl₂, 10 mmol/L β-glycerophosphate, 10 mmol/L NaF, 0.3 mmol/L sodium orthovanadate, 1 mmol/L DTT). The final pellet was resuspended in 25 μL kinase buffer containing 5 μg histone H1 (Sigma, St. Louis, MO), 20 μmol/L ATP, and 5 μCi [γ-³²P]ATP (4,500 μCi/mmol) and incubated for 30 minutes at 30°C with occasional mixing. The reaction was terminated by the addition of 25 μL of 2× concentrated Laemmli sample buffer and separated on a 12.5% SDS-polyacrylamide gel. The gel was dried under vacuum and exposed to Kodak BioMax MS film (Kodak, Rochester, NY) at −80°C.

Prostaglandin E₂ assay. Prostaglandin E₂ (PGE₂) was measured using the PGE₂ EIA kit (Cayman Chemical). A total of 1 × 10⁶ cells were plated in triplicates for each experiment and serum starved for 24 hours. The serum-free cells were treated as indicated with IL-1β (1.0 ng/mL) and BMS-387032 (300 nmol/L). After 8 hours, medium was collected and subjected to PGE₂ assay in a 96-well plate. The product of this enzymatic reaction has a distinct yellow color that absorbs strongly at 412 nm. The intensity of the color is inversely proportional to the amount of free PGE₂ present in the well during the incubation.

RNA analysis. H358 cells were treated as indicated and total RNAs were isolated by using TRIzol (Life Technologies, Grand Island, NY). RPA was done using the RPA II kit. COX-2 open reading frame was subcloned into a pCMV-SPORT6 obtained from Open Biosystem (Huntsville, AL) and linearize with Tth111I enzyme (65°C, overnight). Radiolabeled riboprobes were synthesized from 1 μg of the linearized plasmids using T7 enzyme mix MAXIscript kit (Ambion) for 1 hour at 37°C in a total volume of 20 μL. The reaction buffer contained 10 mmol/L DTT, 0.5 mmol/L ATP, CTP, and GTP, 2.5 mmol/L UTP, and 5 μL of 800 Ci/mmol [α-³²P]UTP at 10 mCi/mL. At the same time, actin riboprobe was also synthesized for loading control. Free nucleotides were removed using the NucAway spin columns (Ambion). Total RNA (10 μg) from H358 cells were incubated at 45°C for 12 hours in hybridization buffer with 5 × 10⁴ cpm COX-2-labeled and 1 × 10⁴ cpm actin-labeled riboprobes. After hybridization, RNase digestion was carried out at 37°C for 30 minutes. The protected fragments were precipitated and then separated on a 5% denaturing polyacrylamide gel at 200 V for 4 hours. The gel was dried under vacuum and exposed to Kodak BioMax MS film overnight at −80°C with intensifying screens.

COX-2 promoter activity. H358 cells were transiently transfected with COX-2 promoter-luciferase reporter plasmids (1.0 μg) along with pRSV-β-gal (0.25 μg) using LipofectAMINE. An equal amount of total DNA was used in transfection for each experiment. Sixteen hours after transfection, cells were serum starved for 24 hours and treated with or without IL-1β (1 ng/mL) in the presence or absence of BMS-387032 for further 8 hours. Luciferase activity was measured by Promega (Madison, WI) luciferase assay system using Monolight 2010 luminometer (Analytical Luminescence Laboratory, San Diego, CA). Luciferase activity was normalized to β-galactosidase activity for differences in the transfection efficiency.

Oligonucleotides. Oligonucleotides containing different COX-2 promoter sites were synthesized by Invitrogen as follows: NF-IL6, 5′-CCCACCGGGCTTACGCAATT-3′ (sense) and 5′-AATTGCGTAAGCCCGGTGGG-3′ (antisense); mutant NF-IL6, 5′-CCCACCGGGCTTACgcttTT-3′ (sense) and 5′-AAaagcGTAAGCCCGGTGGG-3′ (antisense); cyclic AMP–responsive element (CRE), 5′-AAACAGTCATTTCGTCACATGGGCTTG-3′ (sense) and 5′-CAAGCCCATGTGACGAAATGACTGTTT-3′ (antisense); mutant CRE, 5′-AAACAGTCActgattcaCATGGGCTTG-3′ (sense) and 5′-CAAGCCCATGtgaatcagTGACTGTTT-3′ (antisense); NF-κB, 5′-AGGAGAGTGGGGACTACCCCCTCTG-3′ (sense) and 5′-AGAGGGGGTAGTCCCCACTCTCCT-3′ (antisense); and mutant NF-κB, 5′-AGGAGAGTGggtgtgtatcCCTCTG-3′ (sense) and 5′-AGAGGgatacacaccCACTCTCCT-3′ (antisense). The complementary oligonucleotides were annealed and purified following the manufacturer's protocol.

Electrophoretic mobility shift assays. Nuclear extracts were prepared as described previously (21). Probes (double-stranded oligonucleotides) containing the NF-IL6, NF-κB, or CRE-binding site were purchased from Invitrogen and radiolabeled using T4 polynucleotide kinase and [γ-³²P]ATP (22). A typical binding reaction involved a 15-minute preincubation with 5 μg nuclear extract, 1.5 μg poly(deoxyinosinic-deoxycytidylic acid), 200 ng single-stranded oligonucleotide, 20 mmol/L HEPES-NaOH (pH 7.6), 100 mmol/L NaCl, 1 mmol/L DTT, and 2% glycerol followed by a 20-minute incubation with 20 fmol radiolabeled probe. To assess the specificity of DNA protein binding, radiolabeled mutant oligonucleotide was added instead of wild-type radiolabeled oligonucleotide to the binding reaction mixture. In the supershift analysis, C/EBPβ-specific antibodies (1 μg) were added to the mixture and incubated for an additional 10 minutes at room temperature before electrophoresis. Complexes were then resolved by electrophoresis for 90 minutes at 200 V on a 6% native polyacrylamide gel, dried, and processed for autoradiography.

siRNA transfection. H358 cells (1 × 10⁵ per well) were transiently transfected with different concentrations of CDK2 siRNA in six-well plates using Oligofectamine. Cells were rinsed 5 hours after transfection with 1× PBS, and complete medium was added and further incubated for 40 hours. Cells were serum starved for 24 hours and treated with IL-1β for another 8 hours in the serum-free medium. Cells were lysed and subjected to immunoblot analysis for CDK2, COX-2, and COX-1 expression.

Determination of cell death. Cells were plated, serum starved for 24 hours, and treated with specific kinase inhibitors at increasing concentrations for 1 hour before the addition of IL-1β. Cells were further incubated for 8 hours. Cells were trypsinized (keeping all the floating cells) and the cell suspension was diluted 1:1 with 0.4% trypan blue and counted using a hemacytometer. Cells scoring positive for uptake of the dye was considered dead and the % cell death was calculated using the following formula: [(number of trypan blue positive cells) / (number of trypan blue–positive cells + number of trypan blue negative cells)] × 100.

CDK2 inhibitor BMS-387032 inhibits IL-1β-mediated induction of COX-2 expression. Proinflammatory cytokines are reported to induce COX-2 expression in a variety of cell types (23–25). We tested the ability of several cytokines to induce COX-2 expression in the H358 NSCLC cells. These cells were serum starved for 24 hours and incubated with IL-1β, IL-6, IL-8, IFN-γ, and tumor necrosis factor-α (TNF-α). Results in Fig. 1A show that only the addition of IL-1β significantly increased COX-2 expression (Fig. 1A,, lane 2) after 8 hours. Enhancement of the activity of CDK2 by IL-1β has been reported before (19). We therefore explored the possibility that IL-1β-induced COX-2 expression may involve CDK2. In Fig. 1B, H358 cells were pretreated with increasing concentrations of the CDK2 inhibitor, BMS-387032, for 1 hour and then treated with IL-1β (1 ng/mL) for further 8 hours. We observed that IL-1β strongly induced COX-2 expression (Fig. 1B,, lane 2) and BMS-387032 produced a dose-dependent decrease in IL-1β-induced COX-2 expression (Fig. 1B,, lanes 3-7). Significant inhibition started at 150 nmol/L BMS-387032 where >85% COX-2 expression was blocked (Fig. 1B,, lane 5), whereas complete inhibition was achieved at 300 nmol/L (Fig. 1B,, lane 7). BMS-387032 also blocked IL-1β-induced COX-2 expression in a second NSCLC cell line, A549 (Fig. 1C,, lanes 3-7). However, CDK2 inhibition in both cell lines had no effect on COX-1 expression (Fig. 1B  and C, middle). Data from these experiments clearly indicate that BMS-387032 inhibits IL-1β-induced COX-2 expression in these lung carcinoma cells in a dose-dependent manner, whereas endogenous COX-1 expression is not affected by either IL-1β or the combination of IL-1β and BMS-387032.

BMS-387032 inhibits IL-1β-induced CDK2 activity. To show that IL-1β in fact induces CDK2 activity in H358 cells and BMS-387032 blocks this activity, we did in vitro kinase assays. CDK2 was immunoprecipitated from cells treated with IL-1β in the presence or absence of BMS-387032. The immune complex was then analyzed for CDK2 activity using histone H1 as the substrate. We observe that the treatment with IL-1β strongly induced CDK2 activity at 2 hours (Fig. 1D,, lane 2), which is evidenced by the phosphorylation of its downstream target histone H1. Pretreatment with BMS-387032 significantly inhibited IL-1β-induced histone H1 phosphorylation (Fig. 1D , lane 3). Immunoprecipitation of equal amounts of CDK2 protein in each kinase reaction was verified by Western blotting with anti-CDK2 antibody.

CDK2 siRNA abrogates IL-1β-induced COX-2 expression. Our data indicate that BMS-387032 blocks the IL-1β-mediated induction of COX-2 in H358 cells by inhibiting CDK2. Next, we tested the possibility that ablation of the CDK2 protein has a similar effect on IL-1β-induced COX-2 expression. Data show that CDK2 expression is abrogated by the transfection of CDK2 siRNA at increasing concentrations (Fig. 2A). Maximum inhibition of CDK2 expression was observed at 100 nmol/L (Fig. 2A,, lane 5), whereas levels of CDK2 remained unchanged when the cells were transfected with a similar concentration of acyl protein thioesterase 1 (APT1) siRNA (Fig. 2A,, lane 2) and serve as a negative control. Silencing of CDK2 expression resulted in a striking inhibition of IL-1β-induced COX-2 expression (Fig. 2B). We also noticed that transfection with CDK2 siRNA had no effect on COX-1 expression (Fig. 2B). These data correlate well with the effect of the small-molecule inhibitor BMS-387032 and clearly show that the activity as well as expression of the CDK2 is critical for IL-1β-induced expression of COX-2.

BMS-387032 inhibits IL-1β-mediated induction of COX-2 mRNA levels. Our data show a critical requirement of CDK2 during IL-1β-mediated induction of COX-2 expression. We next analyzed whether CDK2 inhibition by BMS-387032 had any effect on steady-state mRNA levels of COX-2. Serum-starved cells are treated with BMS-387032 or IL-1β plus BMS-387032 for 4 hours. Total RNA was isolated and subjected to RPA as described in Materials and Methods. The results show that a substantial induction of COX-2 mRNA is observed in response to IL-1β (Fig. 3A,, lane 2). Addition of BMS-387032 alone had no effect on COX-2 mRNA (Fig. 3A,, lane 4). However, BMS-387032 at 300 nmol/L resulted in a significant (>95% inhibition) inhibition of IL-1β-mediated induction of COX-2 mRNA (Fig. 3A,, lane 3). The RPA result is consistent with our observations indicating that CDK2 inhibition blocks IL-1β-induced levels of the COX-2 enzyme (Fig. 1B and C).

BMS-387032 inhibits IL-1β-mediated induction of PGE₂ synthesis. To determine whether CDK2 inhibition blocks PGE₂ production through the inhibition of COX-2 gene expression, we measured IL-1β-induced PGE₂ release in the presence or absence of the BMS-387032. Results presented in Fig. 3B indicate a significant enhancement of PGE₂ release from 2 to 85 pg/mL (42.5-fold) in 8 hours in the presence of IL-1β. Addition of BMS-387032 alone had no observable effect on the basal level of PGE₂ (4.7 pg/mL). However, >90% of the IL-1β-induced PGE₂ synthesis was inhibited (15 pg/mL) by the addition of BMS-387032. This result is consistent with the observation that CDK2 inhibition leads to a blockade of IL-1β-induced COX-2 mRNA and protein.

BMS-387032 inhibits IL-1β-mediated induction of COX-2 promoter activity. COX-2 expression is tightly regulated at multiple levels, including transcription and mRNA stabilization. The COX-2 gene is transcriptionally regulated by a 1,432-bp minimal promoter and is responsive to IL-1β stimulation (26). The COX-2 promoter harbors sequences for the specific binding of transcription factors, such as NF-κB, NF-IL6-C/EBP, PEA3, CRE, and activator protein-1. We asked whether transcriptional activation of the COX-2 promoter by IL-1β can be inhibited by this CDK2 inhibitor. To address this question, we examined the effect of BMS-387032 on the transcription of a luciferase reporter construct driven by the −1432/+59 and −327/+59 minimal COX-2 promoter containing NF-κB, NF-IL6/C/EBP, and CRE sites. The data indicate that treatment with IL-1β produces a 4.4-fold (−1432/+59) and 4.8-fold (−327/+59) increase in COX-2 promoter activity (Fig. 3C). Pretreatment with BMS-387032 resulted in a significant inhibition of IL-1β-induced COX-2 promoter activity in both cases. The results clearly show that the inhibitory effect of BMS-387032 manifests at the level of transcription and accounts for inhibition of IL-1β-induced COX-2 protein expression and PGE₂ levels.

Effect of BMS-387032 on transcription factors that regulate COX-2 transcription in response to IL-1β. To further elucidate the effect of CDK2 inhibitor in the IL-1β-stimulated binding to COX-2 regulatory sequences, we did EMSAs with radiolabeled oligonucleotides containing the binding sites for NF-κB, CRE, or NF-IL6 elements present in the COX-2 promoter. The data indicate that IL-1β increased binding of nuclear proteins to the NF-IL6 (Fig. 4A,, lane 3) and NF-κB (Fig. 4B,, lane 3) sites of COX-2 promoter. However, no significant increase in binding was observed with the CRE site (Fig. 4C,, lane 3). The binding to all the three key cis-acting elements in the COX-2 promoter was specific because oligonucleotides containing mutations in any of the sequences failed to bind nuclear proteins (Fig. 4A,-C, lane 5). Interestingly, preincubation with BMS-387032 followed by IL-1β treatment showed significantly reduced binding (>90%) to the NF-IL6 site (Fig. 4A,, lane 4). A <15% inhibition of binding to the NF-κB site is also observed in the presence of BMS-387032 (Fig. 4B,, lane 4). On the other hand, binding to the CRE sites (Fig. 4C,, lane 4) of the COX-2 promoter remained unaffected. Moreover, supershift analysis with antibodies against C/EBPβ indicated the presence of C/EBPβ in complex with the NF-IL-6 element (Fig. 4A , lane 7). We also did the supershift analysis using the other subunits of C/EBP (α, γ, and δ); however, we did not observe any significant supershifting in the NF-IL6-binding complex (data not shown). These findings strongly suggest the nuclear factor C/EBPβ is the downstream target of CDK2 because BMS-387032 prevented the IL-1β-induced binding of C/EBPβ to NF-IL6 sequence of the COX-2 promoter.

BMS-387032 does not inhibit the endogenous expression of COX-2. The NSCLC cell line HCC3255 exhibits high constitutive expression of COX-2. Therefore, we wanted to examine whether COX-2 expression in this cell line can be prevented by BMS-387032. Our results show that BMS-387032 does not down-regulate the basal COX-2 as well as COX-1 expression in HCC3255 cells even after 48 hours of treatment (Fig. 5A,, lane 4, 8). We then investigate the effect of IL-1β on the induction of COX-2 expression in this cell line in the presence of BMS-387032. The data in Fig. 5B show that neither IL-1β nor BMS-387032 has any effect on COX-2 expression in HCC3255 cells (Fig. 5A and B).

Effect of BMS-387032 on signaling pathways regulating COX-2 expression. IL-1β induces the activity of several key signaling molecules, such as p38 MAPK, phosphorylated stress-activated protein kinase (SAPK)/c-Jun NH₂-terminal kinase (JNK), phosphorylated AKT, and phosphorylated ERK1/2. Moreover, these kinases are also involved in the regulation of COX-2 expression in carcinoma cells (27, 28). We examined whether p38 MAPK, phosphorylated SAPK/JNK, phosphorylated AKT, or phosphorylated ERK1/2 might lie downstream to CDK2 in the IL-1β-induced signaling pathway leading to COX-2 expression. Our results indicate that, in all cases, BMS-387032 fails to prevent the IL-1β-induced phosphorylation of p38 MAPK, phosphorylated SAPK/JNK, phosphorylated AKT, and phosphorylated ERK1/2 (Fig. 6A). We then examined the relative contribution of the different signaling pathways that lead to IL-1β-induced COX-2 expression. Toward this end, we compared the effects of BMS-387032 with the effects of specific kinase inhibitors, such as PD98059, SB203580, and LY294002, which are also known to inhibit COX-2 expression under induced conditions (27, 28). H358 cells are treated with these inhibitors at various concentrations (5 nmol/L to 50 μmol/L) for 1 hour before the addition of IL-1β. Although p38 MAPK (SB203580), PI3K (LY290004), and MEK (PD98059) inhibitors significantly inhibited IL-1β-induced COX-2 expression in H358 cells (Fig. 6B), the effective concentrations of these inhibitors required to exert this effect were much higher when compared with a similar concentration range of BMS-387032 (Fig. 6C). We measured the approximate EC₅₀s for all the inhibitors used in this study and found 14.5, 10.7, and 2.5 μmol/L for SB203580, LY290004, and PD98059, respectively, whereas the EC₅₀ of BMS-387032 is 0.18 μmol/L. Therefore, data in Fig. 6C show that among all the inhibitors tested BMS-387032 is the most effective in preventing COX-2 induction by IL-1β in these lung cancer cell lines. We also measured the % cell death in response to the treatment with each of these inhibitors under similar experimental conditions. We observed no appreciable cell death (7-10%) after 8 hours of treatment with these inhibitors (Fig. 6C).

Dysregulation of the cell cycle is a critical event in the development of cancer. CDK2 is a key modulator of cell cycle and cell proliferation. The role of COX-2 in the modulation of cell cycle progression and its overexpression in several tumor models has also been documented (29, 30). In this study, we have examined the effect of a potent inhibitor of CDK2 on cytokine-induced COX-2 expression, and our findings showing that CDK2 expression and activity is critical in IL-1β-induced COX-2 expression in the NSCLC cell lines gain particular significance.

COX-2 expression is predominantly controlled at two levels: activation of transcription and stabilization of COX-2 mRNA levels. We showed that IL-1β-induced binding of C/EBPβ subunit to NF-IL6 sites of the COX-2 promoter is inhibited by BMS-387032. C/EBP family of proteins has emerged as a key transactivator for COX-2 expression induced by proinflammatory mediators. Several reports indicated that β and δ isoforms of C/EBP are involved in COX-2 transcriptional activation by proinflammatory mediators in human cells (17). Saunders et al. reported that in human foreskin fibroblast cells C/EBPβ activation is required for the IL-1β-induced COX-2 expression and is inhibited by the addition of either aspirin or salicylate (17). The mechanism by which these drugs inhibit the binding of C/EBPβ to the COX-2 promoter site is still unknown (31).

COX-2 mRNA contains an AU-rich element that regulates COX-2 expression at the post-transcriptional level. In this study, we did not observe any effect of either IL-1β or BMS-387032 on the expression of a cytomegalovirus promoter-driven luciferase reporter under the control of the COX-2 3′ untranslated region (data not shown). It also reported that activation of p38 MAPK signaling cascades are important for the increase in COX-2 stability (26, 32). In this study, we noticed that BMS-387032 did not prevent p38 MAPK activation by IL-1β (Fig. 6A) and also observed that EC₅₀ of SB203580, a specific inhibitor of p38 MAPK, was 80-fold higher (14.5 μmol/L) when compared with the EC₅₀ of BMS-387032 (0.181 μmol/L), indicating that IL-1β-mediated COX-2 expression is more sensitive to CDK2 inhibition in this H358 lung carcinoma cells (Fig. 6B and C).

There is considerable evidence indicating that C/EBP is a downstream target of CDK2. Gorgoni et al. reported that COX-2 mRNA induction and promoter activity was blocked in C/EBPβ^−/− macrophages but was rescued when C/EBPβ was overexpressed (33). The transcriptional activity of C/EBPβ is regulated by phosphorylation (34) and several phosphorylation sites in C/EBPβ are required for its activity (34, 35). In a recent report, Shuman et al. reported that in mouse embryo fibroblast cells (NIH3T3) Ser⁶⁴ and Thr¹⁸⁹ residues within the transactivation domain of C/EBPβ were phosphorylated by CDK2 and phosphorylation on both sites was blocked by CDK inhibitor roscovitin (35). It was also reported that Ser⁶⁴ residue of C/EBPβ is specifically phosphorylated by CDK2 not by either CDK4 or CDK6 (35). However, the precise role of CDK2 in the IL-β-mediated activation of human C/EBPβ is currently unknown and this mechanism remains to be elucidated.

CDK2 was originally thought to be critically required for G₁-S transition along with cyclin E. However, recent findings indicated that blocking of CDK2 expression in human tumor cell lines had little or no effect on their ability to proceed through the cell cycle and CDK2 knockout mice are viable and that their cells are able to proliferate in culture in vitro (36). Previous studies reported that the phosphorylation of Ser⁶⁴ and Thr¹⁸⁹ by CDKs is critical for C/EBPβ to enhance Ras-induced neoplastic transformation in mouse fibroblast cells (35, 37). These observations clearly indicated that the function of CDK2 is not limited to cell cycle regulation and may be indicative of a different role in tumor cell lines. In this study, we reported CDK2 expression is necessary for IL-1β-induced COX-2 expression (Fig. 2).

COX-2 expression in various cellular systems is controlled by multiple signaling pathways. For example, in HCC3255, NSCLC cells that exhibit a high constitutive expression of COX-2 and BMS-387032 did not inhibit the high levels of COX-2 expression in this cell line. Our results indicate that CDK2 activity is critical in IL-1β-induced expression of COX-2 in H358 cells and may not be involved in the high constitutive expression of COX-2 observed in HCC3255 cells. This information suggests the possibility that CDK2 inhibitor might target tumors in which COX-2 expression is cytokine mediated and is less likely to be effective against tumors that exhibit cytokine-independent COX-2 expression. Clinically, anti-COX-2 therapies have been targeted at the level of COX-2 enzyme activity. Recent reports suggest that classic COX-2 inhibitors do not effectively discriminate between tumor COX-2 activity and other physioprotective functions of the COX-2 enzyme. Therefore, our results present an opportunity for tumor control through the selective inhibition of cytokine- and CDK2-driven COX-2 expression in certain subsets of tumors.

Grant support: NIH grants CA90949, CA82117, and CA21661 (H. Choy) and American Cancer Society institutional funding and Flight Attendant Medical Research Institute Young Clinical Scientist Award (D. Saha).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Hiroshi Inoue (Nara Women's University, Nara, Japan) for the COX-2 promoter constructs, Dr. John D. Minna for NSCLC cell lines, and Dr. Chaitanya S. Nirodi (University of Texas Southwestern Medical Center) for advice and discussion.