Prostate cancer chemoprevention agents exhibit selective ... - Nature

Prostate Cancer and Prostatic Diseases (2001) 4, 81±91 ß 2001 Nature Publishing Group All rights reserved 1365±7852/01 $15.00

www.nature.com/pcan

Prostate cancer chemoprevention agents exhibit selective activity against early stage prostate cancer cells YQ Liu1, E Kyle2, S Patel2, F Housseau3, F Hakim2, R Lieberman4, M Pins5, MV Blagosklonny2 & RC Bergan1* 1 Division of Hematology/Oncology, Department of Medicine, Northwest University Medical School and the Robert H. Lurie Cancer Center of Northwestern University, Chicago, USA; 2Medicine Branch, National Cancer Institute, National Institutes of Health, Bethesda, MD, National Institutes, National Institutes USA; 3Surgery Branch, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA; 4Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA; and 5Department of Pathology, Northwestern University Medical School and the Robert H. Lurie Cancer Center of Northwestern University, Chicago, USA

Preclinical models for the identi®cation of prostate cancer chemoprevention agents are lacking. Based upon the notion that clinically useful chemoprevention agents should exhibit selective activity against early stage disease, studies were undertaken to assess whether chemoprevention agents selectively inhibited the growth of early stage prostate cancer, as compared to late stage cancer. First, a series of cell and molecular studies were performed, which, when taken together, validated the use of a panel of prostate cell lines as a model of the different stages of carcinogenesis. Next, therapeutic responsiveness to ten different cytotoxic or chemoprevention agents was evaluated. Chemoprevention agents exhibited selective activity against normal and early transformed prostate tissue, whereas cytotoxic agents were non-speci®c. Selective activity against early versus advanced prostate cancer cells is identi®ed as a potential screening method for chemoprevention agents. Prostate Cancer and Prostatic Diseases (2001) 4, 81±91.

Keywords: prostate cancer; chemoprevention; preclinical models; cell growth; treatment

Introduction Prostate cancer is a prevalent disease which historically has been resistant to chemotherapy, and is incurable in advanced stages.1 Prostate cancer will be diagnosed in an estimated 180 400 Americans in 2000, causing death in approximately 31 900 of them, world wide these numbers are much higher.2 Though initially responsive to hormone therapy, relapse is inevitable once bony metastasis have developed. Chemotherapy has not improved overall therapeutic outcome over the past four decades.1,3 Signi®cant advances in this area will require approaches other than those historically undertaken. While most systemic therapeutic interventions are performed in late stage disease, there are a number of reasons *Correspondence: RC Bergan, Olson 8524, Division of Hematology/ Oncology, Northwestern University, 710 N. Fairbanks, Chicago, IL 60611-3008, USA. Received 9 August 2000; accepted in revised form 5 December 2000

supporting therapeutic intervention at a very early stage in the disease process, using a chemopreventive-type approach. First, recent advances in cancer biology suggest that prostate cancer develops as a result of a multi-step process, with late stage steps dependent upon initial changes.4 ± 6 Thus, therapeutic intervention at an early stage would prevent late stage disease. Advances in awareness, detection and in our understanding of early clinical stages has expanded the ability to correlate clinical stages with changes at the molecular level. Importantly, early clinical processes develop in the context of a process characterized by both distinct pathological entities, as well as one characterized by linear progression.1,6,7 The earliest pathologic change, low grade prostatic intraepithelial neoplasia (PIN), is observed in the third decade of life, while invasive cancer is observed in the sixth decade. Likewise, invasive cancer runs a well recognized course of latency within the prostate gland, followed by local invasion, seeding of local lymph nodes, and then the development of wide spread metastasis.

Identi®cation of prostate chemoprevention drugs YQ Liu et al

82

Another advantage to early therapeutic intervention relates to the fact that systemic pharmaceuticals are directed towards a ®nite number of molecular targets. Signi®cant in that the number of molecular abnormalities of any cancer, including prostate, increase with time and advances in the clinical stage. In fact, abnormalities at the molecular level are believed to be responsible for disease progression.8,9 This means that in more advanced stages of disease, therapeutic interventions must effectively reverse a much larger number of molecular aberrations than in the early stages. Finally, patients have fewer comorbid conditions at early stages of the disease, as compared to those with late stages. This is not only an important prognosticator of therapeutic responsiveness, but also de®nes a patient population more tolerant of side effects of therapy. One of the dif®culties with developing prostate cancer chemopreventive agents, however, relates to the manner by which potentially active agents are identi®ed in the preclinical setting. Currently, there is no standard methodology for the identi®cation of such agents. In part this relates to the historic dif®culty in identifying agents which have clinical activity against prostate cancer. That is, it has been dif®cult to develop meaningful preclinical models for prostate cancer.10 This problem is compounded when one is considering an early intervention/ chemoprevention strategy.11 This is because endpoints for chemoprevention trials differ from those sought in more conventional clinical trials which seek to accrue a cohort of patients with advanced disease. An effective chemoprevention agent should be selective and should have activity at early stages of the disease process. It was therefore hypothesized that chemopreventive agents should exhibit selective activity against less carcinogenic cell lines, as compared to cell lines indicative of more advanced stages of carcinogenesis. In addition, the spectrum of activity of chemopreventive agents should be distinct from that observed with conventional cytotoxic agents. A series of investigations were undertaken to test this hypothesis. Taken together current ®ndings support the notion that chemopreventive agents exert selective activity against early prostate cancer, as compared to late stage disease.

Materials and methods Materials Adriamycin, vinblastine, tamoxifen and 4-hydroxytamoxifen were obtained from Sigma Chemical Company (St. Louis, MO). Genistein, 43% and 90% preparations, were obtained from Protein Technologies International (St. Louis, MO), perillyl alcohol (PAL), perillic acid (PAC) and perillyl alcohol methylester (PALME) were obtained from McKesson BioServices (Rockville, MD). Toremifene citrate was obtained from Orion Pharma (Turku, Finland).

Cell lines and cell culture PC3, LNCaP and DU-145 cells were obtained from the American Type Culture Collection (Manassas, VA). PC3M cells are a metastatic variant of PC3 cells whose origin and characteristics have been described previously.12,13 Prostate Cancer and Prostatic Diseases

Normal prostate epithelial cells (NPECs) were obtained from Clonetics (Palo Alto, CA). Human papillomavirus (HPV) transformed normal epithelial and primary cancer cells were obtained from prostate glands as described.14 Brie¯y, normal and cancer cells from the same patient were microdissected and transformed yielding paired normal and cancer transformed cell lines. Paired lines from two patients were available: 1532NPTX (normal) and 1532CPTX (cancer), and 1542NPTX (normal) and 1542CP3TX (cancer). HPV transformed cell lines were a generous gift from S Topolian (National Cancer Institute, Bethesda, MD). Normal diploid peripheral blood mononuclear cells were harvested as previously described from normal volunteers.15,16 PC3 and PC3-M cells were grown in RPMI 1640 media (Gibco BRL, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS; Bio¯uids, Rockville, MD). DU145 cells were cultured in Dulbecco's modi®ed Eagle's medium (DMEM), supplemented with 5% FBS. NPECs were cultured according to the manufacturer. 1532NPTX, 1542NPTX, 1532CP1TX and 1542CP3TX cells were grown in keratinocyte-serum free medium, supplemented with bovine pituitary extract and human recombinant epidermal growth factor according to the manufacturer (Gibco), and with 5% FBS and 2 mM glutamine (Bio¯uids, Rockville, MD). All cells were maintained at 37 C in a humidi®ed atmosphere of 5% carbon dioxide, with biweekly media changes.

Cell cycle and DNA ploidy analysis Exponentially growing subcon¯uent cells were subjected to ¯ow cytometric analysis as previously described.17 Brie¯y, cells were ®xed in 70% ethanol at 4 C, overnight. Cells were then resuspended in PBS containing 50 kunit/ ml type III-A RNase (Sigma, St. Louis, MD) and incubated for 30 min at 37 C; DNA was then stained by the addition of propidium iodide. The intensity of nuclear staining was analyzed on a FACSCAN (Becton Dickinson, San Jose, CA). The proportion of cells in G0 ± G1, S and G2 ± M was then determined from the resultant DNA histogram using a Mod®t-LT software package (Verity Software House, Inc., Topsham, ME). DNA ploidy of prostate cells was analyzed by comparing the position of the G0 ± G1 peak of prostate cells to that of the corresponding peak for normal diploid peripheral blood mononuclear cells. In these experiments, prostate and peripheral blood mononuclear cells were analyzed separately (ie, individual cell types were analyzed separately in consecutive runs), as well as together (ie, individual cell types were mixed and run together).

Cell adhension assay Cell adhesion assays were performed as previously described, with modi®cation.18 Brie¯y, exponentially growing non-con¯uent cells were detached by treatment with trypsin/EDTA and adjusted to 105 cells/ml. 100 ml of cell suspension were then added to each well of a 96-well plate; plates were either coated with 1 mg/cm2 ®bronectin or not, as indicated. After incubation at 37 C for 10 ± 240 min, non-adherent cells were removed by washing with PBS and tissue culture media added. The number of cells


per high power ®eld were then counted in individual wells. Each time point, for each cell type, was run in replicates of six and all adhesion assays were repeated at least once at a separate time.

Immuno¯uorescence detection of ®lamentous actin Cells were stained for ®lamentous actin as previously described.18 Brie¯y, 66103 cells in 250 ml culture media were plated into each well of an eight chamber tissue culture slide (Franklin Labs, Franklin Lakes, NJ). Cells were then cultured for 24 ± 48 h under conditions of exponential growth and subcon¯uence. Cells were stained for ®lamentous actin with rhodamine-phalloidin according to the manufacturer (Molecular Probes, Eugene, OR). Brie¯y, cells were washed with PBS, ®xed in 3.7% formaldehyde in PBS for 10 min, extracted with 0.1% triton in PBS at ÿ20 C for 3 min and then stained with rhodamine-phalloidin. Slides were then mounted with Slow Fade2 according to the manufacturer (Molecular Probes, Eugene, OR). Cells were visualized on a Nikon Labophot epi¯uorescence microscope (Tokyo, Japan) equipped with an Optronics DEI-750 digital camera (Optronics Engineering, Goleta, CA) linked to a Power Macintosh 8500 workstation (Apple Computer Inc., Cupertino, CA) with Adobe Photoshop2 software (Adobe Systems Incorporated, San Jose, CA).

Assay for transforming growth factor b Cells were grown under TGFb-free conditions as described, with modi®cations, and assayed for TGFb-1 and TGFb-2 production by ELISA using TGFb-1 or TGFb-2 speci®c kits from R&D Systems (Minneapolis, MN), according to manufacturer's instructions.17 Brie¯y, for each of eight individual cell lines tested, 2.5 ± 7.56105 cells were plated into 9 cm tissue culture dishes in the presence of the standard tissue culture media for individual cell lines. 24-h after plating, cells were extensively washed and cultured in serum free media (ie, standard media for individual cell lines, minus FBS and bovine pituitary extract). After allowing cells to condition the media for 24 h, media were collected and contaminating cells removed by centrifugation. Equal amounts of media from individual cell types were used for TGFb assays. By counting the number of viable cells present in each dish, TGFb production per cell could be determined and was reported.

Western blot for c-myc, FAK, and tyrosine phosphorylated proteins Twenty-four hours after 26106 cells were plated, cells were lysed and probed for the presence of c-myc, focol adhesion kinase (FAK) or tyrosine phosphorylated protein as described.18 ± 20 Brie¯y, cells were lysed in RIPA lysis buffer (50 mM Tris, pH 7.5, 0.1% sodium dodecyl sulfate (SDS), 0.5% deoxycholic acid, 150 mM NaCl, 1% NP40) in the presence of the following protease inhibitors: 10 mg/ml aprotinin, 20 mg/ml leupeptin, and 1 mM phenylmethylsulfonyl¯uoride. Othovanadate (1 mM) was also added to the lysis buffer to inhibit endogenous protein-tyrosine phosphatase activity.21 The resultant clar-

i®ed lysates, normalized for protein, were separated on an 8% SDS polyacrylamide gel. Proteins were transferred onto 0.45 mm nitrocellulose (Schleicher & Schuell, Keene, NH) in a wet transfer cell (Bio-Rad, Hercules, CA). For detection of c-myc and FAK membranes were blocked with nonfat dry milk and probed with either c-myc mouse monoclonal antibody-containing ascites diluted 1 : 750 (clone 9E-10 hybridoma from American Type Culture Collection), or with anti-FAK monoclonal antibody (Transduction Laboratory, Lexington, KY) at 250 ng/ml. For detection of phosphotyrosine, blots were blocked with 5% bovine serum albumin (fraction V; Sigma) for 1 h at 45 C and probed with anti-phosphotyrosine antibody (clone 4G10; Upstate Biotechnology Inc., Lake Placid, NY) at 1.3 mg/ml, in blocking solution, for 1 h at room temperature. Proteins were detected with an antimouse-HRP conjugated secondary antibody (Amersham, Arlington Heights, IL) diluted 1 : 1000 ± 1 : 1200 and visualized with the ECL detection system (Amersham, Arlington Heights, IL) according to the manufacturer.

83

Detection of PSA, androgen receptor and beta-actin by RT-PCR RNA was isolated from exponentially growing, subcon¯uent cells using STAT-60 (Tel-Test B, Inc., Friendswood, TX) as described.16 For synthesis of cDNA, 1 mg of total RNA was incubated in a 20 ml reaction mixture containing 16reaction buffer, 5 mM MgCl2, 1 mM deoxynucleotide mix, 50 U RNase inhibitor, 20 U AMV reverse transcriptase (Boehringer Mannheim, Indianapolis, IN) and 1.6 mg oligo-dT primer. The reaction mixture was incubated for 10 min at 25 C, followed by 60 min at 42 C and then 5 min at 95 C. PCR detection of PSA utilized nested primers, while single primer pairs were used for the detection of the androgen receptor and beta-actin as described.15,22 ± 24 Upstream and downstream PCR primers for prostate speci®c antigen (PSA), androgen receptor and beta-actin are located on separate exons. The primer sequences are as follows: PSA outer upstream, CAC AGG CCA GGT ATT TCA GG; PSA outer downstream, CCT TGA TCC ACT TCC GGT AA; PSA inner upstream, TCC AAT GAC GTG TGT GCG CA; PSA inner downstream, GTG TAC AGG GAA GGC CTT TC; betaactin upstream, CAT TGT GAT GGA CTC CGG AGA CGG; beta-actin downstream, CAT CTC CTG CTC GAA GTC TAG AGC; androgen receptor upstream, AGC TAC TCC GGA CCT TAC G; androgen receptor downstream, GTG GTG CTG GAA GCC TCT CCT T. For PSA, 2 ml of cDNA were used in the outer primer reaction in a 50 ml reaction volume containing 1.5 mM MgCl2, 50 mM KCl, 10 mM Tris, pH 8.3, 0.3 mM each primer, 0.1 mM dNTP and 1.25 U Taq polymerase (Perkin-Elmer Cetus, Norwalk, CT). The reaction conditions were 95 C for 45 seconds, 58 C for 45 seconds, 72 C for 90 seconds, for 30 cycles, followed by an extension at 72 C for 8 min. One ml of ®rst reaction product was used for the inner primer reaction which was run under the same conditions, except the annealing temperature was 60 C; this reaction yields a 194 bp product. For the androgen receptor, 2 ml of cDNA were used in a 50 ml reaction volume containing 1.5 mM MgCl2, 16 PCR buffer, 0.15 mM each primer, 0.1 mM dNTP and 1.25 U Prostate Cancer and Prostatic Diseases


84

Taq polymerase. The reaction conditions consisted of an initial denaturation at 95 C for 2 min, followed by 94 C for 45 s, 55 C for 1 min, and 72 C for 1 min, for 30 cycles, with an extension at 72 C for 8 min, yielding a 361 bp product. Beta-actin was used as a control to detect the integrity of the RNA, and the cDNA reaction. For betaactin, 2 ml cDNA were used in a 50 ml reaction volume containing 1.5 mM MgCl2, 16 PCR buffer, 0.1 mM each primer, 0.1 mM dNTP and 1.25 U Taq polymerase. The reaction conditions consisted of an initial denaturation at 95 C for 2 min, followed by 95 C for 1 min, 58 C for 1 min and 72 C for 2 min, for 30 cycles, with an 8 min extension at 72 C, yielding a 231 bp product. In each case, 20 ml of reaction products were separated on a 1.5% agarose gel and visualized by staining with ethidium bromide. Each set of reactions was run with a positive (RNA isolated from LNCaP cells, known to contain AR and PSA), and a negative (no RNA), control. In some experiments, alpha-[32P]-ATP (3000 Ci/mmol, Amersham) was added during the PCR reaction to allow detection by autoradiography, as previously described.16 In these instances, reaction products were separated on a 10% non-denaturing polyacrylamide gel, and visualized by autoradiography.

Results Cell cycle and DNA ploidy Initial investigations sought to characterize a panel of eight prostate cell lines derived from different steps in the carcinogenic process. The ploidy of each of eight cell lines tested is shown in Table 1. All of the virally transformed cell lines (ie, 1532NPTX, 1542NPTX, 1532CP1TX and 1542CP3TX), as well as the normal prostate cell line, NPEC, displayed a diploid pattern. All of the established cell lines (ie, PC3, PC3-M and DU-145) were hyperdiploid, with a modal ploidy of 3.3, 4.0 and 2.8, respectively. Also depicted in Table 1 is the fraction of cells in S, G2 plus M phases of the cell cycle; a direct measure of growth fraction and an important indicator of growth potential. Approximately two thirds of the established cell lines were in S/G2/M phases, with PC3-M cells having the highest growth fraction, at 71%. In contrast, only 40% of virally transformed primary cells and 24% of NPECs were in S/G2/M phases. As expected, none of the peripheral blood mononuclear cells were actively cycling.

Cell adhesion and cytoskeletal architecture Drug sensitivity of prostate cell lines Cell growth inhibition assays were conducted as described previously.13,17 Brie¯y, between 1200 and 2000 cells in a ®nal volume of 100 ml were added to each well of a 96-well plate and allowed to adhere overnight. Each of 10 different drugs, in 10 different serial dilution concentrations, were then added in a volume of 100 ml to each well. After incubating for 3 days, thymidine incorporation was measured as described.13,17 Cells were harvested onto glass ®ber ®lters (Packard, Meriden, CT) with a Packard cell harvester and counted in a Matrix 9600 Microplate counter (Packard). All microtiter assays in this study were run in replicates of six and were repeated at least once at a separate time. For each cell type, cells were plated such that they were growing in an exponential and non-con¯uent manner throughout the assay period. Furthermore, for each cell line ± drug combination, drug concentrations in the assay were adjusted such that approximately half of the treated cells would be exposed to drug concentrations above the IC50 (drug concentration at which cell growth was inhibited by 50%), and half below. Finally, the concentration of drug solvent, such as DMSO, never exceeded 0.5% (v/v) and where applicable, controls were run in the presence of solvent. IC50 values were determined by ®rst plotting growth inhibition vs drug concentration and then extrapolating the concentration at which cell growth was inhibited by 50%. All IC50 values expressed in the results section are the mean /ÿ s.d. of n 2 separate experiments (each in replicates of 6), conducted at separate times. In some experiments, the growth inhibitory effect of TGFb-1 was measured and was determined as previously described.17 Brie¯y, 24 h after plating cells in a 96-well format, cells were switched to serum free media as described above for TGFb assays. Cells were then treated with various concentrations of TGFb-1 for 24 h, after which thymidine incorporation was measured. Prostate Cancer and Prostatic Diseases

Alterations in prostate cell adhesion are associated with metastatic behavior.13,18 To determine if cell growth was associated with metastatic potential, the ability of cells to adhere to a solid matrix was measured. As can be seen in Figure 1, the highly metastatic PC3-M cell variant ®rst attached more rapidly than all other cell types, then entered a phase of spontaneous cell detachment. Cytoskeletal architecture plays an important role in regulating cell adhesion. Changes in cytoskeletal structure were sought by evaluating ®lamentous actin architecture (Figure 2). As shown, compared to all other cell types studied, the actin ®laments of PC3-M metastatic variant cells, were short and disorganized. In addition, a single prominent lamellopodal complex was evident in many cells (indicative of cytoskeletal polarity), in all cell types evaluated except PC3-M (data not shown).

Transforming growth factor b In vivo, TGFb is secreted in an inactive, or latent, form and is activated through an incompletely understood mechanism.25 However, in vitro, total TGFb levels (active plus Table 1

DNA ploidy and cell cycle analysis of prostate cell lines Percentage of cells in phase of cell cycle

Cell type PBL NPEC 1532NPTX 1532CP1TX 1542CP3TX 1542NPTX PC3 PC3-M DU-145

G0/G1

S

G2/M

S G2/M

DNA ploidy

100 76 55 61 61 62 35 30 34

0 8 15 18 15 18 40 28 42

0 16 30 21 24 20 25 43 24

0 24 45 39 39 38 65 71 66

2N 2N 2N 2N 2N 2N 3.3N 4.0N 2.8N


85

Figure 1 Adhesion of prostate cells to solid matrix. Exponentially growing, subcon¯uent prostate cells were detached from tissue culture plates by treatment with trypsin/EDTA. At time 0, detached prostate cells were plated onto 96-well treated tissue culture plates. At the indicated time points, nonadherent cells were washed off, and the remaining attached cells counted. The mean number of adherent cells (n 6) was then expressed as the percentage of cells at the 240 min time point. Standard deviations were less than 10%. Similar results were obtained in a second experiment, run at a separate time (also in replicates of six).

latent forms) can be measured by performing an in vitro acid activation step.26 Under the current assay conditions, all cell lines tested, except 1542NPTX and 1532CP1TX, secreted signi®cantly higher amounts of TGFb1, as compared to normal prostate NPEC cells (Figure 3A). Only PC3 cells secreted detectable amounts of TGFb2 (Figure 3B). As in vitro response pro®les to TGFb subspecies 1 ± 3 tend to be similar across a variety of systems and as this was previously shown to be the case for growth inhibition of prostate, only growth inhibition in response to TGFb1 was measured in the current study.17,25 The IC50s for growth inhibition by TGFb1 are depicted in Table 2. No pattern of sensitivity was evident, with relative resistance (ie IC50 > 0.5 ng/ml) being observed in two of three established cell lines and in three of ®ve non-established cell lines. The two-sided t-test P-value was 0.39 for difference between established and non-established cell lines.

Phosphotyrosine, c-myc and FAK protein expression Twenty-four hours after plating, exponentially growing prostate cells were probed for either phosphotyrosine, cmyc or FAK protein. As can be seen in Figure 3, c-myc protein expression differs between cell lines, with highest levels observed in established cell lines (highest in PC3-M cells) and very low levels in NPEC, normal prostate cells. FAK protein levels differ less than those of c-myc and are highest in 1542NPTX, 1532CP1TX, PC3 and PC3-M cells and lowest in NPEC. A surprisingly conserved pattern of tyrosine phosphorylation was observed in all non-established cell lines (Figure 4). In established PC3, PC3-M and DU-145 cells, changes in phosphotyrosine expression were observed in

closely grouped areas, though the direction of change varied in individual cases. Increases in band intensity were observed for proteins of 120 ± 130 kDa, while a uniform decrease was evident in a 65 kDa band in all three established lines. Notable changes were evident in proteins of 180 ± 190 kDa; both increases and decreases in intensity being evident. A similar pattern was also observed in two of the three established cell lines (ie PC3-M and DU-145 cells) for proteins of 55 ± 60 kDa.

Detection of prostate speci®c antigen (PSA) and androgen receptor (AR) transcripts by reverse transcriptase-polymerase chain reaction (RT-PCR) LNCaP cells are known to express PSA and AR and were therefore used as a positive control.27 RNA was isolated from each cell line and subjected to cDNA synthesis and PCR ampli®cation for either PSA or AR. In each instance, beta-actin was also ampli®ed to verify RNA integrity and cDNA synthesis. As can be seen in Figure 5, PSA was detected in the 1532NPTX, 1532CPTX, PC3, LNCaP and DU-145 cell lines; AR was detected in the NPEC, 1542NPTX, 1542CP3TX, DU-145, LNCaP and PC3 cell lines. Note, the weak signal evident in the current ®gure for AR in 1532NPTX and PC3-M cells did not become more prominent with ®ve more cycles of ampli®cation and so was scored as negative, while that for DU-145 was consistently positive upon further ampli®cation and so was scored as positive. For AR, weak bands detected on ethidium bromide staining were con®rmed as positive by incorporating radiolabeled nucleotide during PCR and subsequent autoradiography. For PSA, no additional bands were detected by radiolabeling (data not shown). Thus for AR and PSA, autoradiography and ethidium bromide detection methods, respectively, are depicted. Prostate Cancer and Prostatic Diseases


86

Figure 2 Cytoskeletal organization in prostate cells. After allowing cells to adhere to tissue culture treated eight chamber slides for 48 h, cells were ®xed and stained for ®lamentous actin with rhodamine-phalloidin, as described in the Methods. Representative ¯uorescent photomicrographs are shown (10006). No ¯uorescent signal was detected in control cells not stained with rhodamine-phalloidin (data not shown).

Drug sensitivity A panel of 10 drugs was selected and the sensitivity of each cell line to each drug was measured. The IC50s for all eight cell lines, and all 10 drugs tested are shown in Table 3. In Table 4 the mean IC50 value for each drug is shown for: (1) all established cancer cell lines; (2) all HPV transformed cancer lines; (3) all cancer cell lines; (4) HPV transformed normal lines; (5) all normal cell lines. Also depicted are the IC50 ratios for the following: (1) HPV cancer to HPV normal cell lines; (2) all cancer to all normal cell lines. These ratios provide measures of relative sensitivity between cancer cells and normal cells. Prostate Cancer and Prostatic Diseases

Ratios above 1.0 indicate higher drug concentrations are required to inhibit the growth of cancer cells relative to that of non-cancer cells. Conversely, ratios below 1.0 indicate cancer cells are more sensitive to a particular agent than are normal cells. When all cancer cells are compared to all normal cells, ratios above 2.0 (representing a two-fold difference in sensitivity) were seen with PAC and with both preparations of genistein (both are drugs currently being evaluated as prostate cancer chemopreventive agents in human trials). Values between 1.5 and 2.0 were seen with PAL, vinblastine, 4-hydroxytamoxifen and toremefene. Ratios below 1.0 were only observed with PALME. A similar trend was observed


87

Figure 4 Protein expression patterns in prostate cells. 24 h after plating non-con¯uent, exponentially growing prostate cells were harvested. After separation of equal amounts of protein from resultant cell lysates by SDS PAGE under reducing conditions, proteins were transferred onto nitrocellulose and probed for either c-myc, FAK or phosphotyrosine (Ptyr). Cell lines evaluated are: lane 1, NPEC; lane 2, 1532NPTX; lane 3, 1532CP1TX; lane 4, 1542NPTX; lane 5, 1542CP3TX; lane 6, PC3; lane 7, PC3-M; lane 8, DU145.

when HPV transformed cancer cells were compared to HPV transformed normal cells; in particular, PAC and genistein both had ratios above 2.0.

Discussion Figure 3 Secretion of transforming growth factor beta (TGFb) by prostate cells. 24-h after plating prostate cells, media were changed to serum free media, as described in the methods. 24 h later the concentration of TGFb1 (A) and TGFb2 (B) in conditioned media was measured by ELISA. By counting the number of viable cells present after 24 h incubation in serum free media, the amount of TGFb produced per cell could be determined and is depicted. Expressed values are the mean s.d. (n 2) of a single experiment; similar results were obtained in a separate experiment (also run in replicates of 2).

Table 2 Transforming growth factor b1 IC50 values for growth inhibition of prostate cell lines IC50, ng/ml Cell line NPEC 1532CPTX 1532NPTX 1542CP3TX 1542NPTX DU-145 PC3 PC3-M

Mean (n 2)

s.d.

1.0 0.62 0.12 0.18 0.89 1.1 1.4 0.19

0.17 0.014 0.014 0.14 0.16 0.14 0.85 0.04

Relatively few comparative studies have been conducted which have sought to characterize factors associated with therapeutic responsiveness to chemotherapy in prostate cancer.1 Thus, there was little prior experience which could be drawn upon to guide these investigations which sought to evaluate therapeutic responsiveness in early stage disease. An emerging understanding of prostate cancer biology guided the selection of cell lines, cell and molecular parameters and therapeutic agents used in this study. Their selection sought to provide an accurate representation of prostate carcinogenesis, pertinent cell and molecular parameters and a panel of therapeutic agents inclusive of both prostate speci®c and non-speci®c agents. The initial goal was to construct and characterize a panel of prostate cells lines which together provided a representative selection of cell lines from initial to ®nal stages of prostatic carcinogenesis. Choosing cell lines derived from tissues representative of different stages of carcinogenic progression was inherently logical. Only recently have normal, non-transformed, prostate epithelial cells (NPECs) become widely available.28 Though limited by eventual senescence, when used appropriately (ie, only using cells from a single batch of frozen cells, and Prostate Cancer and Prostatic Diseases


88

Figure 5 Expression of PSA and androgen receptor (AR) in prostate cells. RNA was isolated from exponentially growing, non-con¯uent prostate cells and reverse transcribed. Resultant cDNA was ampli®ed with exon spanning primers speci®c for either PSA, AR or b-actin. Depicted are photomicrographs of ethidium bromide stained agarose gels for PSA and bactin (194 and 231 bp products, respectively), and an autoradiograph of radiolabeled PCR products separated on a non-denaturing polyacrylamide gel for AR (361 bp). Lanes are: S, standards; lane 1, NPEC; lane 2, 1532NPTX; lane 3, 1532CP1TX; lane 4, 1542NPTX; lane 5, 1542CP3TX; lane 6, PC3; lane 7, PC3-M; lane 8, DU145; lane 9, LNCaP; N, no RNA negative control.

only maintaining them for less than 10 doublings once thawed), these cells can serve as valuable tools. While immortalized cells derived by viral transformation have potential pitfalls due to the continued presence of viral elements, viral transformation has been an invaluable tool for obtaining primary prostate cells.14,29 ± 32 In the current study, paired normal and primary cancer cells obtained from the same patient serve to eliminate potential variables.14 Finally, established cell lines were included as they are widely available and as they have been extensively characterized in the literature.33,34 The PC3-M cell line is a well-characterized variant of the PC3 cell line and represents a more rapidly growing metastatic variant of the parental cells.12,13,18 While widely reported in the literature, the LNCaP cell line was not subjected to phenotypic analysis in this study. This is because the current focus was not upon identi®cation of androgen receptor binding agents and there was concern that tamoxifen may exhibit promiscuous receptor activation, as has been reported with a variety of agents. However, before cell lines which are maintained in culture could be considered as representative of their respective in vivo clinical counterparts, it was necessary to characterize them phenotypically. Taken together, phenotypic analysis supports the notion that the selected panel of cell lines in fact provides a representative model of the spectrum of prostatic carcinogenesis. Phenotypic analysis focused upon characteristics known to be associated with prostate cancer progression. Particular emphasis was placed upon elements whose functional role could clearly be associated with neoplastic progression and whose measurement in an in vitro model could readily be associated with a similar function in vivo. Examples of this include: cell cycle analysis, DNA ploidy, cell adhesion, c-myc expression and FAK expression.13,35,36 In other cases phenotypic characteristics either had a less clear role in carcinogenesis, or had previously recognized problems in associating in vitro expression with in vivo expression. In these cases, phenotypic characteristics were chosen because of their close association with prostate cancer. Examples include: transforming growth factor b (TGFb), androgen receptor, tyrosine phosphorylation and PSA.27,37 ± 46

Table 3 Three day IC50 values for growth inhibition of prostate cell lines Cell lines PC3-M

PC3

DU-145

1542CP3TX

1532CPTX

1542NPTX

1532NPTX

NPEC

Drug

Mean

s.d.

Mean

s.d.

Mean

s.d.

Mean

s.d.

Mean

s.d.

Mean

s.d.

Mean

s.d.

Mean

s.d.

PAL (mM) PALME (mM) PAC (mM) Adriamycin (nM) VBN (ng/ml) 90% Geni (mM) 43% Geni (mM) Tamoxifen (mM) 4OHTAM (mM) Toremifene (mM)

1.17 0.75 1.33 17.5 0.61 20 34 1.6 5.65 4.2

0.06 0.91 0.32 1.0 0.06 0.0 5.0 0.7 0.21 1.4

0.23 0.58 2.25 38.5 0.53 10 18 8.5 6.5 8.5

0.01 0.0 0.07 6.36 0.04 3.5 2.0 1.8 0.71 0.0

0.7 0.22 0.52 4.0 0.53 21 32 10.5 5.25 3.0

0.03 0.0 0.06 0.0 0.0 10 10 0.71 1.77 0.8

0.15 0.57 1.35 17.5 0.21 34.67 31.67 5.07 5.13 10.67

0.04 0.0 0.21 10.61 0.02 14.74 7.02 0.9 1.0 4.04

0.2 0.31 1.1 33 0.35 16 18 7.6 6.0 6.0

0.09 0.09 0.14 7.0 0.11 2.0 2.0 4.81 0.0 2.0

0.4 1.1 0.28 11.33 0.22 12.33 15.17 7.13 3.4 1.8

0.0 0.08 0.11 1.53 0.02 5.13 8.55 2.02 1.78 1.6

0.13 0.13 0.4 12 0.28 7.0 8.0 5.1 3.5 7.0

0.03 0.02 0.06 6.0 0.02 3.5 0.0 1.27 2.12 1.0

0.3 0.51 0.34 23 0.27 3.33 6.0 4.25 1.85 2.9

0.17 0.51 0.04 3.0 0.23 0.58 0.0 0.35 0.92 2.0

Prostate Cancer and Prostatic Diseases


89

Table 4 IC50 averages and ratios for growth inhibition of prostate cell lines Cell lines a

Established lines

b

HPV ca lines

All ca lines

c

HPV nl linesd All nl linese

Drug

Mean

s.d.

Mean

s.d.

Mean

s.d.

Mean

s.d.

Mean

s.d.

HPV ca/HPV nl ratio

All ca/all nl ratio

All est/all nl ratio

PAL (mM) PALME (mM) PAC (mM) Adriamycin (nM) VBN (ng/ml) 90% Geni (mM) 43% Geni (mM) Tamoxifen (mM) 4OHTAM (mM) Toremifene (mM)

0.7 0.52 1.37 20 0.56 17 28 6.87 5.8 5.23

0.47 0.27 0.87 17.39 0.05 6.08 8.72 4.67 0.64 2.89

0.17 0.44 1.23 25.25 0.28 25.33 24.83 6.33 5.57 8.33

0.04 0.18 0.18 10.96 0.1 13.2 9.66 1.79 0.61 3.3

0.49 0.49 1.31 22.1 0.45 20.33 26.73 6.65 5.71 6.47

0.44 0.22 0.62 13.76 0.16 9.1 8.02 3.43 0.56 3.13

0.27 0.62 0.34 11.67 0.25 9.67 11.58 6.11 3.45 4.4

0.19 0.69 0.09 0.47 0.04 3.77 5.07 1.43 0.07 3.68

0.28 0.58 0.34 15.44 0.26 7.56 9.72 5.49 2.92 3.9

0.14 0.49 0.06 6.55 0.03 4.53 4.82 1.48 0.93 2.74

0.65 0.72 3.63 2.16 1.13 2.62 2.14 1.04 1.61 1.89

1.77 0.84 3.87 1.43 1.75 2.69 2.75 1.21 1.96 1.66

2.53 0.89 4.03 1.29 2.18 2.25 2.88 1.25 1.99 1.34

a PC3, PC3-M, and DU-145 cell lines. b1532CP1TX and 1542CP3TX. cPC3, PC3-M, DU-145, 1532CP1TX and 1542CP3TX. d1532NPTX and 1542NPTX. eNPEC, 1532NPTX and 1542NPTX.

Increased growth fraction (ie, percentage of cells in cell cycle), as well as increased DNA ploidy, are well associated with neoplastic progression in a wide array of cancers, including prostate.47 The ®ndings of high growth fraction and DNA ploidy in the current study in cell lines derived from late stage tissue, and a low growth fraction with normal DNA ploidy in normal prostate epithelial cells, is thus not surprising. Cell adhesion is a critical factor which regulates metastatic behavior; to metastasize, cells must ®rst detach.48 The PC3-M cell line is a highly aggressive metastatic variant and it alone exhibited a unique cell adhesion pro®le. Speci®cally, the spontaneous detachment phase seen with PC3-M cells, after initial rapid attachment, is re¯ective of rapid cell detachment and attachment inherent in metastatic behavior. Both changes in c-myc and in FAK expression in prostate cancer have been reported previously.13,35,36 The current study con®rms this and demonstrates changes even in transformed prostate cells. Not unexpectedly, c-myc expression directly correlated with the percentage of cells actively cycling. Interestingly, increased FAK expression was observed in early transformed cell lines, suggesting that changes in FAK may occur early in the process of prostatic carcinogenesis. While this notion is supported by the fact that FAK plays an important role in cell adhesion, changes in FAK expression did not correlate well with changes in cell adhesion. Thus, factors other than FAK appear to be necessary to induce effects upon cell adhesion, and ultimately, metastatic behavior. While a variety of growth factors in¯uence the growth of prostate cells, TGFb occupies a prominent growth regulatory role in prostate and in prostatic carcinogenesis.38 ± 44 Growth inhibitory to normal prostate, early prostate cancer cells down regulate TGFb receptor and lose sensitivity to TGFb, while TGFb appears to stimulate the growth of metastatic cells.38 ± 40 In addition, TGFb expression patterns appear to change as a function of neoplastic progression; however, there is signi®cant heterogeneity in TGFb expression patterns.41 ± 44 In the current study there was no clear pattern to either TGFb growth sensitivity, or production, as a function of neoplastic progression. Natural differences may exist

between individual cell types masking broader changes associated with progression in the context of a relatively small sample size. Tyrosine phosphorylated proteins represent a signi®cant proportion of both oncogenes as well as growth factor related signaling pathways. Both are critical elements with respect to prostatic carcinogenesis.37 It is tempting to assign causation between the uniform pattern of tyrosine phosphorylation observed in non-established cells with their lower growth fraction, and the clustered changes in protein phosphorylation observed in established cells with their higher growth fraction. More studies will need to be conducted in order to determine causation. The current ®ndings are provocative however, and warrant further investigation into the role of these proteins in prostate growth regulation. While a signi®cant body of work has been published on both PSA and the AR, their role in advanced disease is still evolving (for reviews, see references 45 and 46). However, as both AR and PSA are relatively speci®c to prostate cancer, and intimately linked to that disease, they were evaluated in the current study. In this study, there was no association between the expression of AR and PSA: co-expression was seen in three of nine cell lines tested (including the LNCaP cell line), simultaneous lack of expression in one, and differential expression in ®ve. Evaluation of androgen responsiveness is complex, can change with a change in growth conditions and is beyond the scope of the present study.27 Having con®rmed the validity of the model system, studies then sought to evaluate potential differential affects upon normal or early cell lines, as compared to those representative of advanced disease. Adriamycin and vinblastine both possess signi®cant single agent activity in a wide array of neoplasms, including prostate and thus represent non-selective globally active agents.49,50 The remaining agents were included as they are associated with activity which is relatively unique to prostate: (1) perillyl alcohol has exhibited activity against prostate cells in preclinical studies and is currently being evaluated in phase I prostate chemoprevention studies; its active metabolite is perillic acid, with perillyl alcohol methylester being a close chemical derivative of Prostate Cancer and Prostatic Diseases


90

perillyl alcohol.51 ± 53 (2) Genistein possesses antimetastatic activity in prostate, is also growth inhibitory, and is currently undergoing phase I testing as a potential prostate chemopreventive agent.13,18,54 (3) Tamoxifen, at high concentrations (ie, micromolar range), inhibits protein kinase C (PKC), activating prostate TGFb signaling pathways and leading to growth inhibition, and when given to humans in the form of high dose tamoxifen therapy, is associated with clinical activity in advanced prostate cancer.17,55 Toremifene and 4-hydroxy tamoxifen, are chemically related to tamoxifen.17 It is important to note that the cytotoxic agents did not exhibit selective activity against cancer cell lines. This was even the case if one only considered established cancer cell lines, whose marked increase in growth fraction would be expected to sensitize cells to cytotoxic agents. This pattern was present for standard forms of cytotoxic agents (adriamycin, vinblastine) and the experimental agent tamoxifen and its derivatives. In contrast, the chemopreventive agents genistein and perrylic acid were more active against normal prostate cell lines than cancer cell lines. The ultimate reasons for this can only be determined by further investigation. However, both classes of agents exhibit prostate speci®c activity and both are being developed as potential prostate chemopreventive agents based upon this activity.18,51,54 Note, that even though the drug administered to humans is perillyl alcohol, its active metabolite is perillic acid; perillyl alcohol methylester is chemically related to perillyl alcohol. The phenotype of in vivo normal prostate epithelial cells is more closely represented by the normal prostate cell lines than by cancer cell lines. Current ®ndings support the hypothesis that selective activity against normal prostate cells, as compared to prostate cancer cells, is an indicator of the chemopreventive potential of an agent. Additional studies will need to be conducted in order to con®rm this hypothesis. Selective activity against early vs advanced prostate cancer cells is identi®ed as a potential screening method for chemoprevention agents.

References 1 Oesterling JE, Fuks Z, Lee CT, Scher HI. Cancer of the prostate. In: DeVita VT, Hellman S, Rosenberg SA (eds). Cancer: Principals and Practice of Oncology, Vol 1. Lippincott-Raven: New York, 1997, pp 1322 ± 1385. 2 Greenlee RT, Murray T, Bolden S, Wingo PA. Cancer Statistics, 2000. CA Cancer J Clin 2000; 50: 7 ± 33. 3 Oh WK, Kantoff PW. Management of hormone refractory prostate cancer: current standards and future prospects [see comments]. J Urol 1998; 160: 1220 ± 1229. 4 Lara PN Jr, Kung HJ, Gumerlock PH, Meyers FJ. Molecular biology of prostate carcinogenesis. Crit Rev Oncol Hematol 1999; 32: 197 ± 208. 5 Sciavolino PJ, Abate-Shen C. Molecular biology of prostate development and prostate cancer. Ann Med 1998; 30: 357 ± 368. 6 Carter HB, Piantadosi S, Isaacs JT. Clinical evidence for and implications of the multistep development of prostate cancer. J Urol 1990; 143: 742 ± 746. 7 Bostwick DG. Prostatic intraepithelial neoplasia is a risk factor for cancer. Semin Urol Oncol 1999; 17: 187 ± 198. 8 Vogelstein B, Kinzler KW. The multistep nature of cancer. Trends Genet 1993; 9: 138 ± 141. Prostate Cancer and Prostatic Diseases

9 Bishop JM. Molecular themes in oncogenesis. Cell 1991; 64: 235 ± 248. 10 Royai R, Lange PH, Vessella R. Preclinical models of prostate cancer. Semin Oncol 1996; 23: 35 ± 40. 11 Kelloff GJ et al. Progress in cancer chemoprevention. Ann N Y Acad Sci 1999; 889: 1 ± 13. 12 Kozlowski JM et al. Metastatic behavior of human tumor cell lines grown in the nude mouse. Cancer Res 1984; 44: 3522 ± 3529. 13 Kyle E et al. Genistein-induced apoptosis of prostate cancer cells is preceded by a speci®c decrease in focal adhesion kinase activity. Mol Pharmacol 1997; 51: 193 ± 200. 14 Bright RK et al. Generation and genetic characterization of immortal human prostate epithelial cell lines derived from primary cancer specimens. Cancer Res 1997; 57: 995 ± 1002. 15 Bergan R et al. Electroporation of synthetic oligodeoxynucleotides: a novel technique for ex vivo bone marrow purging. Blood 1996; 88: 731 ± 741. 16 Schwartz GN et al. Growth inhibition of chronic myelogenous leukemia cells by ODN-1, an aptameric inhibitor of p210bcr-abl tyrosine kinase activity. Antisense Nucleic Acid Drug Dev 1998; 8: 329 ± 339. 17 Rohlff C et al. Growth inhibition of prostate cancer cells by tamoxifen is associated with protein kinase C inhibition and induction of p21 waf1/cip1. Prostate 1998; 37: 51 ± 59. 18 Bergan R et al. Genistein-stimulated adherence of prostate cancer cells is associated with the binding of focal adhesion kinase to beta-1-integrin. Clin Exp Metastasis 1996; 14: 389 ± 398. 19 Bergan R, Connell Y, Fahmy B, Neckers L. Electroporation enhances c-myc antisense oligodeoxynucleotide ef®cacy. Nucleic Acids Res 1993; 21: 3567 ± 3573. 20 Bergan RC, Kyle E, Connell Y, Neckers L. Inhibition of proteintyrosine kinase activity in intact cells by the aptameric action of oligodeoxynucleotides. Antisense Res Dev 1995; 5: 33 ± 38. 21 Bergan R et al. Aptameric inhibition of p210bcr-abl tyrosine kinase autophosphorylation by oligodeoxynucleotides of de®ned sequence and backbone structure. Nucleic Acids Res 1994; 22: 2150 ± 2154. 22 Dakhama A, Macek V, Hogg JC, Hegele RG. Ampli®cation of human beta-actin gene by the reverse transcriptase-polymerase chain reaction: implications for assessment of RNA from formalin-®xed, paraf®n-embedded material. J Histochem Cytochem 1996; 44: 1205 ± 1207. 23 Jaakkola S et al. Detection of prostatic cells in peripheral blood: correlation with serum concentrations of prostate-speci®c antigen [see comments]. Clin Chem 1995; 41: 182 ± 186. 24 Culig Z et al. DNA sequence of the androgen receptor in prostatic tumor cell lines and tissue specimens assessed by means of the polymerase chain reaction. Prostate 1993; 22: 11 ± 22. 25 Roberts AB. Molecular and cell biology of TGF-beta. Miner Electrolyte Metab 1998; 24: 111 ± 119. 26 Sporn MB, Roberts AB, Wake®eld LM, de Crombrugghe B. Some recent advances in the chemistry and biology of transforming growth factor-beta. J Cell Biol 1987; 105: 1039 ± 1045. 27 Webber MM, Bello D, Quader S. Immortalized and tumorigenic adult human prostatic epithelial cell lines: characteristics and applications. Part I. Cell markers and immortalized nontumorigenic cell lines. Prostate 1996; 29: 386 ± 394. 28 Peehl DM. Culture of human prostatic epithelial cells. In: RA Freshney (ed). Culture of Epithelial Cells (Culture of Specialized Cells). John Wiley & Sons: New York, 1992, pp 159 ± 180. 29 Webber MM et al. Prostate speci®c antigen and androgen receptor induction and characterization of an immortalized adult human prostatic epithelial cell line, Carcinogenesis 1996; 17: 1641 ± 1646. 30 Bello D et al. Androgen responsive adult human prostatic epithelial cell lines immortalized by human papillomavirus 18. Carcinogenesis 1997; 18: 1215 ± 1223. 31 Cussenot O et al. Immortalization of human adult normal prostatic epithelial cells by liposomes containing large T-SV40 gene. J Urol 1991; 146: 881 ± 886. 32 Weijerman PC et al. Lipofection-mediated immortalization of human prostatic epithelial cells of normal and malignant origin using human papillomavirus type 18 DNA. Cancer Res 1994; 54: 5579 ± 5583.


33 Mickey DD et al. Heterotransplantation of a human prostatic adenocarcinoma cell line in nude mice. Cancer Res 1977; 37: 4049 ± 4058. 34 Kaighn ME et al. Establishment and characterization of a human prostatic carcinoma cell line (PC-3). Invest Urol 1979; 17: 16 ± 23. 35 Jenkins RB, Qian J, Lieber MM, Bostwick DG. Detection of c-myc oncogene ampli®cation and chromosomal anomalies in metastatic prostatic carcinoma by ¯uorescence in situ hybridization. Cancer Res 1997; 57: 524 ± 531. 36 Tremblay L, Aprikian A, Chevalier S. Expression of focal adhesion kinase in prostate cancer. Proc Am Ass Cancer Res 1995; 36: 64. 37 Webber MM, Bello D, Quader S. Immortalized and tumorigenic adult human prostatic epithelial cell lines: characteristics and applications. Part 3. Oncogenes, suppressor genes, and applications. Prostate 1997; 30: 136 ± 142. 38 Kim IY et al. Loss of expression of transforming growth factor b type I and type II receptors correlates with tumor grade in human prostate cancer tissues. Clinical Cancer Res 1996; 2: 1255 ± 1261. 39 Sehgal I, Baley PA, Thompson TC. Transforming growth factor betal stimulates contrasting responses in metastatic vs primary mouse prostate cancer-derived cell lines in vitro. Cancer Res 1996; 56: 3359 ± 3365. 40 Wilding G et al. Differential effects of transforming growth factor beta on human prostate cancer cells in vitro. Mol Cell Endocrinol 1989; 62: 79 ± 87. 41 Steiner MS, Zhou ZZ, Tonb DC, Barrack ER. Expression of transforming growth factor-beta 1 in prostate cancer. Endocrinology 1994; 135: 2240 ± 2247. 42 Eastham JA et al. Transforming growth factor-beta 1: comparative immunohistochemical localization in human primary and metastatic prostate cancer. Lab Invest 1995; 73: 628 ± 635. 43 Tu H, Jacobs SC, Borkowski A, Kyprianou N. Incidence of apoptosis and cell proliferation in prostate cancer: relationship with TGF-beta l and bcl-2 expressing. Int J Cancer 1996; 69: 357 ± 363. 44 Perry KT, Anthony CT, Steiner MS. Immunohistochemical localization of TGF beta 1, TGF beta 2, and TGF beta 3 in normal and malignant human prostate. Prostate 1997; 33: 133 ± 140.

45 Ablin RJ. A retrospective and prospective overview of prostate-speci®c antigen. J Cancer Res Clin Oncol 1997; 123: 583 ± 594. 46 Culig Z et al. Expression, structure, and function of androgen receptor in advanced prostatic carcinoma. Prostate 1998; 35: 63 ± 70. 47 Kastan MB. Molecular biology of cancer: the cell cycle. In: DeVita VT, Hellman S, Rosenberg SA (eds). Cancer: Principals and Practice of Oncology, Vol 1. Lippincott-Raven: New York, 1997, pp 121 ± 134. 48 Fidler IJ. Molecular biology of cancer: invasion and metastasis. In: DeVita VT, Hellman S, Rosenberg SA (eds). Cancer: Principals and Practice of Oncology, Vol 1. Lippincott-Raven Publishers: Philadelphia, 1997, pp 135 ± 152. 49 Bender RA, Hamel E, Hande KR. Plant alkaloids. In: DeVita VT, Hellman S, Rosenberg SA (eds). Cancer Chemotherapy: Principles and Practice. JB Lippincott: Philadelphia, 1990, pp 253 ± 275. 50 Myers CE, Chabner BA. Anthracyclines. In: DeVita VT, Hellman S, Rosenberg SA (eds). Cancer Chemotherapy: Principles and Practice. JB Lippincott: Philadelphia, 1990, pp 357 ± 381. 51 NIH Development. Clinical development plan: 1-perillyl alcohol. J Cell Biochem 1996; 26: 137 ± 148. 52 Stewart CF, Ratain MJ. Topoisomerase interactive agents. In: DeVita VT, Hellman S, Rosenberg SA (eds). Cancer: Principals and Practice of Oncology, Vol 1. Lippincott-Raven: New York, 1997, pp 452 ± 467. 53 Rowinsky EK, Donehower RC. Antimicrotubule agents. In: DeVita VT, Hellman S, Rosenberg SA (eds). Cancer: Principals and Practice of Oncology, Vol 1. Lippincott-Raven: New York, 1997, pp 467 ± 482. 54 Steele VE, Pereira MA, Sigman CC, Kelloff GJ. Cancer chemoprevention agent development strategies for genistein. J Nutr 1995; 125(Suppl 3): 713S ± 716S. 55 Bergan RC et al. A phase II study of high dose tamoxifen in patients with hormone refractory prostate cancer. Clinical Cancer Research 1999; 5: 2366 ± 2373.

91

Prostate Cancer and Prostatic Diseases