Purpose: The transcription factor nuclear factor-κB (NF-κB) promotes the production of angiogenic, antiapoptotic, and prometastatic factors that are involved in carcinogenesis.
Experimental Design: Electromobility gel shift assays were used to evaluate NF-κB DNA binding in vitro. The functional relevance of NF-κB DNA binding was assessed by both cDNA array analyses and proliferation assays of prostate cancer cells with and without exposure to an NF-κB inhibitor, parthenolide. Immunohistochemistry staining for the p65 NF-κB subunit was used to determine the frequency and location of NF-κB in 97 prostatectomy specimens. The amount of staining was quantified on a 0–3+ scale.
Results: An electromobility gel shift assay confirmed the presence of NFκB DNA binding in all four prostate cancer cell lines tested. The binding was inhibited by parthenolide, and this agent also decreased multiple gene transcripts under the control of NF-κB and inhibited proliferation of prostate cancer cells. The staining results revealed overexpression of p65 in the prostatic intraepithelial neoplasia and cancer compared with the benign epithelium. Specifically, there was a predominance of 1+ and 2+ with no 3+ staining in benign epithelium, whereas there was only 2+ and 3+ staining (30 and 70%, respectively) in the cancerous areas. These differences were statistically different. There was no correlation with tumor grade or stage.
Conclusions: NF-κB is constitutively activated in prostate cancer and functionally relevant in vitro. Immunohistochemistry of human prostatectomy specimens demonstrated overexpression of the active subunit of NF-κB, p65, and that this occurs at an early stage in the genesis of prostate cancer. This work supports the rationale for targeting NF-κB for the prevention and/or treatment of prostate cancer.
Nuclear factor-κB (NF-κB) is a dimeric transcription factor composed of members of the Rel family (1) . The predominant NF-κB dimers are the transcriptionally active p65:p50 heterodimer and the less active p50:p50 homodimer (2) . NF-κB dimers are bound to inhibitory IκB proteins in the cytoplasm and released from inhibitor of nuclear factor-κB (IκB) after phosphorylation of IκB by Iκ kinases followed by proteosome-mediated degradation of IκB. The release of NF-κB and subsequent binding to DNA occurs in response to a variety of stimuli, including chemotherapy, radiation, and cytokines such as tumor necrosis factor α (TNF-α) and interleukin (IL)-1. Genes activated by NF-κB play a central role in many of the hallmarks of prostate cancer, including invasion (IL-6 and matrix metalloproteinase 9), angiogenesis (IL-8, vascular endothelial growth factor), and inhibition of apoptosis (cIAP 1, c-IAP 2, TRAF-1, TRAF-2, Bfl-1/A1, Bcl-XL, and manganese superoxide dismutase; Refs. 3, 4, 5, 6 ).
Induction of NF-κB promotes resistance to TNF and chemotherapy (6) and inhibition of NF-κB activation by overexpression of IκB restores chemotherapy and TNF sensitivity. This suggests that constitutive NF-κB activation may be an important mechanism of hormonal and chemotherapy resistance. Studies of cell lines in vitro have shown that NF-κB is constitutively activated in many malignancies, including breast cancer and prostate cancer (7 , 8) . Recently, investigators have shown that inhibition of NF-κB by insertion of mutated IκB into cancer cell lines causes decreased vascular endothelial growth factor and IL-8 expression and was associated with decreased in vivo growth with decreased angiogenesis of an ovarian cancer cell line (9 , 10) . Emerging preclinical evidence implicates NF-κB in the development of prostate cancer because it has been shown to regulate bcl-2 transcription (8) and inhibition of NF-κB results in apoptosis (11) and cell cycle arrest (12) .
Therapy directed against the transcription factor NF-κB is therefore a prime target and has the potential to improve the prognosis of prostate cancer patients. This latter point is underscored by findings from the clinical trials of the proteasome inhibitor PS 341. This drug blocks NF-κB (in addition to other pathways) and induces remissions and suppression of proteins under NF-κB control (e.g., IL-6) in patients with multiple myeloma, hormone refractory prostate cancer, and non-small cell lung cancer (13 , 14) .
Electromobility gel shift analyses were undertaken to confirm the presence of NF-κB DNA binding in vitro in a variety of prostate cancer cells lines and to determine whether this could be inhibited by the NF-κB inhibitor parthenolide. The functional relevance of NF-κB DNA binding was assessed by cDNA array analyses and proliferation assays of prostate cancer cells with and without exposure to a NF-κB inhibitor, parthenolide. The immunohistochemistry analysis was undertaken to determine the frequency and location of NF-κB by staining for the p65 subunit of the Rel family.
MATERIALS AND METHODS
Electromobility Gel Shift Assay.
All prostate cancer cells were plated on 100-mm plates and harvested in exponential growth phase. The prostate cancer cells were (a) the hormone-sensitive cell line LNCaP (in three different passage states: low passage, 11 passages; intermediate passage, 38 passages; and high passage, 208 passages); (b) PC-3 (hormone-independent cell line) and (c) C42 (a hormone-independent subclone of LNCaP). The LNCaP P-11 and 208 and PC-3 cells were cultured in MEM with 10% fetal bovine serum and were obtained from American Type Culture Collection (Manassas, VA). The LNCaP P-38, the parental clone of the C42 cells, and the C42 cells were provided by Dr. Thomas Gardner and cultured in T-media (Life Technologies, Inc., Rockville MD). The CWR22Rv1 cell line was kindly provided by Dr. James Jacobberger (15) . The cells were grown in DMEM containing 10% FCS.
Solvent control and parthenolide dissolved in 100% ethanol (Aldrich, Milwaukee WI) were added 3 h before harvesting. Whole cell extracts were prepared as previously described (7) and incubated with a radiolabeled NF-κB probe for 30 min at room temperature. The oligonucleotide probe binds to the NF-κB DNA binding site in the promoter region of the immunoglobulin gene. Electrophoresis and autoradioragraphy were performed as previously described (7) using NF-κB and SP-1 probes (Promega, Madison, WI). The specificity of parthenolide inhibiting NF-κB DNA binding was verified by the use of the SP-1 probe as a control.
cDNA Array Analysis.
Total cellular RNA was extracted from CWR22Rv1 after 6 h of exposure to solvent control and 10 μmol/liter parthenolide using RNeasy Minikit (Quiagen, Valencia, CA) according to the manufacturer’s instructions. The NF-κB pathway GE array kit was obtained from SuperArray Bioscience Corp. (Bethesda, MD). The kit determines expression of 96 genes that either interact with or are controlled by NF-κB. Total RNA from respective samples were used as a template to generate biotin-labeled cDNA probes using GEArray Ampolabelling RT kit (SuperArray Bioscience Corp.). The cDNA probes corresponding to the mRNA population were then denatured, and hybridization was carried out in GEHyb solution to nylon membranes spotted with gene-specific fragments. Membranes were then washed in 2× SSC, 1% SDS twice for 15 min each, followed by 0.1× SSC, 0.5% SDS twice for 15 min each. Chemiluminescence was used to visualize the expression levels of each transcript, and the results were quantified with the GEArray Analyzer. The change in a given gene transcript was estimated by normalizing the signal intensities with the signals derived from glyceraldehyde-3-phosphate dehydrogenase and β-actin.
LNCaP and PC-3 cells were plated in a 96-well U-bottomed plate (Becton Dickinson Labware, Franklin Lakes, NJ) at a concentration of 5000 cells/50 μl media and incubated in 5% CO2 at 37°C for 24 h. Varying parthenolide concentrations in 50 μl of media were added to the media 24 h later. Colorimetric readings were obtained using the Cell Titer 96 AQueous Non-radioactive Cell Proliferation Assay (Promega Corp.; Madison, WI) system and an ELISA plate reader after 48 h of exposure to parthenolide. The readings obtained for each concentration tested were from an average of eight wells. Each experiment was expressed as a percentage of the solvent control and completed at least three times with consistent results. The results presented are an average of three experiments.
Ninety-seven cases of radical prostatectomy and bilateral lymphadenectomy between 1990 and 1994 were obtained from the surgical pathology files of Indiana University Medical Center. Institutional Review Board approval was obtained from the Indiana University Purdue University Institutional Review Board. Complete clinical and pathological data were available for 93 patients. Patients ranged in age from 51 to 78 years (mean, 63 years). Grading of the primary tumor from the radical prostatectomy specimens was performed according to the Gleason’s system. The Gleason grade ranged from 4 to 10. Pathological stage was performed according to the 1997 Tumor-Node-Metastasis system. Pathological stages were as follows: T2a (n = 12 patients); T2b (n = 42 patients), T3a (n = 25 patients, and T3b (n = 14 patients; Table 1⇓ ). Thirteen (14%) patients had lymph node metastasis at the time of surgery. Clinical and complete pathological data were not available for four patients.
The antibody used was a rabbit polyclonal antibody that identifies the NH2-terminal domain of NF-κB p65 of human origin (SC 109; Santa Cruz Biotechnology, Santa Cruz, CA). Serial 5-μm thick sections of formalin-fixed slices of radical prostatectomy specimens were used for the studies. The tissue blocks containing the highest Gleason score and the maximum amount of tumor were selected. One representative slide from each case was analyzed, and we recognized the limitation of sample variation. Samples were removed from the paraffin by placing them three times in xylene for 5 min and then rehydrating through graded ethanol and finally immersion in distilled water. Slides were then rinsed in Tris-buffered saline. Antigen retrieval was performed by using the Dako Target Retrieval kit (Dako, Carpinteria, CA) containing a citrate buffer (pH 6.0) for 20 min at ∼95°C. Dako’s Avidin Biotin blocking system was used for 10 min, and the tissue sections were then rinsed with Tris-buffered saline. The nonspecific binding sites were blocked by incubating with Dako’s Protein Block for 20 min. Tissue sections were then incubated with the polyclonal rabbit antibody against p65 (1:100 dilution at room temperature for 60 min). After washing with Tris-buffered saline, the secondary antibody, Dako Link (Dako LSAB2 kit) was applied for 20 min and then rinsed with Tris-buffered saline. Additional washing was followed by incubation with streptavidin horseradish peroxidase (Dako Label, LSAB2 kit) for 20 min. Immunoreactivity was visualized by incubation of sections with 3,3′-diaminobenzidine in the presence of hydrogen peroxide. Sections were counterstained with light hematoxylin and mounted with a coverslip. All of the procedures were performed at room temperature.
The extent and intensity of staining were evaluated by a single pathologist (L. Cheng) in benign epithelium, prostate intraepithelial neoplasia (PIN), and adenocarcinoma from the same slide for each case. Microscopic fields evaluated and scored were those with the highest degree of immunoreactivity. A numeric intensity score of between 0 and 3 was assigned to each case on a scale of from 0 to 3 (0, no staining; 1+, weak staining; 2+, moderate staining; and 3+, strong staining). Inter- and intraobserver variation of staining was not assessed in the current study. The authors recognized the limitation and inherent subjectivity of immunostaining evaluation. A representative slide with 3+ staining is presented in Fig. 4⇓ .
The intensity of staining in benign epithelium, PIN, and adenocarcinoma were compared using the Cochran-Mantel-Haenszel tests for correlated ordered categorical outcomes. Pairwise comparisons between the tissue types were made if the ANOVA revealed significant treatment effects. P < 0.05 was considered significant, and all P values were two-sided.
Electromobility Gel Shift Analysis.
We have found in vitro that NF-κB DNA binding is present in all prostate cancer cell lines evaluated (Fig. 1)⇓ . The proportion of the p65:p50 heterodimer to p50:p50 homodimer is lowest in the LNCaP cell lines that have undergone the least number of passages in culture compared with (a) the LNCaP cell lines that have undergone a higher number of passages and (b) the hormone independent cell lines—C42 and PC-3 (Fig. 1A)⇓ . Also of note is that parthenolide is able to inhibit the NF-κB DNA binding in a dose-dependent manner starting at 0.5 μmol/liter. EMSA also showed that NF-κB DNA binding, present in the androgen-independent prostate cancer cell line CWR22Rv1 (Fig. 1B)⇓ , and supershift assay on two occasions confirmed the presence of the active subunit p65 (data not shown). NF-κB DNA binding was again inhibited by parthenolide (Fig. 1B)⇓ .
cDNA Array Analysis.
The relative expression level of 96 transcripts was determined using chemiluminescence and analyzed with GEArray Analyzer. All results were normalized by adjusting for the signal derived from glyceraldehyde-3-phosphate dehydrogenase and β-actin spots. The change in a given gene transcript from one membrane/experiment was estimated by comparing the signal intensities of paired specimens: parthenolide treated versus solvent control (Fig. 2)⇓ . Genes that had at least a 10% decrease after 6 h of treatment compared with untreated control in two consecutive experiments were arbitrarily considered to show a magnitude and consistency in effect between experiments to support the hypothesis that genes under NF-κB control are present in prostate cancer cells and provide relevance to the electromobility gel shift findings. The 18 genes that met these criteria and have previously been shown to be under the transcriptional control of NF-κB (16) are detailed in Table 2⇓ . It is of note that genes associated with the hallmarks of cancer (17) were decreased with parthenolide treatment. Specifically, genes associated with evasion of apoptosis, TNF receptor associated factor-1 and TNF receptor associated factor-5, were decreased. Also, genes associated with maintaining cell-cell adhesion and thus preventing metastasis (intercellular adhesion molecule-2, ICAM-5) and genes associated with inflammation (TNF) were also decreased.
To assess if the inhibition of NF-κB in prostate cancer cells by parthenolide has antiproliferative properties we performed an Cell Titer 96 Aqueous Non-Radioactive Cell Proliferation Assay (Promega Corp.) proliferation assay. As detailed in Fig. 3⇓ , 48 h of exposure to parthenolide resulted in a dose-dependent inhibition of prostate cancer cell proliferation with an IC50 of 4 μmol/liter (dose to inhibit proliferation to 50% of control) for both LNCaP and PC-3 cells.
Immunoreactive staining for the p65 subunit of NF-κB was cytoplasmic (Fig. 4)⇓ . No immunoreactivity was seen in stromal cells. When the extent of staining in the normal, PIN, and adenocarcinoma specimens was quantified on the 0–3+ scale, there was a graded increase between the normal epithelial tissues, PIN, and cancer (Table 3)⇓ . Specifically, none of the benign glands had 3+ and 66% exhibited 0 or 1+ staining. The amount of staining in the PIN lesions was intermediate with 74% of lesions having 1+ or 2+ and 25% had 3+. In contrast all of the cancer specimens were 2+ to 3+ (30 and 70%, respectively). The cancer specimens had no 0 or 1+. The difference in amount of staining was significantly greater in the PIN lesions compared with the benign glands (P < 0.0001) and was greater in the invasive neoplastic disease compared with the intraepithelial neoplasia (P < 0.0001).
Multiple pathological prognostic features and clinical parameters were correlated with the amount of staining (0–2+ versus 3+) in the cancer specimen. There was no association between T-stage, grade, prostate-specific antigen relapse, or extraprostatic extension. However, there was a trend toward an association between extent of tumor involvement in the prostatectomy and 3+ staining (P = 0.056).
This data demonstrates the presence of constitutive NF-κB DNA binding in a variety of prostate cancer cells in vitro and that this can be inhibited by parthenolide. Of note is that the amount of NF-κB DNA binding is increased in LNCaP cells that have been in culture longer and the ratio of the p65:p50 heterodimer to p50:p50 homodimer is increased in the hormone-independent cell lines compared with the hormone-dependent cells. The p65:p50 heterodimer is more biologically active than other dimers and is associated with the antiapoptotic properties of NF-κB (18) . This suggests that increased NF-κB activation may be associated with hormone independence. Also, the observation that cells that have been in culture longer develop an increase in NF-κB DNA binding may partially explain the observation by Igawa et al. (19) who have shown the LNCaP cell line becomes androgen independent with greater cell passages. Other investigators have found constitutive NF-κB DNA binding in prostate cancer cells in vitro and that NF-κB can be inhibited by a variety of agents, including selenium (20) , ibuprofen (21) , ibdehydroxymethylepoxyquinomicin (22) , dexamethasone (23) , and genistein (24) .
The relevance of NF-κB DNA binding in prostate cancer cells was confirmed by screening for the presence of and changes in genes under the control of NF-κB. We clearly have shown that many NF-κB-related genes that promote the cancer process are expressed in CWR22Rv1 cells and are decreased by the NF-κB inhibitor, parthenolide. The inhibition of NF-κB DNA binding was associated with an inhibition of cancer cell proliferation.
The p65 subunit is the more relevant component of NF-κB, and this article characterizes its location and frequency in human prostatectomy specimens by immunohistochemical staining. The differential staining (frequency and amount) and graded increase between the normal, PIN, and cancerous areas supports the contention that NF-κB is involved in carcinogenesis and, in particular, the genesis of prostate cancer. The presence of a high amount of staining in all of the prostate cancer specimens in vivo supports the in vitro electromobility gel shift assay data, which shows that increased NF-κB is a universal finding in prostate cancer and that it is involved in the development and propagation of prostate cancer (8 , 11 , 12 , 25) . The lack of nuclear staining is due either to the very short half-life of transcription factors in the nucleus or the fact NF-κB is increased in amount and is yet to be activated. Although this immunohistochemistry staining does not document the presence of activated NF-κB, it at least demonstrates that prostate cancer cells are primed to the deleterious effects of NF-κB activation by cytokines found in the prostate cancer microenvironment such as IL-1α (26 , 27) . Our attempts to employ other antibodies that recognize an epitope that overlaps the nuclear localization signal of the p65 subunit, and hence identify released/activated NF-κB, were unsuccessful because of our inability to minimize the background staining (data not shown).
The lack of association between pathological and clinical prognostic features is probably due to the universal staining at either 2+ or 3+ intensity in the cancer cells. The presence of NF-κB in normal glands at a lower frequency and amount is because it is a ubiquitous transcription factor with a baseline level of activity to promote growth and survival of all cells. In contrast, the universal presence and higher amount in the cancer specimens suggests that this transcription process is involved in the malignant transformation. The observation of the sudden increase in NF-κB expression from the benign epithelium to the PIN lesions suggests that activation of NF-κB is an early event in prostate cancer carcinogenesis. Findings supporting this have been made in breast cancer. Specifically, carcinogen treatment of female Sprague Dawley rats in vivo and in human mammary epithelial cells in culture resulted in NF-κB activation just before malignant transformation of the breast tissue (28) .
This article provides strong support for the hypothesis that NF-κB is increased and relevant at an early stage in prostate cancer because it is found in both PIN and cancerous lesions. This works provides support for the development of NF-κB inhibitors as either preventative agents (prevent the transition of normal cells to PIN or prevent the progression from PIN to cancer) or as a therapy for advanced prostate cancer.
Grant support: Department of Defense Grant DAMD17-02-1-1-0072, American Institute Cancer Research Grant OOA047, and the Walther Cancer Institute.
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: Christopher Sweeney, Indiana University, Room 473, 535 Barnhill Drive, Indianapolis, IN 46202. Phone: (317) 274-3515; Fax: (317) 274-3646; E-mail:
- Received April 15, 2003.
- Revision received April 29, 2004.
- Accepted May 10, 2004.