Cigarette smoke (CS) has been linked to cardiovascular, pulmonary, and malignant diseases. CS-associated malignancies including cancers of the larynx, oral cavity, and pharynx, esophagus, pancreas, kidney, bladder, and lung; all are known to overexpress the nuclear factor-κB (NF-κB)-regulated gene products cyclin D1, cyclooxygenase (COX)-2, and matrix metalloprotease-9. Whether the COX-2 inhibitor, celecoxib, approved for the treatment of colon carcinogenesis and rheumatoid arthritis, affects CS-induced NF-κB activation is not known, although the role of NF-κB in regulation of apoptosis, angiogenesis, carcinogenesis, and inflammation is established. In our study, in which we examined DNA binding of NF-κB in human lung adenocarcinoma H1299 cells, we found that cigarette smoke condensate (CSC)-induced NF-κB activation was persistent up to 24 h, and celecoxib suppressed CSC-induced NF-κB activation. Celecoxib was effective even when administered 12 h after CSC treatment. This effect, however, was not cell type-specific. The activation of inhibitory subunit of NF-κB kinase (IκB), as examined by immunocomplex kinase assay, IκB phosphorylation, and IκB degradation was also inhibited. Celecoxib also abrogated CSC-induced p65 phosphorylation and nuclear translocation and NF-κB-dependent reporter gene expression. CSC-induced NF-κB reporter activity induced by NF-κB inducing kinase and IκB α kinase but not that activated by p65 was also blocked by celecoxib. CSC induced the expression of NF-κB-regulated proteins, COX-2, cyclin D1, and matrix metalloproteinase-9, and celecoxib abolished the induction of all three. The COX-2 promoter that is regulated by NF-κB was activated by CSC, and celecoxib suppressed its activation. Overall, our results suggest that chemopreventive effects of celecoxib may in part be mediated through suppression of NF-κB and NF-κB-regulated gene expression, which may contribute to its ability to suppress inflammation, proliferation, and angiogenesis.
Cigarette smoke (CS) is a major risk factor in the development of many cancers. Smokers have a high incidence of lung cancer, which is the most common cause of cancer in Western countries, accounting for more deaths than those caused by prostrate, breast, and colorectal cancers combined (1) . Recent estimates indicate that CS causes approximately 80–90% of lung cancer in the United States (2) . CS is also implicated in cancers of the larynx, oral cavity, pharynx, esophagus, pancreas, kidney, and bladder. CS is a complex chemical mixture containing thousands of different compounds, of which 100 are known carcinogens, cocarcinogens, mutagens, or tumor promoters (3) . Besides being exposed to these chemicals, active smokers have >25% lower circulating concentrations of ascorbic acid, α-carotene, β-carotene, and cryptoxanthin (4) . CS is also known to induce morphological changes in lungs, placenta, liver, and kidneys (5) . Thus agents that can neutralize the effects of CS are urgently needed.
Celecoxib (1,5-diaryl pyrazole-based compound), a specific cyclooxygenase (COX)-2 inhibitor has been approved recently for the treatment of rheumatoid arthritis and osteoarthritis (6) . This nonsteroidal anti-inflammatory agent has also been shown to reduce the formation of polyposis in familial adenomatous polyposis patients (7) . In vivo, celecoxib has been shown to suppress the growth of cancers of the colon and head and neck (8 , 9) and to enhance the antitumor activity of chemotherapeutic agents (10) . Numerous studies suggest that this drug exhibits chemopreventive activity against colon cancer (11 , 12) , breast cancer (13) , urinary bladder cancer (14) , and skin cancer (15) . Celecoxib has also been shown to suppress angiogenesis (16) , most likely by reducing proliferation and induction of apoptosis in endothelial cells (17) . There is also a report that indicates that celecoxib reduces pulmonary inflammation (18) . In vitro, celecoxib has been shown to induce apoptosis of colon cancer cells (19) , pancreatic cancer cells (20) , and prostate cancer cells (21) .
Because suppression of the nuclear transcription factor nuclear factor-κB (NF-κB) activation has been linked with antitumor, chemopreventive, chemosensitivity, suppression of inflammation, antiangiogenesis, and apoptosis, we postulate that celecoxib must mediate its effects through suppression of NF-κB activation. Indeed, NF-κB has been shown to regulate the expression of a number of genes of which the products are involved in tumorigenesis (22) . These include antiapoptoic genes (e.g., cIAP, suvivin, TRAF, bcl-2, and bcl-xl); COX-2; matrix metalloproteinase (MMP)-9; genes encoding adhesion molecules, chemokines, inflammatory cytokines, and iNOS; and cell cycle regulatory genes (e.g., cyclin D1; Ref. 23 ). In an inactive state, NF-κB is present in the cytoplasm as a heterotrimer consisting of p50, p65, and inhibitory subunit of NF-κB (IκB) α subunits. In response to an activation signal, the IκBα subunit is phosphorylated, ubiquitinated, and degraded through the proteasomal pathway, thus exposing the nuclear localization signals on the p50-p65 heterodimer. The p65 is then phosphorylated, leading to nuclear translocation and binding to a specific sequence in DNA, which in turn results in gene transcription. An IκBα kinase, IKK, has been identified that phosphorylates serine residues in IκBα at positions 32 and 36 (24) .
Because CS can activate NF-κB in a wide variety of different cell types (25) possibly leading to carcinogenesis, we postulated that the chemopreventive effects of celecoxib may be mediated through suppression of NF-κB. In the present report we thus investigated whether celecoxib could block CS-induced NF-κB activation in human non-small cell lung carcinoma cells and whether it abrogates the expression of genes implicated in carcinogenesis. The results to be described show that celecoxib is quite effective in suppressing CS-induced NF-κB activation and NF-κB-regulated gene expression.
MATERIALS AND METHODS
Celecoxib, purchased from LKT Laboratories, Inc. (Minneapolis, MN), was dissolved in DMSO as a 100 mm stock solution and stored at −20°C. Penicillin, streptomycin, RPMI 1640, fetal bovine serum, and lipofectAMINE 2000 were obtained from Invitrogen (Grand Island, NY). The following polyclonal antibodies were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA): anti-p65, anti-p50, anti-IκBα, and anti-cyclin D1. Phospho-IκBα (Ser32) antibody was purchased from Cell Signaling (Beverly, MA). Anti-IKKα and anti-IKKβ antibodies were kindly provided by Imgenex (San Diego, CA). Anti COX-2 antibody was purchased from BD Biosciences (San Diego, CA), and anti-MMP-9 antibody was purchased from Cell Sciences, Inc., (Norwood, MA). The COX-2 promoter (−375 to +59) amplified from human genomic DNA by using the primers 5′-GAGTCTCTTATTTATTTTT-3′ (sense) and 5′-GCTGCTGAGGAGTT CCTGGACGTGC-3′ (antisense) was kindly provided by Dr. Xiao-Chun Xu (M.D. Anderson Cancer Center).
The cell lines used in our studies included immortalized human bronchial epithelial cells (BEAS-2B), human non-small cell lung carcinoma cells (H1299), and human lung epithelial cell carcinoma cells (A549) were all kindly provided by Dr. Reuben Lotan (M. D. Anderson Cancer Center). Human embryonic kidney cells (293) were obtained from the American Type Culture Collection (Manassas, VA). BEAS-2B were maintained in keratinocyte serum-free medium, H1299 cells were cultured in RPMI 1640, and 293 cells were cultured in MEM, all supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 μg/ml streptomycin.
Preparation of CS Condensate (CSC).
The CSC, prepared from the University of Kentucky Reference Cigarette 1R4F (9 mg tar and 0.8 mg nicotine/cigarette; Ref. 25 ), was kindly provided by Dr. C. Gary Gairola (University of Kentucky, Lexington, KY).
To determine NF-κB activation, we carried out EMSA essentially as described previously (26) . Briefly, nuclear extracts prepared from cells (2 × 106/ml) treated with CSC were incubated with 32P-end-labeled 45-mer double-stranded NF-κB oligonucleotide. The DNA-protein complex formed was separated from free oligonucleotide on 6.6% native polyacrylamide gels. The dried gels were visualized, and radioactive bands were quantitated using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) using Imagequant software.
IκBα Phosphorylation and Degradation.
To determine the effect of celecoxib on CSC-dependent IκBα phosphorylation and degradation, cytoplasmic extracts were prepared as described (27) from H1299 (2 × 106/ml) pretreated with celecoxib for 4 h and then exposed to 10 μg/ml CSC for various times. Thirty to 60 μg of cytoplasmic extracts were analyzed by Western blot analysis using antibodies, which recognize IκBα and phosphorylated IκBα.
To determine the effect of celecoxib on CSC-induced IKK activation, we analyzed IKK by a method essentially as described previously (27) . Briefly, IKK complex was precipitated from whole-cell extracts with antibody to IKKα and IKKβ, assayed in kinase assay using glutathione S-transferase-IκBα 1–54(1–54) as a substrate. To determine the total amounts of IKKα and IKKβ in each sample, 30 μg of the whole-cell extract protein was resolved on a 7.5% polyacrylamide gel and analyzed by Western blot analysis using antibodies against IKKα and IKKβ.
NF-κB-Dependent Reporter Gene Transcription.
The effect of celecoxib on CSC-induced NF-κB-dependent reporter gene transcription was measured as described previously (28) . Briefly, H1299 cells (5 × 105 cells/well) were plated in six-well plates and transiently transfected next day with 2 μg of the secretory alkaline phosphatase (SEAP) expression plasmid (pNF-κB-SEAP) by lipofectAMINE 2000 method using the manufacturer’s protocol. To examine the effect of celecoxib on CSC-induced reporter gene expression, we treated the cells with various doses of celecoxib for 4 h and then with CSC. Twenty-four h later, the cell culture medium was harvested and analyzed for SEAP activity.
Western Blot Analysis for COX-2, MMP-9, and Cyclin D1.
To determine the expression of cyclin D1, COX-2, and MMP-9, whole-cell extracts were prepared from treated cells (2 × 106 cells in 2 ml of medium), and 30–50 μg of protein was resolved on 10% SDS-PAGE, electrotransferred, and probed with specific antibodies against cyclin D1 (1:1000), MMP-9 (1:1000), and COX-2 (1:250) as described (29) .
Immunolocalization of NF-κB p65.
The effect of celecoxib on CS-induced nuclear translocation of p65 was examined using the method described previously (30) .
Assay of COX-2 Promoter Activity.
H1299 cells were seeded at a concentration of 1.5 × 105 cells per well in six-well plates. After overnight culture, the cells in each well were transfected with 2 μg DNA consisting of COX-2 promoter-luciferase reporter plasmid, along with 6 μl of lipofectAMINE 2000 by following the manufacturer’s protocol. After 6-h exposure, the cells were incubated in medium containing celecoxib for 4 h. The cells were washed and then exposed to CSC, harvested 36 h later, and luciferase activity was measured by using the Promega luciferase assay system according to the manufacturer’s protocol using Monolight 2010 (Analytical Luminescence Laboratory, San Diego, CA). All of the experiments were performed in triplicate and performed at least thrice to prove their reproducibility.
The aim of the current study was to determine whether celecoxib, a chemopreventive agent, can suppress CSC-induced NF-κB activation and NF-κB-regulated gene expression that mediates carcinogenesis. For most of the experiments, human non-small cell lung adenocarcinoma cells were used. Immortalized human bronchial epithelial cells, which are nonmalignant, were also used for some experiments. The concentration of celecoxib used and the duration of exposure had minimal effect on the viability of these cells as determined by trypan blue dye exclusion test.
Celecoxib Inhibits CSC-Induced DNA Binding of NF-κB.
H1299 cells were preincubated with different concentrations of celecoxib, then treated with CSC, prepared the nuclear extracts, and examined NF-κB activity by EMSA. Results showed that celecoxib, even up to 100 μm, by itself did not activate NF-κB, CSC activated NF-κB by 3-fold, and celecoxib inhibited CSC-mediated NF-κB activation in a dose-dependent manner, with maximum inhibition occurring at 100 μm (Fig. 1A) ⇓ . Celecoxib inhibited CSC-induced NF-κB activation by 50% at 210 min, but complete inhibition occurred at 4 h (Fig. 1B) ⇓ .
When we incubated nuclear extracts from CSC-activated cells with antibodies to the p50 (NF-κB1) and the p65 (RelA) subunit of NF-κB, both antibodies shifted the band to a higher molecular mass, thus suggesting that the CSC-activated complex consisted of both subunits (Fig. 1C) ⇓ . Neither preimmune serum nor irrelevant antibody had any effect. Addition of excess unlabeled NF-κB (cold oligo; 100-fold) caused complete disappearance of the band. EMSA results showed that celecoxib did not modify the DNA-binding ability of NF-κB proteins prepared from cells by treatment with CSC (data not shown).
Whether the effect of CS on NF-κB activation was persistent was investigated. Results in Fig. 2A ⇓ show that CSC-induced NF-κB activation remained persistent up to 24 h. NF-κB activation could also be seen even 72 h after exposure of cells to CS (data not shown). Pretreatment of cells to celecoxib, however, completely abrogated the activation of NF-κB. Whether cotreatment of celecoxib and CSC is also effective in suppressing NF-κB activation was also examined. We observed that celecoxib need to be present for 4 h to inhibit the NF-κB activation and the inhibitory effect persisted until 24 h (Fig. 2B) ⇓ . Whether administration of celecoxib after exposure to CS will also suppress NF-κB activation was also examined. The results demonstrate that celecoxib inhibited CSC-induced NF-κB activation even 12 h after exposure to CSC (Fig. 2C) ⇓ . These results indicate that celecoxib will inhibit the CSC-induced NF-κB activation, whether treated before, simultaneously, or after the CS exposure.
Inhibition of NF-κB Activation by Celecoxib Is Not Cell Type-Specific.
Some reports suggest that distinct signal transduction pathways mediate NF-κB induction in epithelial and lymphoid cells (31) . Whether CSC could activate NF-κB in other cell types and whether celecoxib could inhibit this activation was investigated. We examined the effect of celecoxib on immortalized human bronchial epithelial cells (BEAS-2B), lung epithelial cell carcinoma (A549), T-cell leukemia (Jurkat), and myeloid leukemia (KBM-5) cells. Cells were pretreated with celecoxib, then exposed to CSC and examined for NF-κB. CSC activated NF-κB in all of the cell types, and celecoxib completely inhibited this activation (Fig. 3) ⇓ , indicating a lack of cell type specificity.
Celecoxib Inhibited CSC-Dependent IκBα Degradation.
To determine whether inhibition of CSC-induced NF-κB activation was due to inhibition of IκBα degradation, normally a condition for translocation of NF-κB to the nucleus (32) , we pretreated cells with celecoxib and then exposed them to CSC for different times. CSC activated NF-κB in the control cells in a time-dependent manner but had little effect on celecoxib-pretreated cells (Fig. 4A) ⇓ . CSC induced IκBα degradation in control cells as early as 15 min, but in celecoxib-pretreated cells CSC had no effect on IκBα degradation (Fig. 4B) ⇓ .
Celecoxib Inhibited CSC-Dependent IκBα Phosphorylation.
CSC induced IκBα phosphorylation at 15 min, and celecoxib almost completely suppressed IκBα phosphorylation (Fig. 4C) ⇓ . These results indicate that celecoxib inhibited CSC-induced NF-κB activation through the inhibition of phosphorylation and degradation of IκBα.
Celecoxib Inhibited CSC-Induced IKK Activation.
Because celecoxib inhibits the phosphorylation of IκBα, we tested the effect of celecoxib on CSC-induced IKK activation, which is required for CSC-induced phosphorylation of IκBα (29) . As shown in Fig. 4D ⇓ , in an immunocomplex kinase assay, CSC activated IKK and the activation occurred 15 min after CSC treatment (top panel). Celecoxib treatment completely suppressed this activation. CSC or celecoxib had no direct effect on the expression of either IKKα (middle panel) or IKKβ (bottom panel) proteins. When we immunoprecipitated IKK from CSC-treated cells and then incubated it with different concentrations of celecoxib, we found that celecoxib has no direct effect on IKK activity (data not shown).
Celecoxib Inhibited CSC-Induced Phosphorylation and Nuclear Translocation of p65.
We also tested the effect of celecoxib on CSC-induced phosphorylation of p65, which is also required for transcriptional activity of p65 (24) . As shown in Fig. 5A ⇓ , CSC induced the phosphorylation of p65 in a time-dependent manner, and celecoxib treatment suppressed p65 phosphorylation almost completely. Western blot analysis (Fig. 5B) ⇓ and immunocytochemistry (Fig. 5C) ⇓ indicated indicate that CSC induced the nuclear translocation of p65, and celecoxib treatment abrogated the p65 translocation.
Celecoxib Represses CSC-Induced NF-κB-Dependent Reporter Gene Expression.
Although we showed by EMSA that celecoxib blocked NF-κB activation, DNA binding alone does not always correlate with NF-κB-dependent gene transcription, suggesting that there are additional regulatory steps (33) . CSC induced NF-κB-dependent reporter (SEAP) gene expression, and dominant-negative IκBα transfection abolished it, indicating specificity (Fig. 6A) ⇓ . Treatment of cells with celecoxib inhibited CSC-induced NF-κB reporter activity in a dose-dependent manner (Fig. 6A) ⇓ . These results demonstrate that celecoxib inhibits not only DNA binding but also NF-κB-dependent reporter gene expression induced by CSC.
How CSC activates NF-κB is not fully understood, but the results indicated above suggest that IKK is required for NF-κB activation. Because IKK is activated by numerous kinases including NF-κB-inducing kinase (NIK; Ref. 34 ), to delineate the site of action, we transfected cells with NIK, IKK, and p65 plasmids and then monitored NF-κB-dependent SEAP expression in celecoxib-untreated and -treated cells. As shown in Fig. 6B ⇓ , NIK, IKK, and p65 plasmids induced gene expression; celecoxib suppressed reporter gene expression induced by NIK and IKK plasmids but had no effect on that induced by p65. Because phosphorylation of IκBα and p65 is needed for NF-κB activation (24) , we suggest that celecoxib inhibits the kinase involved in their phosphorylation.
Celecoxib Inhibits CSC-Induced COX-2, MMP-9, and Cyclin D1 Activation.
Our results indicated that CSC activates NF-κB through activation of IKK, leading to phosphorylation and degradation of IκBα and that NF-κB is transcriptionally active. Because cyclin D1, COX-2, and MMP-9 are NF-κB-regulated genes (35, 36, 37) , we investigated whether expression of CSC-induced and NF-κB-regulated expression of products of these genes is abrogated by celecoxib. H1299 cells, either untreated or pretreated with celecoxib, were exposed to CSC for a different time, and whole-cell extracts were prepared and analyzed by Western blotting. CSC induced COX-2, cyclin D1, and MMP-9 expression in a time-dependent manner (Fig. 6C) ⇓ , and celecoxib abolished the CSC-induced expression of these gene products. Whether celecoxib can suppress the expression of these genes in normal bronchial epithelial cells was also examined by using BEAS 2B cells. Like H1299 cells, CSC induced COX-2, cyclin D1, and MMP-9 expression, and celecoxib abolished the expression (Fig. 6D) ⇓ .
Whether celecoxib can suppress CS-induced COX-2 promoter activity was also investigated. For this, cells were transiently transfected with COX-2 promoter-luciferase reporter plasmid then exposed to CSC in the presence and absence of celecoxib (Fig. 6E) ⇓ . These results also showed that CSC induced COX-2 promoter activity and celecoxib down-regulated it.
Because of the central role of NF-κB and NF-κB-regulated genes in CS- induced carcinogenesis and because of chemopreventive role of celecoxib, we investigated the effect of celecoxib on the CS-induced activation of NF-κB. In our study, exposure of human lung cells to CSC activated NF-κB and celecoxib abolished the activation. Our results also show that the effect of CS exposure on NF-κB activation was persistent and can be seen as late as 24 h after the exposure. Celecoxib can suppress the effect of CS even when administered 12 h after exposure to CS. This effect was not cell type-specific. CSC-induced activation of IKK, IκBα phosphorylation, and degradation were also inhibited by the COX-2 inhibitor, as were CSC-induced p65 phosphorylation and nuclear translocation and NF-κB-dependent reporter gene expression. CSC-induced NF-κB reporter activity induced by NIK and IKK but not that activated by p65 was also blocked. CSC induced the expression of the NF-κB-regulated proteins COX-2, cyclin D1, and MMP-9, and celecoxib abolished the induction.
Our results demonstrate that celecoxib by itself did not activate NF-κB but rather abolished CSC-induced NF-κB activation. These results agree with a recent report that showed that celecoxib inhibits interleukin 1-induced NF-κB activation in rat mesangial cells (38) . Our results, however, differ from this report in that celecoxib by itself did not activate NF-κB at higher doses (50 μm; Ref. 38 ). It is unlikely that this difference is due to the cell types used, as none of the four cell types tested in our study showed activation of NF-κB by celecoxib alone. Niederberger et al. (38) did not show by either supershift or competition assays that the celecoxib-induced DNA binding observed was indeed that of NF-κB. How celecoxib suppresses NF-κB activation was also not investigated by these workers.
Our studies indicate that celecoxib blocked CSC-induced IKK activation, IκBα phosphorylation, and degradation, leading to abrogation of NF-κB activation. How celecoxib inhibits IKK is less clear. Our results show that celecoxib is not a direct inhibitor of IKK activity. It is possible, however, that celecoxib suppresses an upstream kinase involved in the activation of IKK. What kinase activates IKK is also not clear, but several potential candidates have been identified. NIK, for example, can activate IKK. Our results indicate that celecoxib can suppress the NF-κB activation induced by NIK. Thus, whether celecoxib suppresses NF-κB activation by directly inhibiting NIK or some other kinase cannot be ruled out at present.
Our studies indicate that celecoxib suppressed the CSC-induced expression of NF-κB-regulated COX-2, MMP-9, and cyclin D1. Whereas celecoxib is known to inhibit the activity of COX-2, our results indicate that it can also suppress the synthesis of COX-2, which is regulated by NF-κB (36) . COX-2 has been implicated in carcinogenic processes, and its overexpression by malignant cells has been shown to enhance cellular invasion, induce angiogenesis, regulate antiapoptotic cellular defenses, and augment immunological resistance through production of prostaglandin E2 (39) . Additionally, it has been demonstrated that COX-2 is overexpressed in patients with lung cancer, head and neck cancer, or breast cancer (13 , 40 , 41) .
A previous report on the suppression of angiogenesis by celecoxib (16) indicated that this activity may also be mediated in part through inhibition of NF-κB-regulated MMP-9 synthesis (37) . MMP-9 plays a crucial role in tumor invasion and angiogenesis by mediating degradation of the extracellular matrix in breast cancer cells (42) ; it also mediates destruction of lung elastin in chronic obstructive pulmonary disease (43) . Inhibition of MMP activity has been shown to suppress lung metastasis (44) . Also, cyclin D1 is overexpressed in a wide variety of tumors (for references see Ref. 45 ), and it is regulated by NF-κB (35) . Altered expression of cyclin D1 is an early event in non-small cell lung carcinoma development, and cyclin D1 is known to be required for cells to advance from G1 to S-phase of the cell cycle (46) . CSC-induced NF-κB-mediated down-regulation of cyclin D1 may explain the antiproliferative effects of celecoxib.
Whether the concentrations of celecoxib used in our studies are clinically relevant and achievable is very important. Davies et al. (47) have reported a human serum concentration of celecoxib up to 8 μm. Several points need to be considered when comparing drug doses in vitro versus that in vivo. First, concentration of celecoxib similar to that used by us have been used by others in vitro (19 , 21) for suppression of PDK1 and AKT; second, we expose cells to celecoxib for short-term (just a few hours), but in vivo exposure occurs for long-term (days); third, proteins have been identified in the serum, which bind to celecoxib and neutralize its activity (20) ; fourth, the true relevant celecoxib concentration in the tissue is unclear. Thus, it is difficult to correlate the celecoxib concentration used in vitro to that achievable clinically.
NF-κB activation has also been implicated in chemoresistance (48) . Thus, our results indicate that celecoxib behaves much like other nonsteroidal anti-inflammatory agents such as sulindac and aspirin, both of which have been shown to suppress NF-κB activation (49 , 50) . The suppression of CS-induced NF-κB activation by celecoxib may explain its antitumor, chemopreventive, chemosensitivity, antiangiogenesis, antiproliferative, and apoptosis-inducing activities.
We thank Walter Pagel for a careful review of the manuscript.
Grant support: PO1 Grant (CA91844) from the NIH on Lung Chemoprevention. B. Aggarwal is a Ransom Horne, Jr. Distinguished Professor of Cancer Research.
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.
Requests for reprints: Cytokine Research Laboratory, Department of Bioimmunotherapy, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030. Phone: (713) 792-3503, extension 6459; Fax: (713) 794-1613; E-mail:
- Received January 21, 2004.
- Revision received April 12, 2004.
- Accepted May 5, 2004.
- ©2004 American Association for Cancer Research.