The mechanisms of action of the anticancer agent perillyl alcohol (POH), presently in Phase II clinical trials, were investigated in advanced rat mammary carcinomas. Gross and ultrastructural morphology of POH-mediated tumor regression indicated that apoptosis accounted for the marked reduction in the epithelial compartment. Characterization of cell growth and death indices revealed that apoptosis was induced within 48 h of chemotherapy, before the induction of cytostasis. RNA expression studies, based on a multiplexed-nuclease protection assay, demonstrated that cell cycle- and apoptosis-related genes were differentially expressed within 48 h of POH treatment; p21Cip1/WAF1, bax, bad, and annexin I were induced; cyclin E and cyclin-dependent kinase 2 were repressed; and bcl-2 and p53 were unchanged. Next, a potential role for transforming growth factor β (TGF-β) signaling in POH-mediated carcinoma regression was explored. RNA expression studies, again based on a multiplexed-nuclease protection assay, showed that TGF-β-related genes were induced and temporally regulated during POH treatment: (a) c-jun and c-fos were transiently induced within 12 h of chemotherapy; (b) TGF-β1 was induced within 24 h of chemotherapy; (c) the mannose 6-phosphate/insulin-like growth factor II receptor and the TGF-β type I and II receptors were induced within 48 h of chemotherapy; and (d) smad3 was induced during active carcinoma regression. In situ protein expression studies, based on fluorescence-immunohistochemistry in concert with confocal microscopy, confirmed up-regulation and demonstrated colocalization of TGF-β1, the mannose 6-phosphate/insulin-like growth factor II receptor, the TGF-β type I and II receptors, and Smad2/Smad3 in epithelial cells. Nuclear localization of Smad2/Smad3 indicated that the TGF-β signaling pathway was activated in regressing carcinomas. Subpopulations of Smad2/Smad3-positive and apoptotic nuclei colocalized, indicating a role for Smads in apoptosis. Thus, Smads may serve as a potential biomarker for anticancer activity. Importantly, none of the POH-mediated anticancer activities were observed in normal mammary gland.
The naturally-occurring monoterpenes LIM3 and POH have proven to be effective chemopreventive and chemotherapeutic agents in several rodent organ-specific tumor models (1, 2). Our laboratory has extensively characterized their anticancer activities in DMBA- and N-methyl-N-nitrosourea-induced rat mammary carcinoma models (3, 4, 5). Dietary administration of 10% LIM or 2% POH caused regression of ∼84% of advanced DMBA- or N-methyl-N-nitrosourea-induced rat mammary carcinomas, where ∼60% of the carcinomas completely regressed. Furthermore, the only toxicity observed during treatment was initial weight loss due to food aversion (6, 7). Because of the high efficacy and low toxicity of monoterpenes during preclinical development, evaluation of LIM and POH has been conducted in Phase I clinical trials in England (8) and the United States (9, 10). Favorable toxicity profiles and objective responses were documented with both monoterpenes. Multiple Phase II clinical trials are now in progress.4
Alterations in gene expression associated with monoterpene-mediated carcinoma regression were identified by screening 10% LIM-treated, actively regressing advanced rat mammary carcinomas using subtractive display (11). Several identified cDNAs were consistent with a differentiation/remodeling process of tumor regression. The subtractive display screen identified M6P/IGF2R and TβIIR as induced genes. Similarly, in a rat liver tumor model, POH induced expression of M6P/IGF2R and TβIR, TβIIR, and TGF-β type III receptor (12). The subtractive display result and our previous finding that TGF-β1 and M6P/IGF2R expression were up-regulated in LIM-treated regressing mammary carcinomas (13) suggest the TGF-β signaling pathway is involved in mammary carcinoma regression.
Before TGF-β initiates its signaling cascade, its inactive form, LAP-TGF-β1, must be proteolytically activated. M6P/IGF2R, a tumor suppressor gene in human breast cancer (14) and hepatocellular carcinomas (15), facilitates LAP-TGF-β1 activation, as well as targets the mammary mitogen IGF2 to lysosomes for degradation (16). TGF-β initiates its signaling pathway by facilitating oligomerization and activation of transmembrane serine/threonine kinase receptors. An activated signaling complex forms by TGF-β1 binding to TβIIR, followed by recruitment of TβIR into a hetero-oligomeric complex. TβIIR exhibits intrinsic kinase activity, resulting in phosphorylation of TβIR, which then phosphorylates receptor-bound Smad2 and Smad3. Phorsphorylated Smad2 and Smad3 form hetero-oligomeric complexes with non-receptor-bound phosphorylated Smad4. These Smad complexes translocate to the nucleus to regulate transcription of specific genes through direct or indirect binding with DNA and other transcription factors (17, 18). In addition to signaling through Smads, TGF-β activates Ras and several members of the MAPK superfamily, including extracellular signal-regulated kinases and SAPK/JNK (19). Activation of these signaling components occurs within minutes of TGF-β treatment and is blocked by a dominant-negative Ras. Furthermore, dominant-negative mutants in the SAPK/JNK pathway, including MAPK kinase kinase 1, MAPK kinase 4, JNK1, and c-Jun abrogate TGF-β signaling (20).
One downstream event consistent with TGF-β signaling, relevant to carcinoma regression, is induction of cytostasis. TGF-β is a potent inhibitor of mammary gland growth (21) and induces cytostasis through inhibition of G1 cyclin-CDK complex activities (22). The ability of TGF-β to inhibit activity of G1 cyclin-CDK complexes is partially due to transcriptional induction of CDK inhibitor p21Cip1/WAF1 (23), which subsequently binds and regulates cyclin E-CDK2 complexes. In mammary cell culture, POH-mediated inhibition of cyclin-E-dependent CDK2 activity has been associated with an increase in p21Cip1/WAF1-cyclin E-CDK2 complex formation (24).
A second downstream event consistent with TGF-β signaling relevant to carcinoma regression is induction of apoptosis. TGF-β-treated mammary cells undergo apoptosis, and administration of antisense TβIIR oligomers blocks apoptosis (25). Furthermore, induction of TGF-β expression is associated with postlactational mammary gland involution, a large scale tissue remodeling process whereby apoptosis mediates the reabsorption of the bulk of the epithelium (26). In LIM-treated regressing mammary carcinomas, the subtractive display screen identified lipocortin 1, also termed annexin I (11), a marker of apoptosis in the involuting mammary gland (27). Furthermore, POH-treated liver tumors (12), pancreatic cells (28, 29), and colon tumors (30) exhibit increased levels of apoptotic cells (28).
We explore POH-mediated activation of the TGF-β signal transduction pathway and its downstream effects. Characterization of POH-treated regressing mammary carcinomas includes a study of morphologic remodeling events during tumor regression, determination of cell proliferation and apoptotic indices, RNA expression analysis of cell cycle- and apoptosis-related genes, and RNA and in situ protein expression analyses of TGF-β signaling components.
MATERIALS AND METHODS
Advanced Mammary Carcinoma Generation.
At 50–55 days of age, virgin female Wistar Furth rats (Harlan Sprague Dawley, Madison, WI) were administered a single dose of 50 mg/kg DMBA (Eastman Kodak, Rochester, NY) in sesame oil by gastric intubation. The rats were provided Teklad Lab Blox chow and acidified water ad libitum, unless otherwise noted. All rats were palpated biweekly until advanced mammary tumors (≥10 mm in diameter) developed; subsequently, the rats were randomized to the acute or chronic treatment protocols described below. Rats who did not develop advanced mammary tumors within 15 weeks after DMBA administration were euthanized. All animal use was in compliance with NIH guidelines for humane care and was approved by the University of Wisconsin-Madison Medical Center Animal Use Committee.
Acute Treatment Protocol.
Rats in the acute treatment group were randomly paired and assigned to treatment timecourse groups of 1 h, 4 h, 12 h, 24 h, or 48 h and, subsequently, to POH or CON treatment groups. Paired rats were initially administered 0.1 g/kg POH (>96% pure by gas-chromatographic analysis; Aldrich, Milwaukee, WI) in sesame oil or vehicle alone (CON treatment) via gastric intubation. After initial administration of POH or vehicle alone, paired rats were again dosed every 10 h until the end of their assigned timecourse, (i.e., rats in the 48-h timecourse group were given POH or vehicle alone a total of five times). Paired rats were also given 12.5 ml/kg BrdUrd labeling reagent (Amersham Life Sciences, Arlington Heights, IL) via i.p. injection 1 h before the end of their assigned timecourse, at which time rats were sacrificed, and tumors, mammary gland, and intestinal tissues were resected.
Chronic Treatment Protocol.
Rats in the chronic treatment group were randomly paired and were provided a 2% POH (w/w) or pair-fed CON diet as described previously (7). Tumors were followed for regression by twice-weekly palpation over a course of 10 weeks and scored for active regression (>50% reduction of the maximum diameter of the tumor). Once the POH-treated rat in a paired group was found to bear an actively regressing tumor, a BrdUrd slow release pellet (Innovative Research of America, Sarasota, FL) was implanted s.c. into both rats 2 days before their removal from the study, at which time rats were sacrificed and tumors, mammary gland and intestines were resected. All remaining rats were euthanized 10 weeks after diet randomization.
To address possible POH-mediated alterations in normal mammary gland, rats who were not given DMBA were also placed into a chronic POH or pair-fed CON treatment protocol. After 4 weeks of chronic POH or CON treatment, rats were implanted s.c. with a BrdUrd slow release pellet (Innovative Research of America) and sacrificed 2 days later, at which time mammary gland and intestinal tissues were removed.
Tissue Section Preparation and Gross Morphology.
At necropsy, a small portion of each tissue was fixed with freshly prepared 4% paraformaldehyde in PBS (PBS Tablets; Sigma Chemical Co.; pH 7.4) for 4 h on ice, dehydrated in a graded series of ethanol, and embedded in paraffin. Tissue sections were cut 4 μm thick, mounted on poly-L-lysine-coated glass slides, deparaffinized with xylenes, rehydrated in a graded series of ethanol, and washed in ddH2O and in PBS. All tumors used in these investigations were confirmed to be carcinomas by gross morphologic analysis of H&E-stained tissue sections.
At necropsy, a small portion of tissue from selected carcinomas in the chronic treatment group was immediately placed in 3% gluteraldehyde on dental wax. The tumors were cut into 1-mm cubes and transferred to vials with 3% gluteraldehyde in 0.1 M cacodylate (pH 7.4) for 2 h, washed three times with 0.1 M cacodylate buffer containing 7.5% sucrose, and postfixed with 1% OsO4 for 1.5 h on ice. After dehydration in a graded series of ethanol, the tissue was infiltrated first with 1:1 propylene oxide:Eponate 12 (Ted Pella, Redding, CA), then overnight with Eponate 12 and embedded in fresh Eponate 12. Thin sections were cut with a diamond knife on a Reichert Ultracut E ultramicrotome and stained with uranyl acetate and lead citrate. Electron micrographs were taken with a Hitachi H-7000 electron microscope operated at 75 kV.
Circulating POH Metabolite Analysis.
At necropsy, blood was withdrawn from euthanized rats via heart puncture with a syringe flushed with 20 mg/ml heparin (Sigma Chemical Co.) in 0.9% saline. Plasma, prepared by centrifugation of blood, was stored at −20°C, where monoterpene metabolites are stable for 1 yr. POH metabolite concentrations in organic extracts of plasma were determined by capillary gas chromatography with on-line flame ionization and infrared detection as described previously (31).
Digital images of tissues were acquired with a laser-scanning confocal microscope, consisting of a Bio-Rad MRC-1000 laser scan head (Bio-Rad Life Science Research, Hercules, CA) equipped with a 15-mW krypton/argon laser and mounted transversely to an inverted Nikon Diaphot 200 microscope (Nikon, Melville, NY). The excitation color wheel of the confocal microscope, containing the 522DF32 (green), 605DF32 (red), and 680DF32 (far red) filter blocks, allowed separation of emitted green light from FITC into the green channel, red light from LSRC or PI into the red channel, and far red light from Cy5 into the blue channel. The software controlling the microscope and its settings was the BioRad MRC-1024 Laser Sharp (version 2.1T) program. The laser power output and gain control settings within each individual experiment were identical between POH- and CON-treated tissue samples. Digital images were stored as uncompressed TIFF files and exported into Photoshop 4.0 software (Adobe Systems, San Jose, CA) for image processing and hard copy presentation. Confocal microscopy was carried out at the Keck Neural Imaging Laboratory in the Center for Neuroscience (University of Wisconsin-Madison Medical School, Madison, WI).
Cellular S Phase and Apoptosis Staining.
Tissue sections were stained first for apoptosis by TUNEL using the In Situ Cell Death Detection Kit-Fluorescein (Boehringer Mannheim, Indianapolis, IN) and then for cells in S phase by immunolocalization of incorporated BrdUrd using the BrdUrd Labeling and Detection Kit I (Boehringer Mannheim). Tissue sections were pretreated with trypsin for 5 min at room temperature. TUNEL- and BrdUrd-stained tissues were visualized by confocal microscopy. TUNEL-stained nuclei fluoresced green due to the FITC-dUTP label. To differentiate the TUNEL-stained nuclei from the BrdUrd-labeled nuclei in separate color channels by confocal microscopy, a Cy5-donkey antimouse IgG1 2° antibody (Jackson ImmunoResearch Labs) diluted 1:100 in 1% BSA (IgG- and protease-free; Jackson ImmunoResearch Laboratories)/PBS was substituted for the FITC-sheep antimouse IgG1 2° antibody supplied with the cell proliferation kit. Thus, BrdUrd-labeled nuclei were visualized in the blue channel. Nuclei were counterstained with PI (Molecular Probes) and visualized in the red channel. Slides were mounted with ProLong Antifade medium (Molecular Probes). Intestinal tissues were used as a positive control for both TUNEL and BrdUrd labeling. Tissue sections were stained in the absence of TdT, and the monoclonal mouse anti-BrdUrd antibody was used as negative controls for TUNEL and BrdUrd staining, respectively.
TUNEL and BrdUrd-labeled nuclei were scored from six randomly captured images/sample at ×600 original magnification with the UTHSCSA ImageTool version 1.27 program (University of Texas Health Science Center at San Antonio, San Antonio, Texas).5 TUNEL and BrdUrd indices were calculated by normalizing the total TUNEL- and BrdUrd-stained nuclei to the total PI-stained nuclei in all six images/sample. TUNEL and BrdUrd labeling indices for individual carcinomas in each treatment group were averaged, and SDs were calculated. Statistical significance was determined by single factor ANOVA.
Multiplexed-NPAs using oligodeoxynucleotides as probes were carried out using a Multi-NPA kit (Ambion, Austin, TX). NPA oligodeoxynucleotide probes were designed using Oligo 5.0 software (National Biosciences, Plymouth, MN) and the cDNA sequence of the appropriate gene.6 Genes of interest were grouped into panels, with the probe of each gene designed to a specified size such that the entire panel of probes would be resolved by PAGE. Selection of candidate NPA probes was based on specificity as determined by BLAST sequence similarity searches (32). Candidate NPA probes, which returned BLAST queries showing an exact sequence match with only the appropriate gene and not other family members, were selected. Random sequences, designed into the 3′ end of the probe, allowed differentiation by PAGE of protected probes from full-length probes. NPA oligodeoxynucleotide probes were synthesized and PAGE purified by Genosys Biotechnologies (The Woodlands, Texas).
The NPA probe sequences are given below; the names of the NPA probes are underlined, the protected sizes are in parentheses, and random sequences at the 3′-end of the full-length probes are in lower case text:
annexin I (98-mer) 5′-GCCTTGCAATAGGAAGAAAACAACGGCTA-AGAGATGTTCTCCTCGGAATCTTACAGAGCAGTTGGGATGTTTAGT-TTCCTCCACACAGAGCCACCAGGcggcgcgtcgcag-3′
CDK2 (86-mer) 5′-GGCCCAGGATCAGGTCAGACCACAGGTGAAG-AGGGCTTTGGGAAGGACATCAGAGTCGAAGGTGGGGCACTGGTTT-AGTCACATCCgatgcacgctat-3′
cyclin E (75-mer) 5′-GCTGTGGCTCTGCATCCCATGCTTGCTCACG-ACCACTCGCCGTACCCTATCAACAGCAACCCTGCGACACCCAGCct-catatctag-3′
p21Cip1/WAF1 (75-mer) 5′-CACAAATAATAATTAAGACACACTGAATGAAGGCTAAGGCAGAAGATGGGGAAGAGGCCTCCTGAtatgtctact-3′
18S rRNA (56-mer) 5′-CGGGTCGGGAGTGGGTAATTTGCGCGCCTGCTGCCTTCCTTGGATGTGGTAGCCGTggagtcata-3′
bax (47-mer) 5′-CGGAGGAAGTCCAGTGTCCAGCCCATGATGGT-TCTGATCAGCTCGGGtgagcgcac-3′
bad (41-mer) 5′-GAACATACTCTGGGCTGCTGGTCTCCACCGCTT-CCTCCCGCagatag-3′
p53 (75-mer) 5′-CATACGGTACCACCACGCTGTGCCGAAAAGTC-TGCCTGTCGTCCAGATACTCAGCATACGGATTTCCTTCCACCCttcgg-ctgcac-3′
bcl-2 (65-mer) 5′-GATCCAGGTGTGCAGATGCCGGTTCAGGTACT-CAGTCATCCACAGAGCGATGTTGTCCACCAGGGactgtcgaga-3′
M6P/IGF2R (98-mer) 5′-CTGTAGGCACCACGGAGTCAGATGTGTAGGAGGTCCTCGTCACTGTCATCATGGAAGGACACCAGCTTGGC-AGGGGTTGTTGGCTTGCGCTGTCCGGggactcgacgtatc-3′
TβIIR (86-mer) 5′-GCGAAGCGGCCCTTCCCCACCAGCGTGTCCA-GCTCGATGGGCAGCAGTTCCGTATTGTGGTTGATGTTGTTGGCGCA-CGTAGAGCTtgcactctcata-3′
c-jun (75-mer) 5′-CATCTTTGCAGTCATAGAACAGTCGGTCACTT-CACGCGGGGTTAGCCTGGGCTGTGCGCAGAAGTTTCGGGGCCGtgc-tgtaatat-3′
TβIR (65-mer) 5′-TCGCCGCCGCCACCAACACGATGAGAAGCAG-GCAGCGACGCAAAGCAGCCGACGCCGCCTCCATgagtattacta-3′
TGF-β1 (56-mer) 5′-GCGTATCAGTGGGGGTCAGCAGCCGGTTAC-CAAGGTAACGCCAGGAATTGTTGCTAcgccgacta-3′
c-fos (48-mer) 5′-AACATCATGGTCGTGGTTTGGGCAAAGCTCGG-CGAGGGGTCCAGGGGTctgagtca-3′
18S rRNA (41-mer) 5′-CGGGTCGGGAGTGGGTAATTTGCGCGCCT-GCTGCCTTCCTTatgcagc-3′
smad3 (35-mer) 5′-TGACCGCCTTCTCGCACCACTTCTCCTCCTG-CCCGcgacaa-3′
The NPA probes were 5′-end radiolabeled using [γ-32P]ATP (6000 Ci/mmol, 150 mCi/ml) and T4 polynucleotide kinase (Life Technologies, Inc., Gaithersburg, MD), followed by removal of unincorporated radiolabel using a MicroSpin G-25 sephadex column (Pharmacia Biotech, Piscataway, NJ). Equal amounts (1 × 106 cpm/RNA sample) of radiolabeled probes were pooled. Due to the abundance of endogenous 18S rRNA, the radiolabeled 18S rRNA probe was prediluted with 250-fold excess unlabeled 18S rRNA oligonucleotide to lower its specific activity. Total RNA was isolated from frozen tissues using the RNeasy Total RNA Isolation System (Qiagen, Los Angeles, CA). Pooled probes were combined with 50 μg total RNA/sample. To ensure complete digestion of nonprotected probe and the 3′ random sequence of the protected probes, the nuclease digestion mixture supplied by the manufacturer was supplemented with 6.5 μl of S1 Nuclease (Life Technologies, Inc.). Samples were resolved on an 8 M urea/polyacrylamide gel. The gels were exposed to PhosphorImager screens (Molecular Dynamics, Sunnyvale, CA) and quantitated with ImageQuant software (Molecular Dynamics).
Relative expression of each gene was calculated by first normalizing the expression of the target gene to 18S rRNA, and then averaging the expression of the target gene among all the samples within the treatment group. This normalized and averaged expression for the gene of interest in each CON-treated timecourse group was defined as 100% expression and used to determine relative differential expression of the target gene in the matched POH-treated group. SDs were calculated, and statistical significance was determined by single factor ANOVA.
The IHC procedures used to localize and determine qualitative protein expression profiles were based on a standardized protocol developed for these studies. Tissue sections were pretreated with 1 mg/ml trypsin (Trypsin Tablets; Sigma Chemical Co.) for 5 min at room temperature and washed three times in PBS, unless otherwise stated. The sections were blocked for 1 h at 37°C in a humidified chamber with a solution of 5% donkey serum (Jackson ImmunoResearch Laboratories) and 1% BSA prepared in PBS. Appropriate polyclonal 1° antibodies were combined and diluted as necessary with 1% BSA/PBS and incubated with the tissue sections overnight at 4°C in a humidified chamber. Negative controls for each experiment omitted the 1° antibody mixture. The 1° antibodies were detected using donkey-raised, species-specific, fluorochrome-conjugated 2° antibodies (Jackson ImmunoResearch Laboratories) and incubated with the tissue sections for 1 h at 37°C in a humidified chamber. FITC-, LSRC-, and Cy5-conjugated 2° antibodies were diluted 1:200, 1:200, and 1:100, respectively, in 1% BSA/PBS. After incubation with 2° antibodies, slides were mounted with ProLong Antifade medium (Molecular Probes) and visualized by confocal microscopy.
The experiment staining for LAP-TGF-β1 (green channel), active TGF-β1 (red channel), and M6P/IGF2R (blue channel) used the following 1° and 2° antibodies, respectively: goat anti-LAP-TGF-β1 IgG (R&D Systems, Minneapolis, MN) diluted to 20 μg/ml and FITC-antigoat IgG; chicken anti-TGF-β1 IgY (R&D Systems) diluted to 20 μg/ml and LSRC-antichicken IgY; rabbit anti-M6P/IGF2R IgG7 diluted 1:800 and Cy5-antirabbit IgG.
The goat anti-LAP-TGF-β1 antibody specifically recognizes the inactive form of TGF-β1 and the chicken anti-TGF-β1 antibody specifically recognizes the active form of TGF-β1. Because proteolysis can convert LAP-TGF-β1 to active TGF-β1, the trypsin pretreatment step was omitted in the LAP-TGF-β1, active TGF-β1, M6P/IGF2R staining experiment to maintain the specificity of the antibodies. However, in the following staining experiments, the trypsin pretreatment step was included and converted LAP-TGF-β1 to its active form and, hence, the chicken anti-TGF-β1 antibody recognized total TGF-β1.
The experiment staining for TβIIR (green channel) and TGF-β1 (red channel) used the following 1° and 2° antibodies, respectively: rabbit anti-TβIIR (L-21) IgG (Santa Cruz Biotechnology, Santa Cruz, CA) diluted to 2 μg/ml and FITC-antirabbit IgG; chicken anti-TGF-β1 IgY diluted to 20 μg/ml and LSRC-antichicken IgY.
The experiment staining for TβIIR (green channel) and TGF-β1 (red channel) used the following 1° and 2° antibodies, respectively: rabbit anti-TβIR (V-22) IgG (Santa Cruz Biotechnology) diluted to 2 μg/ml and FITC-antirabbit IgG; chicken anti-TGF-β1 IgY diluted to 20 μg/ml and LSRC-antichicken IgY.
The experiment staining for apoptotic nuclei (green channel), TGF-β1 (red channel), and Smad2/Smad3 (blue channel) was modified from the standard IHC protocol by initially staining tissue sections for apoptotic nuclei by TUNEL, as described above. After TUNEL staining, TGF-β1 and Smad2/Smad3 staining used the following 1° and 2° antibodies, respectively: goat anti-TGF-β1 IgG (Santa Cruz Biotechnology) diluted to 20 μg/ml and LSRC-antichicken IgY; rabbit anti-Smad2/Smad3 (E-20) IgG (cross-reacts with both Smad2 and Smad3 as determined by the vendor; Santa Cruz Biotechnology) diluted to 20 μg/ml and Cy5-antirabbit IgG.
Acute and Chronic POH Treatment Protocols.
Rats bearing advanced DMBA-induced mammary carcinomas, defined as ≥10 mm in diameter, were treated with POH according to the chronic or acute protocol. Rats placed into the chronic treatment protocol were fed 2% POH in their diets, and carcinomas were tracked by palpation. Carcinomas that shrank to one-half of their maximum diameter were defined as actively regressing and were resected. During the 10-week chronic treatment protocol, 83% (n = 35 total tumors in group) of POH-treated advanced mammary carcinomas underwent active regression compared with 5.6% (n = 18 total tumors in group) of CON-treated carcinomas. In the chronic POH treatment group, the average time until active carcinoma regression was 23 ± 13 days (mean ± SD).
Rats placed into the acute treatment protocol were serially given 0.1 g/kg POH or vehicle alone (CON-treatment) via gastric intubation every 10 h over a 48-h timecourse. Because metabolism of POH occurs too rapidly to be detected in plasma (7), total circulating levels of the two primary metabolites of POH, perillylic acid, and dihydoperillylic acid, were compared between acute- and chronic-treated rats. The total level of POH metabolites [(perillylic acid) + (dihydoperillylic acid)] in plasma at 12 h (714 ± 102 μm), 24 h (852 ± 218 μm), and 48 h (734 ± 222 μm) of acute POH treatment was not significantly different from the total level of circulating POH metabolites at the time of active carcinoma regression (808 ± 207 μm; P = 0.4 at 12 h, P = 0.7 at 24 h, and P = 0.5 at 48 h of treatment). Therefore, the acute POH treatment protocol achieved physiologically relevant levels of POH metabolites by 12 h and maintained these plasma concentrations throughout the acute treatment protocol.
Gross and Ultrastructural Morphology of Mammary Carcinoma Regression.
Sections of all collected tissues were H&E-stained to examine gross morphologic alterations associated with tumor regression. The CON group carcinomas were characterized by dense anaplastic epithelium (Fig. 1,A) typical of DMBA-induced rat mammary carcinomas (33). As previously reported for LIM-treated regressing carcinomas (6), the POH-treated regressing carcinomas also displayed some regions of anaplastic epithelium, but predominantly exhibited a shrinking epithelial compartment. The remaining epithelial cells showed marked compaction and manifested as a simple epithelium lining pseudolumen structures reminiscent of normal mammary gland (Fig. 1 C). In contrast, examination of chronic POH- and CON-treated normal mammary gland (never exposed to DMBA) on therapy for 4 weeks did not reveal any gross morphological changes (data not shown).
The ultrastructural morphology of monoterpene-mediated regression was investigated using electron microscopy. The CON-treated carcinoma cells were relatively well differentiated as evidenced by numerous microvilli (Fig. 1,B). A large increase in apoptotic cells was found in POH-treated regressing carcinomas compared with CON carcinomas (Fig. 1, C–F). Cells were determined to be apoptotic using morphologic criteria of chromatin margination, cytoplasmic condensation, lucent vacuoles, loss of microvilli, nuclear compaction and fragmentation, and residual bodies (34). Apoptotic cells generally resided at the periphery of epithelial cords (Fig. 1,D) and pseudolumen structures (Fig. 1,E) or were sloughed into pseudolumens (Fig. 1, E–F).
Quantitation of Cytostasis and Apoptosis.
Cellular proliferation and apoptosis were simultaneously quantified using a combination of BrdUrd labeling to detect cells actively synthesizing cellular DNA and TUNEL staining of cells exhibiting fragmented cellular DNA. Tissue sections were counterstained with PI to detect all nuclei. Slides were visualized by confocal microscopy. Quantification of BrdUrd and TUNEL indices involved summation of all respective labeled nuclei normalized to total PI-stained nuclei in six randomly chosen fields/sample at ×600 original magnification (Fig. 2).
Carcinomas in the chronic treatment groups were BrdUrd-labeled for an extended period of 2 days versus the 1-h BrdUrd labeling period in the acute treatment groups in an effort to identify the potentially infrequent cycling of cells in the POH-treated actively regressing carcinomas. Hence, the 48-h acute CON-treated carcinomas exhibited a much lower BrdUrd labeling index than the chronic CON-treated carcinomas. There was no significant change in the BrdUrd labeling indices between the 48-h CON-treated (2.9 ± 0.84%) and the 48-h POH-treated carcinomas (4.1 ± 1.6%; P = 0.14; Fig. 2,C and compare Fig. 2, A with B). However, a comparison of the BrdUrd labeling indices in the chronic CON-treated carcinomas (12.9 ± 1.7%) versus chronic-POH treated actively regressing carcinomas (1.4 ± 0.54%) revealed a 9.2-fold reduction with POH treatment (P = 5.6 × 10−7; Fig. 2,F and compared Fig. 2, D with E).
Control carcinomas showed a low basal level of apoptosis (Fig. 2, A and C), as has been reported previously in DMBA-induced rat mammary carcinomas (35). The TUNEL indices of acute 48-h POH-treated carcinomas (2.0 ± 0.50%) showed a 4.3-fold increase relative to 48-h CON-treated carcinomas (0.47 ± 0.13%, P = 2.0 × 10−4; Fig. 2,C and compare Fig. 2, A with B). The number of TUNEL-stained cells was further increased to 14.1-fold in chronic POH-treated regressing tumors (3.9 ± 1.3%) relative to chronic CON-treated carcinomas (0.27 ± 0.067%, P = 5.6 × 10−5; Fig. 2,F and compared Fig. 2, D with E).
Comparison of chronic POH-treated normal mammary gland (never exposed to DMBA) with chronic CON-treated normal mammary gland on therapy for 4 weeks showed no change in the frequency of BrdUrd- or TUNEL-stained cells (data not shown). Furthermore, a comparison of intestinal tissues from either chronic POH-treated or chronic CON-treated rats, with or without carcinomas, showed no change in the frequency of BrdUrd- or TUNEL-stained cells.
RNA Expression Analysis of Cell Cycle- and Apoptosis-related Genes.
RNA expression of several genes involved with cell cycle and apoptosis regulation, specifically p21Cip1/WAF1, cyclin E, CDK2, annexin I, bad, bax, bcl-2, and p53 was quantified in multiple independent carcinomas using a multiplexed-NPA and a panel of antisense oligodeoxynucleotide probes. Analysis of the data from the NPAs revealed that the cell cycle- and apoptosis-related genes examined were differentially regulated by POH treatment relative to CON treatment (Fig. 3). Expression of the CKI p21Cip1/WAF1 was induced 1.6-fold (P = 0.004) at 48 h of POH treatment and further increased to 2.0-fold (P = 0.0001) during active regression of chronic POH-treated carcinomas. In contrast, cyclin E and CDK2 were repressed to 67.5% (P = 0.01) and 68.8% (P = 0.01), respectively, of control levels at 48 h of POH treatment and further repressed to 51.7% (P = 0.001) and 53.1% (P = 0.0009), respectively, of control levels during active regression of chronic POH-treated carcinomas. RNA expression of the mammary gland apoptosis marker, annexin I, was induced 2.6-fold (P = 0.0001) at 48 h of POH treatment versus CON treatment and induced 2.9-fold in actively regressing POH-treated carcinomas relative to chronic CON-treated carcinomas (P = 0.0001). The RNA expression of the proapoptosis factors bad and bax followed a similar pattern of regulation as annexin I. bad and bax were both induced 1.6-fold (P = 0.001 and P = 0.008, respectively) at 48 h of POH treatment and further increased their induction during active carcinoma regression to 2.2-fold (P = 0.00006) and 2.3-fold (P = 0.0001), respectively. The expression of the antiapoptosis factor, bcl-2, was not significantly altered in chronic POH-treated carcinomas relative to chronic CON-treated carcinomas (97.2%, P = 0.7). Additionally, p53 expression was not significantly different in chronic POH-treated versus chronic CON-treated carcinomas (104.6%, P = 0.5).
Fig. 4 shows the expression profile of the cell cycle- and apoptosis-related genes in multiple independent carcinomas during active regression of chronically treated mammary carcinomas as determined by the NPA. Carcinomas treated chronically with POH, but which did not regress, were included in the analysis. These POH-treated nonresponsive carcinomas showed no significant changes in expression of the cell cycle- and apoptosis-related genes compared with CON-treated carcinomas. In addition, NPAs of chronic POH-treated normal mammary gland (never exposed to DMBA) and chronic CON-treated normal mammary gland on therapy for 4 weeks showed no change in RNA expression of the cell cycle- or apoptosis-related genes examined (data not shown).
RNA Expression Analysis of TGF-β Signaling Components.
RNA expression of many genes involved with TGF-β signal transduction, specifically c-jun, c-fos, TGF-β1, M6P/IGF2R, TβIIR, TβIR, and smad3, was quantified in multiple independent carcinomas with a multiplexed NPA and a panel of antisense oligodeoxynucleotide probes. The NPAs revealed that the TGF-β-related genes were induced and temporally modulated by POH treatment relative to CON treatment (Fig. 5). c-jun and c-fos were transiently induced 2.0-fold (P = 0.0003) and 1.9-fold (P = 0.0005), respectively, at 12 h of POH treatment. TGF-β1 RNA expression was first significantly induced at 24 h of POH treatment (1.8-fold, P = 0.004) and further potentiated at 48 h of POH treatment (3.3-fold, P = 0.0002) and during active carcinoma regression (3.3-fold, P = 0.0001). The M6P/IGF2R RNA expression was first induced to a statistically significant level by 48 h of POH treatment (2.2-fold, P = 0.0003) and was maintained during active regression (2.3-fold, P = 0.0001). The RNA expression of both TβIIR and TβIR followed a pattern similar to M6P/IGF2R. TβIIR and TβIR were induced at 48 h of POH treatment [3.2-fold (P = 0.0001) and 2.4-fold (P = 0.00008), respectively], and induction was maintained during active regression [3.2-fold (P = 0.0004) and 2.5-fold (P = 0.001), respectively]. smad3 was induced only during active carcinoma regression (3.5-fold, P = 0.0007).
Fig. 6 shows the expression profile of the TGF-β-related gene panel in multiple independent carcinomas during active regression of chronically treated mammary carcinomas as determined by the NPA. Nonregressing chronic POH-treated carcinomas were included in the analysis and showed no significant differences in expression of the TGF-β-related genes compared with CON-treated carcinomas. In addition, NPAs of chronic POH-treated normal mammary gland (never exposed to DMBA) and chronic CON-treated normal mammary gland on therapy for 4 weeks showed no change in RNA expression of the tested TGF-β-related genes (data not shown).
In Situ Protein Expression Analysis of TGF-β Signaling Components.
The protein expression profiles of many TGF-β signaling genes were characterized in situ by fluorescence-IHC in concert with confocal microscopy. The investigated signaling components included LAP-TGF-β1, activated TGF-β1, M6P/IGF2R, TβIIR, TβIR, and Smad2/Smad3. Expression of the TGF-β signaling components was compared between chronic POH-treated regressing carcinomas and chronic CON-treated carcinomas. Staining for TGF-β1 was included in all multiple-labeling IHC experiments as a common reference to determine cell-type expression and colocalization of each of the TGF-β-signaling molecules. The intensity of emitted fluorescence within each IHC experiment was controlled by using the same laser power and gain settings on the confocal microscope between the POH- and CON-treated tissue sections. The fluorescence-IHC data are presented, first, as grayscale images to indicate the staining intensity of each protein in its respective color channel and, second, as a composite pseudocolored image, which reflects relative localization of each of the tested proteins.
The first set of multilabeling IHC experiments demonstrated LAP-TGF-β1, active TGF-β1, and M6P/IGF2R expression colocalized and was upregulated in the POH-treated carcinomas relative to CON-treated carcinomas (Fig. 7). Expression of these proteins was restricted to the epithelium, confirmed by positive staining with the mammary epithelial cell differentiation marker cytokeratin 18 (data not shown).
The second set of IHC studies addressed expression of TβIIR concurrently with TGF-β1 (Fig. 8) and expression of TβIR concurrently with TGF-β1 (Fig. 9). TβIIR and TβIR were upregulated in POH-treated carcinoma epithelium (Fig. 8, A and D, and Fig. 9, A and D, respectively). Additionally, TβIIR and TβIR staining colocalized with TGF-β1 staining (Fig. 8,C and 9 C, respectively).
The third set of IHC studies examined expression of Smad2/Smad3 (Fig. 10). Pilot staining experiments for Smad2/Smad3 demonstrated some Smad2/Smad3-positive cells were present in regions consistent with apoptotic cells. Therefore, tissue sections were stained with the TUNEL assay and next stained for TGF-β1 and Smad2/Smad3. Smad2/Smad3 expression was induced and exhibited nuclear localization in POH-treated regressing carcinomas (Fig. 10, C and G). Also, Smad2/Smad3-stained nuclei were primarily localized to regions of the epithelium expressing TGF-β1. Interestingly, subpopulations of nuclei were positive for either Smad2/Smad3, or TUNEL, or both Smad2/Smad3 and TUNEL (Fig. 10, A, C, and D).
IHC studies of chronic POH- and CON-treated normal mammary gland (never exposed to DMBA) on therapy for 4 weeks showed no alterations in the in situ protein expression of the TGF-β-related genes examined (data not shown).
POH-mediated carcinoma regression was characterized by apoptotic deletion of the majority of the dense anaplastic epithelial compartment. Generally, cells were initially deleted from the periphery of the epithelial cords, the remainder of which manifested as simple epithelium surrounding pseudolumen structures, resembling the architecture of the normal mammary gland. These morphologic alterations are consistent with known effects of TGF-β on mammary-derived tissues; TGF-β slow-release pellets implanted into the mammary gland reversibly inhibit mammary gland ductal growth due to elimination of the proliferating stem cell layer (cap cells) and rapid involution of ductal end buds (21). However, unlike direct treatment of mammary cells with TGF-β, POH treatment selectively affected mammary carcinoma tissues rather than normal mammary gland.
TUNEL staining and immunolocalization of BrdUrd-labeled cells demonstrated that apoptosis was induced before cytostasis in POH-treated, regressing mammary carcinomas. POH has been shown to induce apoptosis in hepatic (12) and colon carcinomas (30) using in vivo rodent models and pancreatic cells in culture (28, 29); however, POH treatment did not significantly regulate cellular proliferation in the hepatic and pancreatic models (not determined in the colon model). Thus, POH may mediate distinct anticancer effects by alternate processes in different cell types.
NPAs designed to examine expression of G1-associated cell cycle-related genes in POH-treated carcinomas were based on the finding that POH inhibits the proliferation of cultured mammary cells in G1 (24). p21Cip1/WAF1 was found to be induced, whereas cyclin E and CDK2 were repressed, suggesting a G1 block in the cell cycle. It has been documented that the induction of p21Cip1/WAF1 in response to radiation is mediated by p53-dependent transcription, but induction of p21Cip1/WAF1 in response to TGF-β is independent of p53 (36). NPAs with a p53 probe showed no change in p53 expression. This observation, taken with the demonstrated induction and nuclear localization of Smad2/Smad3 indicating active TGF-β signaling, suggests the p21Cip1/WAF1 induction may have been p53-independent.
RNA expression of apoptosis-related genes, including annexin I, a marker of apoptosis in the mammary gland (27), and p21Cip1/WAF1, which has been documented to be involved in apoptosis (37), were upregulated in POH-treated carcinomas relative to control levels. As stated above, p53 RNA expression did not change. Li et al. (38) have shown with a transgenic mouse model that mammary gland involution proceeds through a p53-independent mechanism. Thus, it is suggested that the POH-mediated apoptosis in mammary carcinomas is p53-independent. Additionally, the NPAs showed that RNA expression of bcl-2 was not changed, whereas bax and bad were up-regulated. A predominance of Bcl-2 blocks apoptosis by enhancing proton efflux at the outer mitochondrial membrane, which maintains the transmembrane potential and integrity. In contrast, high levels of Bax and Bad promote apoptosis by causing a permeability transition via formation of channels in mitochondrial membranes. The consequent release of cyctochrome C and apoptosis-protease-activating factor into the cytosol initiates the caspase cascade and culminates in cellular death (39, 40).
We demonstrated TGF-β signaling components were induced and temporally regulated in the following order: (a) c-jun (transient) and c-fos (transient); (b) TGF-β1; (c) M6P/IGF2R, TβIIR, and TβIR; and (d) smad3. Others have observed that TGF-β treatment leads to activation of SAPK/JNKs (20, 41) and that SAPK/JNKs phosphorylate the transactivation domain of Jun in vivo (42). Phosphorylated Jun facilitates potent AP-1 activity and, thus, transcriptional regulation of many genes, including autoinduction of c-jun (43). Notably, TGF-β autoinduction is also mediated by AP-1 sites (44), and TGF-β up-regulates expression of TβIR and TβIIR (45). Thus, it can be interpreted that the demonstrated POH-mediated transient induction of c-jun and c-fos implies increased AP-1 activity, which may have contributed to the induction of TGF-β and the subsequent induction of its receptors in POH-treated mammary carcinomas.
Recent reports describe the central role of Smads in TGF-β signaling as TGF-β-dependent phosphoproteins that translocate to the nucleus and transcriptionally regulate TGF-β-responsive genes (17, 18). Dissection of Smad nuclear functions shows that Smads regulate transcription by directly binding DNA, such as a consensus Smad binding element recently identified from a pool of randomized oligodeoxynucleotides (46), or specific DNA elements found in promoters for the type VII collagen gene (47) and the human plasminogen activator inhibitor-type 1 gene (48). Additionally, Smads potentiate transcription by serving as coactivators at AP-1 (49) or Sp1 sites (23). Curiously, the TβIR (50) and TβIIR promoters (51) contain Sp1 sites and, as noted above, TGF-β can induce expression of itself (44) and its receptors (45). Thus, Smads operating through AP-1 sites, Sp1 sites, and/or as-yet unidentified Smad binding elements, may form the basis of a positive feedback loop to maintain induction of TGF-β and its receptors during regression of POH-treated mammary carcinomas.
Carcinomas treated with POH that did not regress showed no changes in RNA expression of all the genes examined. These nonresponsive carcinomas may have contained alterations that abrogated their ability to activate the TGF-β signaling pathway and, consequently, to block the induction of cytostasis and apoptosis.
IHC staining of LAP-TGF-β1, activated TGF-β1, M6P/IGF2R, TβIIR, TβIR, and Smad2/Smad3 demonstrated that these TGF-β signaling components were dramatically induced in chronic POH-treated carcinomas relative to chronic CON-treated carcinomas and that expression of these components colocalized in the epithelium. A small area of intense LAP-TGF-β1 staining lacking active TGF-β1 and M6P/IGF2R staining was observed in Fig. 7 D. The NPAs suggest this subpopulation of cells likely reflects an early step in the temporally regulated induction of TGF-β signaling components. Furthermore, these data underscore the importance of M6P/IGF2R in facilitating the proteolytic activation of TGF-β (16). IHC staining for Smad2/Smad3 exhibited nuclear localization, indicating active TGF-β signaling, since phosphorylation of Smads in response to TGF-β is required for translocation of Smads to the nucleus (52, 53). Taken together, these data suggest the use of Smad-phosphorylation status or Smad-nuclear localization as potential biomarkers of anticancer activity.
During POH-mediated carcinoma regression, a subpopulation of Smad2/Smad3-positive nuclei and apoptotic nuclei colocalized, indicating a role for Smads in apoptosis. Overexpression of Smad4, also termed DPC4, has been shown to induce apoptosis through the SAPK/JNK pathway (54). Also, Smad4 synergizes with Smad2 and Smad3 in numerous TGF-β-dependent activities, including transcriptional activity due to formation of heteromeric complexes (52, 55). Thus, though we did not test Smad4, it is likely that Smad2, Smad3, and Smad4 cooperate in their potential role(s) during apoptosis. However, a subpopulation of TUNEL-stained cells were not positive for Smad2/Smad3 nuclear overexpression, and a subpopulation of Smad2/Smad3-positive nuclei did not colocalize with TUNEL staining during POH-mediated carcinoma regression. One possibility is that Smads may participate in apoptosis in some cells and also in processes like cytostasis and/or differentiation in other cells during carcinoma regression. A second possibility is that Smad nuclear translocation and DNA-binding activities are transient; therefore, in some cases, Smad nuclear localization may have preceded the DNA fragmentation detected by the TUNEL assay. In support of the notion that Smad nuclear localization may be a transient event, Yingling et al. (49) have shown that Smad-DNA complexes form within 5 min, peak at 15 min, and disassemble after 4 h of TGF-β treatment. Also, Liu et al. (55) have shown that in TGF-β-treated cells, approximately 30% of those cotransfected with smad4 and smad2 expression vectors and 65% of those cotransfected with smad4 and smad3 expression vectors exhibited nuclear staining of Smad4.
Importantly, all the above described molecular and cellular events were specific to carcinoma tissues and were not found in normal mammary gland or in intestinal tissues chronically treated with POH. Similarly, the POH-mediated induction of TGF-β receptors and apoptosis in a liver tumor model were absent in its tissue of origin (12). Additionally, the apoptotic effects of POH in malignant pancreatic cells were significantly more pronounced than in nonmalignant pancreatic cells (29). In an analogous manner, human trials evaluating POH have noted objective responses with favorable toxicity profiles (8, 9, 10). The tumor-specific events described here may, in part, account for the very high therapeutic index of POH. It is now important to begin to explore the mechanisms underlying this tumor specificity.
The data presented here temporally correlate POH-mediated mammary carcinoma regression and activation of the TGF-β signaling pathway, but a causal relationship has yet to be established. Efforts in our laboratory are addressing a possible causal relationship between POH treatment, activation of TGF-β signaling, and anticancer activity. However, although correlative, POH-mediated induction of the TGF-β pathway was supported by the observed up-regulated RNA expression of c-jun (transient), c-fos (transient), TGF-β1, M6P/IGF2R, TβIIR, TβIR, and smad3 in a temporal order consistent with the TGF-β pathway. POH-mediated induction of the TGF-β pathway was also supported by the up-regulated and colocalized protein expression of TGF-β1, M6P/IGF2R, TβIIR, TβIR, and Smad2/Smad3 in situ. Moreover, active TGF-β signaling in POH-treated regressing carcinomas was indicated by nuclear localization of Smad2/Smad3. Consistent with active TGF-β signaling downstream events, POH treatment resulted in the induction of apoptosis, followed by cytostasis. POH-mediated cytostasis likely reflected the observed coordinated induction of p21Cip1/WAF1 in concert with repression of cyclin E and CDK2. Induction of bax and bad, but not p53 or bcl-2, may have been mechanistically involved in POH-mediated apoptosis, whereas induction of annexin I may have been a marker of apoptosis. Taken together, the data presented suggest a model in which monoterpenes promote the coordinated temporal induction and activation of the TGF-β signal transduction pathway, which leads to the induction of apoptosis closely followed by the induction of cytostasis, resulting in mammary carcinoma regression.
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.
Supported in part by USPHS NIH Grant R37-CA38128. E. A. A. was supported by the Predoctoral Fellowship DAMD17-94-J-4041 granted by the United States Army Medical Research and Materiel Command. C. A. S. was supported by National Cancer Institute Grant CA-07175.
The abbreviations used are: LIM, d-limonene; POH, perillyl alcohol; DMBA, dimethylbenz-[a]-anthracene; M6P/IGF2R, mannose 6-phosphate/insulin-like growth factor type II receptor; TGF-β, transforming growth factor β; LAP-TGF-β1, latent associated peptide-TGF-β1; IGF2, insulin-like growth factor II; TβIR, TGF-β type I receptor; TβIIR, TGF-β type II receptor; MAPK, mitogen-activated protein kinase; SAPK/JNK, stress-activated protein kinase/c-Jun N-terminal kinase; CDK, cyclin-dependent kinase; BrdUrd, bromodeoxyuridine; LSRC, lissamine rhodamine; PI, propidium iodide; TUNEL, TdT-mediated dUTP nick end labeling; NPA, nuclease protection assay; BLAST, basic local alignment search tool; IHC, immunohistochemistry; CON, control.
H. H. Bailey and M. N. Gould, personal communication.
Available from the Internet by anonymous FTP at maxrad6.uthsca.edu.
GenBank sequences available from the National Center of Biotechnology Information at http://www.ncbi.nlm.gov/.
M. J. Ellis, manuscript in preparation.
We thank Wendy S. Kennan and Chern-Sing Goh for technical assistance with animal experimentation, Jennifer L. Ariazi for technical assistance with tissue collection and critical review of the manuscript, and Dr. Laurie A. Shepel for critical review of the manuscript.