Phenolic antioxidants attract increasing attention for their potential as cancer chemopreventive agents.1, 2, 3, 4, 5, 6, 7, 8, 9, 10 Resveratrol (RE), a phytoalexin present in grapes and berries, is one of the most promising agents for prostate cancer prevention.11, 12, 13, 14, 15, 16, 17, 18 RE has antioxidant and antimutagenic activities that induce phase II drug-metabolizing enzymes; it mediates anti-inflammatory effects and inhibits cyclooxygenase and hydroperoxidase functions.19, 20, 21, 22In vitro research and preclinical studies strongly support the anticancer effects of RE. RE induces cell cycle arrest, inhibits DNA synthesis and induces apoptosis in prostate cancer cells and thus holds great promise for development as a chemopreventive agent for prostate cancer.17, 23, 24, 25, 26, 27, 28, 29, 30 RE is strongly linked with androgen receptor regulation. Recent studies by Mitchell et al.23 and Hsieh et al.31 have shown that RE represses several classes of androgen-upregulated genes at the mRNA level in LNCaP cells. Our earlier studies with RE indicated an inhibition of the AR co-activator ARA-24 in LNCaP cells consistent with previous reports.15 Although these cited studies point to the importance of RE as a potential chemopreventive agent, the effect of RE on transcriptional activation or repression of primary and secondary targets of p53 and AR in prostate cancer has not been investigated.
Numerous studies provide evidence for the anticarcinogenic activity of RE,17, 19, 23, 24, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45 but the precise mechanisms involved in the modulation of oncogenic precursors of prostate carcinogenesis remain to be elucidated. Earlier studies indicated that the p53 protein is stabilized and activated after exposure of mammalian cells to DNA-damaging agents.46 However, it is not known whether p53 activation by RE follows the same pathway as that initiated by other agents that induce G1 arrest and apoptosis. There are indications that the nature of p53 response depends on the levels of the p53 protein, the type of inducing agent and the cell type employed. Significant cellular death was observed only when several of the genes controlled by p53 were expressed in concert, suggesting that p53 needs to activate parallel apoptotic pathways to induce programmed cell death.47, 48
To delineate the complete cascade of molecular events in response to RE treatment of prostate cancer requires a comprehensive study on gene expression at the transcription level. In our study, we employed human cDNA microarray analysis to obtain a genetic profiling of p53-targeted genes. We focused on determining whether RE activates a cascade of p53-directed genes and transcription factors that are involved in apoptosis mechanism(s) in prostate cancer cells. In addition, we examined the p53-activated apoptotic pathway in conjunction with the modification of acetyltransferase p300 and caspase activator Apaf-1 that is induced by RE in the androgen-sensitive prostate cancer cell line (LNCaP).
RE obtained from CALBIOCHEM (La Jolla, CA) was purified by crystallization from ethanol until a purity of 99% was demonstrated in high performance liquid chromatography (HPLC) as described by Palomino et al.49 The stock solution of RE (10−3 M) was prepared in DMSO. After the HPLC purification, aliquots of RE were protected from light and stored at −20°C until use.
Cell culture and treatment
Prostate cancer cells (LNCaP) were grown at 37°C in RPMI, GIBCO brand (Invitrogen Life Technologies, Carlsbad, CA) supplemented with L-glutamine, 7.5% fetal bovine serum (FBS) and antibiotics in a water-saturated atmosphere of 5% CO2. To determine the effect of RE over time, 75% confluent cells were treated with 10−5 M and observed for 24 hr and 48 hr. This concentration at these time points were selected on the basis of our earlier observations that showed irreversible cell growth inhibition associated with apoptosis in a nontoxic manner. In our study, trypsinized and PBS-washed cells were counted with a Coulter particle counter to quantify cell growth in response to RE. Cells harvested from a parallel set of experiments were used for cell cycle analysis by means of flow cytometry and RNA extractions for gene expression analysis. We used DMSO as the solvent control appropriately in our experiments to compare the efficacy and gene expression altered after RE treatment.
Detection of apoptotic cells and DNA fragmentation analysis
Apoptosis induced by RE (10−5 M) over 24 hr and 48 hr in treated LNCaP cells was localized with DAPI staining. Cells showing fragmented and condensed nuclear material characteristic of apoptosis were localized at 40× magnifications under a research grade fluorescence microscope (AX70-Olympus). The percentage of apoptotic cells with characteristic morphologic changes was determined from 3 identical experiments with RE and compared to that in control experiments. To further confirm the induction of apoptosis, a DNA fragmentation analysis was conducted simultaneously by extracting DNA from a similar set of experiments at different time periods as described by us earlier.13, 15
Flow cytometric analysis
LNCaP cells treated with 10−5 M of RE or with control were harvested at 24 hr and 48 hr time periods, washed in PBS and fixed in 1% formaldehyde for 15 min on ice. After rewashing with PBS, the cells were fixed in 80% ethanol for 30 min. Cell suspensions having approximately 3 × 106 cells were centrifuged and the pellets of cells were resuspended with PBS and further treated with 1 mg/mL of RNase at room temperature for 30 min. Propidium iodide (1 mg/mL, final concentration in PBS) was added and flow cytometric analysis was performed after 30 min. For cell cycle analysis, we employed Epics XL MCL (Phoenix Flow Systems; San Diego, CA).
Immunofluorescence detection of p300
LNCaP cells grown in 35 mm cell culture dishes were treated with 10−5 M of RE for 48 hr. Cells washed with 1× PBS were then fixed in formalin for 15 min before incubation with appropriate concentrations of anti-p300 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) for 1 hr. After washing with PBS twice, cells were again incubated with FITC-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) for 45 min, followed by a final wash with PBS and mounted in DAPI/Antifade (ONCOR, Gaithersburg, MD). Fluorescence signals for p300-positive cells were visualized under 40× magnification using the fluorescence microscope (AX-70 Olympus). The percentages of p300-positive cells determined from 3 identical experiments treated with RE were compared to control experiments.
cDNA microarray analysis
Human cDNA microarrays purchased from Perkin Elmer-NEN Life Sciences (Boston, MA) spotted with 2,400 genes per array (known genes and ESTs), sequence-verified by Alpha Gene, were used for gene expression profiling. More than 60% of the genes are full-length copies of known genes, which include two types of control genes, a complement of 3 nonmammalian (plant) genes and 35 housekeeping genes (human).
Total RNA was isolated from RE-treated and control LNCaP cells using Trizol reagent and Qiagen columns (Invitrogen Life Technologies and Qiagen, Valencia, CA, respectively). RNA from control cells was labeled with cyanine cy3-dUTP and RNA from RE-treated cells was labeled with cy5-dUTP. Briefly, an aliquot of RNA (20 μg) for each labeling was precipitated with 2.5 volumes of cold ethanol and 0.1 volume of 3 M sodium acetate. The precipitated RNA was then harvested by centrifugation and resuspended in 17 μl of DEPC (diethylpyrocarbonate)-treated water. This was added to 2 μl of dNTP primer mix (PE/NEN Life Sciences). The reaction mixture was kept at 65°C for 10 min to denature the nucleic acid, then cooled to 25°C for 5 min. It was then reacted with 2.4 μl of 10× RT buffer, 2 μl of AMV RT/RNase inhibitor mix, 4 μl of cy3-dUTP and 2 μl of cy5-dUTP (PE/NEN Life Sciences) for the reverse transcription reaction that was carried out for 1 hr at 42°C. The probes were cooled down to 4°C for 10 min, and 2.5 μl of 0.5 M EDTA as well as 2.5 μl of 1 N NaOH were added before incubation at 65°C for another 30 min. After cooling the above mixture to 4°C for 5 min, 6.5 μl of 1 M Tris-HCl was added. The cDNA was purified by isopropanol precipitation or by using a spin column. The dried pellet was resuspended in 13 μl of RNase-free water; the cy3-dUTP-labeled control and cy5-dUTP-labeled products were combined and 10 μg of carrier DNA was added. After adding 2 μl of 5 M ammonium acetate and 50 μl of isopropanol, the probe mixture was pelleted at 14,000 rpm for 10 min at 4°C. The pellets were washed again with 70% and 95% ethanol respectively and were stored at −20°C for further hybridization.
The microarray to the probe was hybridized according to manufacturer's instructions (PE/NEN Life Sciences) with appropriate modifications as described in our earlier study.15 The mixed probe pellet was resuspended in 25 μl of hybridization buffer and immediately added to human cDNA arrays: a 22 × 22 cover slip was placed carefully on the arrays that were then transferred into hybridization cassettes especially made for submerging the arrays in a 65°C water bath for 12 hr. After hybridization, the slides were washed with microarray wash buffer (0.5 × SSC, 0.01% SDS) until the cover slip fell off. The slides were washed twice more, once with 0.5 × SSC, 0.01% SDS and again with 0.06 × SSC, 0.01% SDS for 15 min. A final wash of the slides were with 0.06 × SSC for 15 min and the cleaned slides were spin-dried before scanning.
Scanning and data analysis
The Axon GenePix 4000B scanner, a nonconfocal scanning instrument containing 2 lasers that excite cyanine dyes at 635 nm for Cy5, and 532 nm for Cy3, respectively, and high-resolution (10 μm pixel size) photo multiplier tubes that detect fluorochrome emission, were used for scanning the hybridized cDNA microarrays (Axon Instruments, Foster City, CA). GenPix Pro software was used to analyze the images and to extract the data sets into a Microsoft Excel spreadsheet. The data sets consist of signal and background intensity, standard deviation of signal and background intensity and ratio of median and/or mean of total intensity, including flags. A macroprogram in Microsoft Excel enabled normalization; we used the intensity of each spot for variations in the overall intensity of the image with respect to control image to perform normalization. A GeneSpring Bioinformatics software package was used for data analysis (Silicon Genetics, Redwood City, CA) as described earlier.15
Validation of cDNA microarray results
To verify the gene expression pattern, experiments were repeated with 3 similar sets of arrays containing similar sets of genes and prosite motifs. The level of expression pattern was compared to similar spots present in the respective quadrants in all slides simultaneously; 87% of the spots showed similarity in their expression in all arrays that were hybridized with similar samples. High- and low-expressed genes were confirmed with RT-PCR using sequence-specific primers. RT-PCR conditions and a detailed protocol are described elsewhere.15
Western blot analysis
LNCaP cells treated with RE (10−5 M) for 48 hr were harvested by trypsinization. Cellular protein was extracted and Western blotting for p53 and p300/CBP protein was carried out using mouse monoclonal antibody for p53 (DO-1) and p300/CBP purchased from Santa Cruz Biotechnology. The reactive protein bands were developed with chemiluminescent detection reagents (ECL Kit, Amersham Biosciences, Piscataway, NJ).
RE-induced G1 arrest and apoptosis
The anticancer effect of RE was studied on exponentially growing LNCaP cells that were treated with the agent (10−5 M) for 24 hr and 48 hr periods. RE-induced cell death via apoptosis was examined by DAPI staining (Fig. 1a) and DNA fragmentation analysis (Fig. 1b). We demonstrated by DAPI staining a remarkable decrease in cell growth inhibition by RE via apoptotic cell death after 48 hr. Cell death analysis indicated 30% apoptotic cells (Fig. 1c) with characteristic morphologic and nuclear material changes.
Cell cycle analysis showed a minor G1 arrest after 24 hr as determined by flow cytometry (Fig. 2). Figure 2a and b indicate that there was a pronounced G1 (84%) peak after 48 hr with far fewer cells in the S phase (8%). The reduced number of cells in the S phase in RE-treated cells may indicate a possible cell cycle arrest at the G2/M phase. Control cells did not exhibit much of an effect on the G1 fraction; on the other hand, it exhibited a higher DNA content with 26.0% of cells in the G2/M phase after 24 hr and 31.6% after 48 hr (Fig. 2c,d), indicating a 3- to 4-fold increase in the cells in G2/M phase.
Effect of RE on gene expression profile
Measurements on the expression levels of differentially expressed genes are presented in Figure 3 as a simple bivariate scatter plot. Upregulated genes amounted to 5.25% and downregulated genes accounted for 17.88% of all expressed genes out of a total of 2,400 genes. To get reproducible results, the experiments were repeated with the same source of RNA from LNCaP cells treated with RE and were compared with control cells. Genes expressed were confirmed with RT-PCR (Fig. 4), using sequence-specific primers designed for selected genes.
The results presented here are the averages from triplicate experiments. A list of differentially expressed genes shows at least a moderate level of expression, varying more than 2-fold between the control and RE-treated cells (Table I). A 2-fold difference in the gene expression was preferred because (i) the signal intensities of nonhomologous plant genes as internal controls present in the array were used; (ii) with respect to the intensity ratio of plant genes in our analysis, we decided to keep the signal:background ratio as 1.0. This ratio was used to normalize the entire array so that all the slides could be compared directly for a 2-fold increase. Although many of our microarray data analyses were performed with a cutoff for 5-fold increase as preferred by other investigators,50 in our study we aimed to separate the genes of interest; for instance, with respect to p53-responsive genes, a 2-fold increase permitted us to identify a large number of transcription factors related to their activation. Although a 5-fold increase is a highly significant signal for the expressed genes, a 2-fold increase is an acceptable level of difference in expression as indicated by several earlier studies51, 52 within a 95% confidence interval as in our present study.
Table I. Differentially Expressed Genes in LNCaP Cells Exposed to Resveratrol Related to P53-Mediated Molecular Targets
Differential expression of primary and secondary targets of p53 genes
Given the significant role that the tumor suppressor gene p53 appears to play in human cancer, we analyzed whether RE activates more than a few known p53 targets or whether there may be expression of other gene transcripts that contributes to p53-mediated apoptosis. Surprisingly,this approach revealed for the first time the differential expression of more than 34 transcripts that are either primary or secondary targets of p53 and are either activated or repressed by RE (Table I). Sixteen of these 34 transcripts, including the programmed cell death factor 2 (PDCD-2), p300, Apaf-1, CPP32, PIG 7 and PIG 8, BAK protein, p57 (Kip2) and an isoform of zinc finger protein 135, that were expressed at significantly higher levels (increased > 2-fold, p < 0.01) are involved in apoptosis. Likewise, a significant fraction of genes that are involved in DNA damage, cell cycle, oxidative stress and microtubule-disturbing activity were downregulated.
RE-induced apoptotic genes
RE has been shown to induce apoptosis and to inhibit the growth of certain cancer cells in vitro. In our study, using LNCaP cells, we demonstrated activation of several pro-apoptotic genes that are listed in Table I. The results presented here are consistent with the activation of p53 target genes that are involved in cell growth inhibition and apoptosis. Activation of Apaf-1, BAK protein, cystein protease CPP32 isoforms, DNA fragmentation factor and apoptosis-specific proteins provides additional evidence for an effect of RE on cell death induction in LNCaP cells.
Effect of RE on transcription factors
As shown in Table I, we observed activation of acetyltransferase p300 associated with alterations in the expression of other transcription factors, including transcription activator TFIIE β, zinc finger protein, seryl tRNA synthetase and putative tumor suppressor ST13. Changes in the expression of these genes seem to be consistent with the shutdown of several other oncogenic precursors, such as serine threonine kinase, phospholipase C and phospholipase A2, K-ras oncogene, androgen receptor-associated protein and prostate-specific antigen precursor protein. Detection of another set of pro-apoptotic genes, namely Apaf-1, 14-3-3 and IGF-binding protein 2, suggests a potential role of RE in the modification of these gene transcripts in G1-arrested LNCaP cells.
Expression of p300 and Apaf-1
In addition to microarray and RT-PCR analysis, we observed an increase in the total protein for p300 in RE-treated LNCaP cells (Fig. 5). Expression of transcriptional co-activator p300 that could acetylate human p5333, 34 was high along with that of the apoptosis-specific protein Apaf-1. Nuclear localization by immunofluorescence detection (Fig. 5a) and quantification (Fig. 5b) corresponds with the results of Western blot analysis (Fig. 5c) for total protein expression. Overall, these findings support the expression and activation of p53 and p300, the key transcription factors that are involved in cell growth arrest and apoptosis.
Effect of RE on androgen receptor and PSA
Proteins that are known to interact with several steroid receptors also influence the functional activity of the wild-type AR.53, 54, 55, 56 Two reference genes, namely androgen receptor-associated protein (ARA 24) and PSA, were first identified by microarray and then confirmed by repeated RT-PCR (Fig. 4). Results from microarray and RT-PCR analyses indicated a downregulation of androgen receptor-associated protein ARA-24 and PSA, suggesting a potential role for RE in inducing cell growth inhibition via anti-androgenic mechanisms.
NF-kB regulation by RE
Results from the microarray analyses indicated significant downregulation of NF-kB p65 (Table I), while the level of NF-kB p50 expression remained unchanged. RT-PCR analyses confirmed a lower expression level for PPARα and NF-kB p65 (Fig. 4). However, an inverse association between the p300 and NF-kB p65 confirmed a possible p53 targeted transcriptional inactivation of nuclear factors that may be cell or organ specific. This requires further investigation. Delineation of the differential regulation of p300, PPARα and the NF-kB family of genes by RE in LNCaP cells is in progress. Comparing the level of expression of NF-kB and the anti-apoptotic Bcl2 family of genes shown in Table I, there is differential expression within the Bcl2 family of genes. For example, there was a higher activation of BAK and another phosphorylated isoform of Bcl2 (identified by Western blot, data not shown) that suggests a possible reprogramming of these genes by RE.
Gene expression analysis by DNA microarray presented here is part of our ongoing research on antioxidant-regulated genes in human prostate cancer. To determine the effect of RE on genes involved in apoptosis and to study the associated cascade of molecular events in LNCaP cells, we used a human cDNA microarray with 2,400 genes. Results of our study with respect to gene expression analysis are limited due to the single time-point measurement in one cell type (LNCaP) at the specified 48 hr point after RE treatment. Although deeper biologic insight with respect to transcriptional regulation of specific genes is likely to develop from cDNA microarray analyses at multiple time points with multiple cell lines, our aim in this study was to gain a general understanding of the genetic response of LNCaP cells to RE treatment. Toward this aim, we used a time (48 hr) and dose (10−5 M) evaluation for RE that had already been tested in the induction of cell cycle arrest and apoptosis in androgen-sensitive LNCaP cells. We preferred such a focused approach mainly to identify those molecular targets with a specific dose and time regimen that may lead to a different phenotype. Although studies with LNCaP cells generally describe a poor correlation between RNA levels and the corresponding proteins,50 findings from this study support the hypothesis that RE induces differential expression of genes at the transcription level that are directly or indirectly involved in cell cycle control and apoptosis mediated by p53-dependent mechanisms. Exploring transcriptional and post-translational modifications of target genes and proteins modified by RE in a time- and dose-dependent manner may provide more insight into differences in the RNA vs. protein expression. Overcoming these limitations may be possible via further perfection in technologies, such as aDNA microarray and 2D gel/MS analysis, to detect mRNA and the corresponding protein interactions in the same sample.
RE exerts chemopreventive activity by inhibiting cellular events during tumor initiation, promotion and progression.17, 19, 28, 37, 41, 44 Using cDNA microarray as well as RT-PCR, immunofluorescence and Western blot analyses, we have demonstrated alterations in the expression of p53 target genes associated with cell cycle arrest and apoptosis as a result of RE treatment in androgen-dependent LNCaP cells. Activation of pro-apoptotic genes including p53, p21(waf1/Cip1), PIG 7 and PIG 8, p300/CBP, Apaf-1, BAK and zinc finger proteins by RE was consistent with the downregulation of ARA 24 and PSA, NF-kB p65 and Bcl2. This demonstrates a potential role of RE in inducing apoptosis while downregulating the expression of prostate cancer precursor genes either by activating p53 signaling mechanisms and/or by acting in synergy with blocking other androgen-signaling pathways. In our earlier studies with human cancer cells, we have found a remarkable increase in the level of p53 mRNA and protein, in conjunction with an increase in the level of p21 (Waf1/Cip1) after exposure to phenolic antioxidants.13, 15 The current study indicates involvement of more than one apoptotic pathway in LNCaP cells in response to RE. Activation of Apaf-1, a novel protein induced by RE, has been reported to participate in the cytochrome-C-dependent activation of caspase 3,57 which triggers a cascade of events in apoptosis. More than 30 genes in this category, specifically proteases and protease inhibitors, are stabilized and activated when mammalian cells are exposed to a variety of stressors including DNA-damaging agents.58 Expression and activation of p300 and the tumor suppressor gene p53 in LNCaP cells suggests a possible role for RE in inducing acetylation at the carboxyl terminus of p53; this type of modification is known to increase the level of sequence-specific DNA binding of p53.48, 59 It is also important to note that RE may induce the antiproliferative effects in LNCaP cells through post-translational modifications of p53, followed by a post-transcriptional activation of p300 after DNA damage in G1-arrested cells.60 Direct evidence demonstrating a role for p300 in human tumors was lacking until recently, but now several studies strongly support a role for p300/CBP as a tumor suppressor.61 The results from our present study underscore the tumor suppressor effect of highly activated p300 in G1-arrested LNCaP cells exposed to RE. In addition, AR transactivation activity been shown to be triggered by androgens or by growth factors; moreover, CBP/p300 has been reported to play a major role in the ligand-dependent AR transactivation activity.62 Cell culture studies have shown that p300 and CBP can act as transcriptional co-activators for a large number of transcription factors. The co-activation potential of p300 and CBP is in part brought about by an acetyltransferase (AT) activity located in the central part of these 2 proteins. In vitro, histones and certain transcription factors, such as p53, are acetylation substrates of p300.48, 63 However, the biologic significance of RE-induced p300 in ligand-dependent or -independent AR transactivation is as yet unknown. Recent studies indicate that AR can be modified by acetylation both in vitro and in vivo;64 acetyltransferase co-activators, such as RE, have important roles in targeting AR function and the associated signaling mechanisms involved in prostate carcinogenesis. RE also activates ERKs and p38 kinase, which, in turn, mediate apoptosis through phosphorylation of p53 at serine 15.28 MEKKI-activated p300-mediated transcription may be involved in RE-induced apoptosis; however, this effect may be due to modulation of more than one signaling pathway. A mechanistic model developed on the basis of currently available comprehensive data generated from this study, by microarray analysis, RT-PCR, Western blot and immunofluorescence detection in LNCaP cells with or without RE treatment for 48 hr, together with information from earlier studies,65 is presented in Figure 6.
In summary, a number of the p53 target genes revealed in our microarray analysis provide new molecular targets that may also serve to assess the efficacy of other potential chemopreventive agents for prostate cancer besides RE. There are indications that the nature of p53 response depends on the levels of p53 protein, the type of inducing agent and the cell type employed. Significant cellular death was observed only when several of the genes controlled by p53 were expressed in concert, suggesting that p53 appears to activate parallel apoptotic pathways to induce programmed cell death.47, 48 The full delineation of the cascade of molecular events in response to RE treatment of prostate cancer requires a comprehensive study on gene expression at the transcription level.
This study was presented at the American Association for Cancer Research 93rd Annual Meeting (Abstract #830, Cancer Prevention III, Prevention and Survivorship Research 2, Poster Discussion Session), 2002. We thank Ms. I. Hoffmann for editing the manuscript.