In vivo molecular mediators of cancer growth suppression and apoptosis by selenium in mammary and prostate models: lack of involvement of gadd genes

Abstract

We used acute selenium (Se) treatments (i.e., daily single oral gavage of 2 mg Se per kg body weight for 3 days) of female Sprague-Dawley rats bearing 1-methyl-1-nitrosourea (MNU)-induced mammary carcinomas to increase the probability of detecting in vivo apoptosis and the associated gene/protein changes in the cancerous epithelial cells. The results show that whereas control carcinomas doubled in volume in 3 days, Se-methylselenocysteine (MSeC) and selenite treatments regressed approximately half of the carcinomas, accompanied by a 3-4 fold increase of morphologically observable apoptosis and ~ 40% inhibition of BrdU index of the cancerous epithelial cells. The mRNA levels of growth arrest-DNA damage inducible (gadd)-34, 45 and 153 genes were, contrary to expectation, not higher in the Se-treated carcinomas than in the gavage or diet restriction control groups. The gadd34 and 153 proteins were localized in the non-epithelial cells, and not induced in the cancer epithelial cells of the Se-treated carcinomas. On the other hand, both Se forms decreased the expression of cyclin D1, and increased levels of P27Kip1 and JNK activation in a majority of the mammary carcinomas. Furthermore, the lack of induction of gadd genes in vivo by methylseleninic acid was confirmed in human prostate xenograft model in athymic nude mice. In summary, these experiments demonstrated the induction of cancer epithelial cell apoptosis and inhibition of cell proliferation by Se in vivo, through potential involvement of Cyclin D1, P27Kip1 and JNK pathways. They cast doubt on the 3 gadd genes as mediators of Se action in vivo.

INTRODUCTION

Selective induction of neoplastic cell apoptosis may be a potential mechanism to mediate the anti-cancer activity of selenium (Se). In vivo, it has been shown that Se-enriched garlic, of which Se-methylselenocysteine (MSeC) constitutes a major Se component, exerts a lasting protective effect when provided for 1 month immediately after 1-methyl-1-nitrosourea (MNU)-exposure as great as when provided throughout the duration of the chemoprevention study (1, 2). This finding indicates that chemopreventive intake of Se may exert such permanent protection against cancer development by inducing the deletion of transformed mammary epithelial cells in vivo through apoptosis.

We and others have investigated the pro-apoptotic effects of Se on mammary and prostate cancer cells and leukemia cells in vitro (3-7) and have identified distinct pathways of signaling and execution with respect to different Se metabolite pools (8). The objectives of the current work were to investigate the in vivo pro-apoptotic and therapeutic effects of Se in preclinical animal models and to characterize the expression changes of selected genes as potential in vivo molecular targets/biomarkers of the anti-cancer action.

Regarding the choice of potential molecular targets for investigation, we have, in earlier work with cell culture models, documented a differential induction of three growth arrest and DNA damage inducible (gadd) genes in mammary cancer epithelial cells by Se compounds that entered two different metabolite pools: selenite as a precursor for the genotoxic hydrogen selenide pool and MSeC and methylseleno-cyanate as precursors to the non-genotoxic methylselenol pool (5, 9). The gadd genes were initially identified by their inducibility after Chinese hamster ovary cells were exposed to a genotoxic agent methyl methanesulfonate (10). These genes are often, but not always, coordinately induced upon exposure of mammalian cells to genotoxic stress or growth arrest conditions (11-13). gadd34 is a homolog of the murine myeloid differentiation gene MyD116 (14). gadd45 is homologous to, but distinct from the murine MyD118 (14) and has been found to interact with proliferating cell nuclear antigen to mediate DNA repair activity and is a transcriptional target of p53 (15). gadd153 product is a homolog of murine CHOP-10, a member of the C/EBP family of transcription factors (13). It is highly inducible by DNA cross-links and by nutrient deprivation (12) and has been linked to endoplasmic reticulum-stress responses (16). The growth arrest and pro-differentiation properties of the gadd gene products and our observation of their induction by Se exposure of mammary cancer epithelial cells in cell culture models (5, 9) prompted our initial effort, more than a decade ago, to link the in vitro biochemical changes with in vivo pro-apoptotic action. Recent studies with PC-3 prostate cancer cells have confirmed the induction of gadd153 in vitro by methylseleninic acid (MSeA), another putative methylselenol precursor compound (7, 17, 18). However, whether the gadd genes play a mediator role in vivo for the anti-cancer effects of Se has not been evaluated.

Tumor size is largely a function of the balance between cell proliferation rate and cell death rate. Aberrant mitogenic and survival signaling pathways are often associated with neoplasia and interfering with such signaling can lead to cell cycle arrest and apoptosis. While the MAPK1/2/ERK1/2 pathway has been primarily involved with proliferation and survival, the stress-activated protein kinases/c-Jun N-terminal kinases (SAPK/JNK) have recently been linked to cellular apoptosis response to stress factors including DNA damage, genotoxic agents, UV and ionizing radiation, cold and heat shock, osmotic pressure, genotoxicity, cytokines or growth factor withdrawal in several cell lines (19). The balance of MAPK1/2 and JNK appears to be a critical determinant of neuronal cell survival (20) and other cell types (21). Cyclins and cyclin dependent kinases (CDK) and their regulatory partners such as P27Kip1 and P21Cip1 are crucial for controlling cell cycle entry and progression (22, 23). Cyclin D1 appears to be particularly critical for mammary epithelial proliferation because in adult D1 knockout mice the mammary epithelial compartment fails to undergo the massive proliferative changes associated with pregnancy despite normal levels of ovarian steroid hormones (24).

We therefore examined expression of the 3 gadd genes, the phosphorylation status of ERK and c-Jun (JNK substrate) as well as the expression of cyclin D1, P21Cip1 and P27Kip1 in Se exposed mammary carcinomas to explore their potential as in vivo targets of Se regulation of apoptosis and cell proliferation. In addition, we sought to extend and cross-validate the findings from the mammary carcinoma model into a human prostate cancer xenograft model in athymic nude mice.

MATERIALS AND METHODS

Chemicals and reagents

Sodium selenite pentahydrate was purchased from J.T. Baker, Inc. (Phillipsburg, NJ). Se-Methylselenocysteine (MSeC) used for mammary cancer study was kindly provided by Dr. Howard Ganther, University of Wisconsin, Madison, WI. MSeC used for the xenograft study was purchased from LKT Labs (St. Paul, MN). Methaneseleninic acid (same as MSeA, >95%, white powder) was purchased from Sigma Chemical Company (St. Louis, MO). Antibodies to gadd proteins, β-actin and Bcl-2 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The phospho-specific antibodies to pERK1/2 and p-Ser63-JUN were purchased from New England Biolabs (Beverly, MA). P27kip1 and cyclin D1 antibodies were purchased from NeoMarker, Inc. (Union City, CA). Anti-BrdU antibody was purchased from Becton Dickinson (Bedford, MA).

Rat mammary carcinogenesis model

This experiment was reviewed and approved by the Institutional Animal Care and Use Committee of AMC Cancer Research Center, Denver, CO and was carried out by HP and JL. Female Sprague-Dawley rats were purchased from Taconic Farms (Germantown, NY) at 20 days of age. Rats were fed a modified AIN-93 diet. Rats were housed 3 per cage in an environment-controlled animal room in the AMC Animal Facility maintained at 22 ± 1°C with 50% relative humidity and a 12-h light/12-h dark cycle. At 21 days of age, animals were each given an intraperitoneal injection of 50 mg MNU (dissolved in acetic acid-acidified saline, pH 4 at a concentration of 14 mg/ml) per kg body weight for the induction of mammary carcinogenesis (25). Starting 4 weeks post carcinogen administration, all rats were palpated 3 times per week for the detection of mammary tumors. Tumor dimensions were measured using a vernier caliper. In this model, palpable tumor growth was very rapid, doubling in volume every 3 days. The majority of the tumors in this short-term model were estrogen responsive as in the standard model (26).

Selenium treatment and other experimental manipulations

When the largest mammary tumor in a tumor bearing rat (TBR) attained or exceeded approximately 1 cm on the longest dimension, the TBR was randomly assigned to one of five treatment groups (n = 4~6 rats): 1) Water oral gavage control using a 1-ml plastic tip delivered through a Pipetman to the back of the tongue, 2) once per day oral gavage of 2 mg Se as MSeC per kg body, 3) once per day gavage of 2 mg Se as sodium selenite per kg body weight, 4) diet restriction (DR, to 70% ad lib intake), and 5) bilateral ovariectomy (OVX). The Se treatment was repeated daily for a total exposure of 6 mg of Se per kg body weight over the 3-day duration. The Se dosage chosen was approximately 5-10 times of the daily Se intake in a typical chemoprevention setting. Tumor samples were harvested 24 h after the third Se dose.

We included OVX and DR groups for comparison with the following rationale. The majority of MNU-induced mammary carcinomas are dependent on and sensitive to ovarian hormones (26). Therefore, OVX was used as a positive control for apoptosis and growth inhibition of cancerous mammary epithelial cells to compare and contrast similarities and distinctions in the apoptosis and growth regulatory mechanisms induced by Se and ovarian oblation. Because intake of Se in much excess of its nutritional requirement can lead to adverse effects on food intake and body weight, DR was used to control for non-specific effects associated with body weight loss.

Tumor size and necropsy

The size of a tumor was measured as its two largest perpendicular dimensions and these data were used to compute volume using the formula for an ellipsoid. Tumor-bearing rats were killed 24 h after the third Se dose. A 5-bromo-2′-deoxyuridine (BrdU) pulse labeling technique was used to assess the proliferation rate of the cancerous epithelial cells (27). Rats were given an i.p. injection of 50 mg BrdU (dissolved in saline) (Sigma Chemical Co.) per kg body weight 3 h prior to euthanasia. All tumors (including those for which size measures were not taken) were excised, and cut in the middle. Two slices were fixed in neutral buffered formalin and methacarn (acidic alcohol fixative), respectively. The rest of the tumor tissue was frozen in liquid nitrogen for assessment of biochemical and gene expression changes. Because of heavy demand of tissues for the various assays, it was necessary to perform some assays on different sets of tumors and therefore not all measured parameters could be correlated.

Human DU145 prostate cancer xenograft model in athymic nude mice

The animal use protocol was approved by the Institutional Animal Care and Use Committee of the University of Minnesota, and carried out at the Hormel Institute's animal facility. Male Balb/c athymic nude mice were purchased from NxGen BioSciences (San Diego CA), at 4-5 weeks of age. They were housed in a specific-pathogen free room with free access to water and commercial rodent chow. Animals are maintained in clean HEPA-filter top covered cages. Cages, bedding, cage tops and water bottles were washed and maintained separately from other animal materials and supplies. After 2 weeks of quarantine, each mouse was inoculated by s.c. injection with DU145 cells and tumor measurements and selenium treatments have been described recently (28).

In order to assess the early dynamic processes of apoptosis signaling in the target human prostate cancer cells in vivo, we treated tumor-bearing mice with MSeA at 4 mg/kg body weight for 24 h and 72 h when the xenograft has grown to about 150 cubic mm (3 weeks after inoculation). Tumors from 5-6 mice of each group were pooled (approximately equal weight) for preparation of tissue lysate for Western blot analysis of biomarkers related to caspase-mediated apoptosis and Gadd protein abundance.

Apoptotic indices

For the rat mammary cancer model, apoptosis of the cancerous mammary epithelial cells was morphologically determined on hematoxylin and eosin stained sections of the formalin-fixed portion of the carcinomas. Primary criteria for the identification of apoptotic cells were cell shrinkage, nuclear condensation, fragmentation, apoptotic bodies, and lack of inflammatory components (29). The apoptotic index, i.e., the percentage of apoptotic cells over total cells counted, was determined on 20 200X fields of cancerous epithelial cells, excluding regions of gross necrosis. For the xenograft tumors, apoptosis was detected by TUNEL assay according to manufecturer's instructions (28).

BrdU index for cell proliferation estimation

Methacarn-fixed mammary carcinomas were thin-sectioned (5 μm) and stained for BrdU immunohistochemically by a method as described by McGinley et al (27). Anti-BrdU antibody (1:40) was used to detect nuclei that have incorporated BrdU into their DNA. After a light nuclear counterstain with hematoxylin, the BrdU labeled and unlabeled nuclei were counted at 400X with a computer assisted image analyzer using the Quantitative Nuclear Antigen Program Version 3.0 (CAS-200, Becton-Dickinson/Cellular Analysis Systems, San Jose, CA) as described elsewhere (30). The proliferation index, i.e., the percentage of labeled cells over total cells counted, was determined by counting 20 fields of cancerous epithelial cells per slide (approximately 2000 cells), excluding gross necrotic regions.

Steady state mRNA level by RT-PCR

Total RNA was extracted from the frozen carcinoma tissues using the RNAzol™ B reagent (Tel-Test, Inc. Friendswood, TX). One half (0.5) μg of total RNA was reverse transcribed using anchored oligo(dT)₁₀ as the primers. The PCR primers were synthesized commercially (Integrated DNA Technologies, Inc., Coralville, IA). The sequences of primers and PCR setting information were described earlier (9).

For real time RT-PCR, total RNA were extracted using RNeasy kit (QIAGEN). One μg of total RNA was used for cDNA synthesis with oligo-dT primer and SuperScript II Reverse Transcriptase (Invitrogen) in a 20μL reaction system, and 5 μl of diluted cDNA (1:20) were used in each 25-μl real-time PCR reaction using the Fast Start Universal SYBR Master with ROX (Roche) with an ABI 7500 Real Time PCR System (Applied Biosystems). A standard curve was generated in each assay and used to derive the copy number of each transcript. Beta-actin was selected as an internal standard, and all the raw data were expressed as the ratio of the copy number of selected gene to beta-actin. Statistical significance was determined by an unpaired, two-tailed Student's t-test.

Western blot analyses

The expression levels of growth regulatory- and cell death-associated proteins were assessed by Western immunoblot analyses of mammary carcinoma extracts using enhanced chemoluminescence detection (31). Total protein content of extracts was determined by Bradford dye binding assay (Bio-Rad Laboratories, Richmond, CA). After probing for the protein of interest, the membranes were stripped and re-probed for β-actin to control for gel loading differences and protein integrity. The X-ray films were digitized using a transmission scanner and the signal intensity was quantitated using the UN-SCAN-IT gel scanner software (Silk Scientific, Inc. Orem, Utah). The signals (pixels) for the proteins were normalized to that of the corresponding β-actin. The normalized expression data were used for statistical evaluations.

Immunohistochemical localization of proteins

Formalin-fixed and paraffin embedded mammary carcinomas were thin sectioned (5 μm) and immunohistochemically stained for gadd34, gadd153 and P27Kip1 proteins. Antigen retrieval was performed on these sections in 10 mM sodium citrate buffer, pH 6.0 by microwave heating.

Statistical analyses

Unless otherwise stated, data are reported as mean ± standard error. Analysis of variance (ANOVA) and/or rank-order tests, when the variance was not normally distributed, were carried out with the Systat package (Systat, Inc. Evansville, IL).

RESULTS

Response of established rat mammary carcinomas to acute Se exposure

Daily single oral gavage treatment (delivered to the back of tongue) of mammary tumor-bearing rats with either MSeC or selenite (SEL) for 3 days resulted in a 6% or 3% body weight loss in comparison with pre-treatment weight respectively, while the control rats (CTRL) increased body weight in the same time by 5% (Fig 1A). These treatments elevated the liver Se content 4.6 fold and 3.3 fold respectively (Fig 1B). Because sustained caloric restriction and associated body weight loss have been shown to inhibit mammary carcinogenesis in this model (32), we therefore instituted a dietary restriction (DR) control to investigate the pro-apoptotic activity of Se exposure that was independent of body weight loss.

Figure 1.

Effects of daily single gavage of selenium compounds, dietary restriction (DR) or ovariectomy (OVX) in tumor-bearing rats on (A) body weight, (B) liver Se content, (C) carcinoma volume, (D) the rate of mammary tumor epithelial cell apoptosis and (E) the rate of mammary tumor epithelial cell proliferation assessed at 3 days after the initiation of each experimental manipulations. CTRL, gavage control group; MSeC, methylselenocysteine treated group, 2 mg Se/kg body weight per day; SEL, selenite treated group, 2 mg Se/kg body weight per day. Panel A: Body weight relative to day 0. N= 4-6 rats. Panel B: Liver Se content on wet tissue basis. ND-not determined for the DR and OVX groups. N= 4-6. Panel C: Tumor volume change in reference to pre-treatment value on day 0. N=5-8. Panel D: Incidence of morphologically identifiable apoptotic cells in cancerous mammary epithelial cells, excluding gross necrotic areas. Panel E: % of cancerous mammary epithelial cells incorporating BrdU, excluding gross necrotic areas. Data were expressed as mean ± Sem. Statistical difference from gavage control: * p<0.05; ** p<0.01; *** p<0.001.

The tumor volume of the gavage control rats increased in 3 days by 125 ± 48 % (mean ± sem) (Fig 1C). The vast majority of the tumors in either MSeC or selenite treated tumor-bearing rats did not grow in the same duration. In fact 4 out 8 MSeC-treated tumors and 4 out of 7 selenite-treated tumors showed regression, respectively, relative to pre-treatment volumes. The magnitude of Se-induced tumor size changes was almost as much as that induced by bilateral OVX, a positive control for apoptosis induced by estrogen-deprivation (Fig 1C). On the other hand, diet restriction (DR) that resulted in a comparable degree of body weight loss (i.e., 6% of pre-treatment weight) did not inhibit tumor growth with a tumor volume increase in 3 days of 105 ± 18% (Fig 1C). Taken together, the tumor volume data (Fig 1C) indicated that the vast majority of established mammary tumors were responsive to the acute delivery of Se compounds and such a response could not be accounted for by the associated body weight loss.

Mammary tumor epithelial cell apoptosis in vivo

As a positive control of mammary tumor epithelial apoptosis, OVX led to a 5-fold increase in the incidence of morphologically observable epithelial apoptosis when examined 3 days after surgery (Fig 1D). Tumors in MSeC-treated and SEL-treated rats showed a 3-fold and 4-fold increase in apoptosis, respectively. Dietary restriction (DR) of the tumor-bearing rats did not increase the incidence of apoptosis of the mammary tumor epithelial cells.

Mammary tumor epithelial cell proliferation in vivo

As would be expected of the inhibitory effect of estrogen depletion on mammary epithelial proliferation, the rate of cell proliferation measured as BrdU incorporation in the mammary tumors was decreased by 65% by OVX (Fig 1E). The epithelial BrdU index of mammary tumors that showed regression after MSeC or selenite exposure were decreased by 40% in comparison to the gavage control tumors (Fig. 1E). Dietary restriction (DR) did not decrease the BrdU index of the mammary tumor epithelial cells, indicating that the weight loss associated with Se exposure could not account for the decreased BrdU incorporation.

Steady state transcript (mRNA) level of gadd and selected growth regulatory genes

The mRNA level in individual mammary tumors from the various groups was detected by RT-PCR (Fig 2). We tested the linearity of RT-PCR detection of increasing amount of input RNA as demonstrated for the two house keeping genes, GAPDH and beta-actin (Fig. 2A). For the tumor samples, the expression level of actin, when normalized to gapdh used as an internal control, was essentially constant across all groups (Figure 2B). Diet restriction did not affect the mRNA level of all the genes examined. Contrary to the induction effects reported for cell culture (5, 9), exposure of tumor-bearing rats to either form of Se did not increase the mRNA level of the 3 gadd genes in the mammary tumors that underwent significant growth inhibition and even regression (Fig 2B). In fact, the expression level of gadd45 and gadd153 in a significant proportion of the Se-treated tumors was lower than the gavage- or DR- controls (Fig 2B). The mRNA level of p21cip1 was not increased by Se treatments, and was even lower in some of the selenite-treated carcinomas. The mRNA level of cyclin D1 was decreased by 50% and 30% by MSeC and selenite treatment, respectively (Figure 2B). The Se treatment effects on cyclin D1 and gadd153 expression were in similar trend as those of OVX. However, OVX led to modest increases of the expression of gadd34 and gadd45 and much greater induction of p21cip1, in contrast to the Se treatments, which decreased the mRNA abundance of these genes.

Figure 2.

RT-PCR analyses of the steady state mRNA level in mammary carcinomas. A. Serial dilution of a mammary tumor cDNA sample to demonstrate linearity of detection of two house keeping genes gapdh and actin. B. RT-PCR analyses of the steady state mRNA level of gadd, p21 and cyclin D1 genes in mammary carcinomas after acute treatments with selenium compounds or ovariectomy (OVX). The PCR primers and conditions were as published (9). An equivalent of 20 ng of total RNA was used for each sample. Each lane represents an individual carcinoma. The house keeping genes gapdh and actin were evaluated as normalization controls for loading differences. The normalized pixel intensity values are presented as mean ± sem. Data were analyzed by analyses of variance with post-hoc ranking. Means among groups in a same row with different superscripts were significantly different (p<0.05).

In situ localization of gadd proteins and P27Kip1

Because RT-PCR was performed on mammary tumor tissues with no ability to distinguish the cell type origin (cancer epithelial vs. stromal and other cell types) of the gene/protein expression changes, we did immunohistochemistry to determine the expression patterns of gadd34 and gadd153 proteins in the rat mammary tumors. Gadd34 protein was localized to the non-epithelial stromal components, especially in endothelial cells lining tumor microvessels (Fig 3). Gadd34 protein expression was not detected in mammary tumor epithelial cells of tumor-bearing rats exposed to either MSeC or selenite (Fig. 3, note the absence of staining in glandular epithelial cells). Gadd153 staining had an even more restrictive pattern than that of gadd34, localizing almost exclusively to microvessels (not shown). Selenium treatment of the tumor bearing rats did not increase expression of gadd153 protein in the tumor epithelial cells. The non-epithelial expression patterns along with the lack of induction at the mRNA level did not support an association of the gadd expression with the observed in vivo anti-proliferative and pro-apoptotic effects of Se on mammary tumor epithelial cells.

Figure 3.

Immunohistochemical detection of gadd34 and P27 proteins in mammary carcinomas after acute treatments with selenium compounds or ovariectomy (OVX). The gadd34 staining (upper row) was localized in the stromal cell components (S) of the mammary carcinoma, most notably in endothelial cells lining microvessels, and absent in the epithelial cells (E) before and after Se treatment. P27kip1 staining (lower row) was weak in untreated carcinomas. The P27 staining in selenite-, MSeC- or OVX-treated carcinomas increased over the gavage control carcinomas and was predominantly in the cancerous epithelial cells. The relative P27 staining intensity (by CAS image analyses) in the cancerous epithelial cells in comparison to the control group was presented below respective groups.

The P27kip1 staining in the gavage control carcinomas was weak in the epithelial cells (Fig. 3), whereas un-involved mammary ducts contained strong epithelial staining (not shown). Selenite and OVX significantly increased the expression of P27 protein in the mammary cancer epithelial cells, however the effect of MSeC on P27 staining was less dramatic than that of selenite: The relative P27 staining intensity (by CAS image analyses) in the cancerous epithelial cells in comparison to the control group (1 ± 0.1, mean ± sem, n = 4) were 1.8 ± 0.5 (n = 5), 5.6 ± 1.2 (n = 7) and 4.6 ± 1.5 (n = 4) for MSeC, selenite and OVX groups, respectively (Fig. 3). P21Cip1 staining was nuclear and localized predominantly in mammary epithelial cells (not shown). In contrast to the more uniform staining of P27 in epithelial cells, P21 staining was sporadically distributed among cancer epithelial cells and the proportion of cells staining positive for P21 was not increased by Se treatment.

Expression changes of selected growth- and apoptosis-regulatory proteins

We next examined expression level of cell proliferation related cyclin D1 and ERK activation status (by phosphorylation) and the apoptotic-regulatory BCL-2 expression and JNK activation by Western blot analyses (Fig 4). The cyclin D1 protein level was decreased in most tumors in the Se-treated rats and in ovariectomized rats (Fig 4), in agreement with mRNA expression patterns (Fig. 3B). The levels of phosphorylated ERK (pERK1/2) were not significantly decreased in tumors treated with either Se or by OVX (Fig 4).

Figure 4.

Western blot analyses of growth- and apoptosis-regulatory proteins in mammary carcinomas after acute treatments with selenium compounds or ovariectomy (OVX). Forty μg (by Bradford dye assay) of tumor protein extract was loaded from each sample. β-actin expression was probed as internal control for loading differences. The normalized pixel densities (expression level) were presented in as mean ± sem. Means among groups within a same row with different superscripts were significantly different (p< 0.05). Due to limitation of number of wells on each gel, the carcinomas from each group were analyzed in 2 different sets and images were compiled later.

JNK activation as measured by the level of phosphorylation of c-Jun on Ser63 was observed in 3 out of 5 tumors in MSeC-exposed rats and 4 out of 7 tumors from selenite-treated rats (Fig. 4). The level of JNK activation was at least one order of magnitude lower than that observed in OVX rats. The level of the cell survival protein Bcl-2 was not decreased in tumors from Se treated rats or OVX (Fig 4), indicating no disabling of Bcl-2 function in mammary tumors containing cells actively engaged in apoptosis induced by Se treatments or OVX.

Human Prostate Cancer xenograft model

To determine whether the lack of in vivo induction of gadd genes by acute Se treatment was specific to the mammary tumors, we tested the gadd expression changes in DU145 prostate cancer cells in vitro (Fig. 5A) and in xenograft after treatment with MSeA or MSeC (Fig. 5B and C). We confirmed that in cell culture with DU145 and LNCaP prostate cancer cells, exposure to an apoptotic dose of MSeA for 6 h strongly activated gadd153 expression (Fig. 5A), as has been reported in PC-3 cells (7, 17). We have previously shown that MSeA and MSeC, given to nude mice at a dosage of 4 mg per kg body weight starting one day after DU145 cell inoculation, resulted in greater than 50% inhibition of tumor growth in 5 weeks with little adverse effect on body weight (28). Then we analyzed these xenograft tumors for either the human (Fig. 5B) or murine (Fig. 5C) mRNA for gadd34, gadd45 and gadd153 by real time RT-PCR and did not detect any increase due to the Se treatments.

Figure 5.

Real time RT-PCR detection of gadd mRNA in DU145 prostate cancer cells in cell culture or xenograft from nude mice. A. Induction of gadd153 mRNA level by MSeA after 6 h treatment of DU145 cells. B. mRNA level of human gadd34, gadd45 and gadd153 in DU145 xenografts after the tumor bearing mice has been treated with daily oral gavage 4 mg/kg of MSeA or MSeC for 5 weeks [See ref. (28)]. C. mRNA level of murine gadd34, gadd45 and gadd153 in DU145 xenografts from B.

In order to characterize the kinetics of apoptosis in the human xenograft tumors, we carried out an acute treatment experiment in tumor-bearing nude mice after the xenografts had grown to an average size of 150 mm3 with MSeA (4 mg Se /kg, oral single dose, daily). Tumors were frozen at −70°C until analyzed. Tumors from 5 mice of each group were pooled (approximately equal weight) for preparation of tissue lysate for Western blot analysis of biomarkers related to caspase-mediated apoptosis and gadd protein expression.

Increased cleavage of PARP was detected at both 24 h and 72 h of MSeA exposure (Fig. 6A). Concomitantly, we detected an activation of caspase-3 cleavage as early as 24 h and sustained through 72 h (Fig. 6A). Pro-caspase-8 cleavage displayed a similar pattern as did caspase-3 (not shown). TUNEL assay for apoptotic fragmented DNA ends detected increased positive signals (Fig. 6B). These data were consistent with increased caspase-mediated apoptosis in human xenograft tumors by MSeA acute exposure.

Figure 6.

Apoptosis response of human DU145 prostate cancer xenograft in nude mice after acute exposure to MSeA. A. Western blot analyses of growth- and apoptosis-regulatory proteins in DU145 prostate cancer xenograft after acute treatments of tumor-bearing mice with MSeA (4 mg/kg) for 24 or 72 h. Protein extract was prepared from pooled sample of 5-6 tumors. β-actin expression was probed as internal control for loading differences. B. TUNEL detection of increased apoptosis (dark stains) after 72 hr treatment with MSeA. C. Schematic pathway links of in vivo Se treatment to apoptosis and growth inhibition in cancer cells.

Western blot analyses did not detect any increased gadd-34 abundance or gadd153 (Fig. 6A). A modest decrease of cyclin D1 abundance at both 24 and 72 h of MSeA treatment was observed while a modest induction of P27Kip1 was detected by 72 h (Fig. 6A).

DISCUSSION

The present work used two acute exposure models to investigate apoptosis induction by selenium compounds and the potential in vivo molecular targets/biomarkers of the chemotherapeutic action. In the MNU-induced rat mammary carcinogenesis model, we observed that a daily single gavage of either selenite or MSeC (2 mg Se per kg body weight) for 3 days in tumor-bearing female rats led to the regression of a vast majority of the established mammary tumors (Fig 1C). The observed tumor volume reduction was associated with significantly increased incidence of cancer epithelial apoptosis (Fig. 1D) and moderately decreased rate of BrdU incorporation in the cancer cells (Fig. 1E). The dosage of selenite or MSeC used here approximated 5 -10 fold of typical chemopreventive intake (8, 33). Since Ip and co-workers have shown that established tumors in the mammary model did not respond to the typical chemopreventive Se intake (1), we chose the higher dose of Se intending to elicit a measurable tumor volume response in an acute therapy context to enable us to examine whether the gadd gene expression changes could be observed in vivo and associated with the in vivo apoptosis in the cancerous mammary epithelium. The high dose was not acutely lethal to the rats, but nevertheless toxic to them, resulting in a loss of body weight in comparison to gavage control rats (Fig. 1A). However, the tumor regressing action of Se exposure could not be accounted for by weight loss per se as evident by the lack of any effect of DR that led to a comparable degree of body weight loss on tumor volume. DR did not inhibit epithelial cell proliferation, nor did it induce apoptosis or gene/protein expression changes (Fig 2).

In an effort to identify potential biomarkers or targets of the pro-apoptotic and anti-proliferative activities of Se in vivo observed in this tumor model, we compared the expression patterns of gadd gene products and selected cell cycle and apoptosis-regulatory proteins (Figs 2 to 4) in the tumors from gavage control and Se-treated tumor-bearing rats. In contrast to the induction effects of Se compounds on gadd gene expression in mouse mammary tumor cells observed in cell culture models (5, 9), the steady state level of mRNA for all 3 gadd genes examined was not higher in Se-treated tumors than in those from the gavage- or DR-control animals (Fig 2). Furthermore, the gadd34 and gadd153 protein products were localized in the non-epithelial cells and no induction by Se treatment was detected in the epithelial cells (Fig 3). These data were inconsistent with a direct mediator role of the gadd gene products for the in vivo anti-proliferative and pro-apoptotic activity of Se in the mammary tumor epithelial cells, although we could not rule out the possibility that our single time point for harvesting the mammary tumors for analyses might have missed the window of induction of these genes.

To further address this issue and to determine whether the findings were specific to mammary tumors, we examined human prostate cancer xenograft models in nude mice (Fig. 5). The additional work was significant because the prostate has been found to be the most responsive organ site for selenized-yeast intervention of cancer risk in a clinical trial led by the late Dr. Larry Clark (34). Prompted by these encouraging results, the Selenium and Vitamin E Cancer Prevention Trial (SELECT) study (35) was testing the prostate cancer preventive efficacy of selenomethionine (SeMet), which is the major Se species present in selenized-yeast, and/or vitamin E supplementation in North American men (www.cancer.gov/clinicaltrials/digestpage/SELECT). The enrollment of some 35, 500 subjects was completed in June 2004. The SeMet intervention was designed for an average of 7 years. The National Cancer Institute stopped the trial in late October 2008, several years ahead of the scheduled completion date, due to a possible increase of diabetes risk in SeMet-supplemented subjects and a rise in prostate cancer risk by the vitamin E-supplemented subjects. In hindsight, preclinical prostate cancer models conducted prior to (36) and since SELECT was initiated (28, 37) did not support any in vivo anti-cancer activity of SeMet. Whereas the SELECT results convincingly ruled out SeMet for prostate cancer prevention, in vivo preclinical mechanistic studies of other Se forms will help to critically judge their potential utility as candidate compounds in future clinical trials. Much work since the original Clark study has focused on prostate cancer cell lines with Se forms much more active than SeMet, identifying numerous signaling pathways (8) including gadd genes (7, 17, 18). However, none of the studies dealt with in vivo validation of the molecular targets or cellular pathways.

Same as in the mammary tumors, we did not detect increased mRNA level for the 3 gadd genes in the human DU145 prostate cancer xenograft after MSeA or MSeC treatment (Fig. 5B and C), in spite of significant induction in cell culture by MSeA as shown in Fig. 5A or as reported in PC-3 cells (7, 17, 18). In the acute exposure context, we failed to detect increased gadd34 or gadd153 protein (Fig. 6A) even though we detected increased caspase-mediated apoptosis (Fig. 6A and B). Taken together, both the mammary and prostate models failed to link the in vivo apoptosis and anti-proliferation actions of Se to these gadd genes.

The disconnect between in vitro induction of gadd genes by Se and the lack of involvement in the two in vivo cancer models is not surprising. Cell culture models measure, by and large, the direct effects of Se compounds on target cancer cells, but rarely simulate in vivo hepatic and systemic Se metabolism. In vivo metabolism could have resulted in Se forms that came in contact with the cancer cells very different from what were administered to the animals. Our current work highlights the risk of missing the mark when cell culture models are used as target discovery tools for compounds that may undergo extensive in vivo metabolism.

On the other hand, our results suggested the potential involvement of cyclin D1, P27Kip1 and the JNK pathway in growth arrest and apoptosis induced by Se exposure in vivo. Cyclin D1 expression, at both mRNA and protein levels, was decreased in most of the Se-treated tumors (Figs 2 and 4) and in prostate xenograft (Fig. 6). P27Kip1 expression was induced in the tumor epithelial cells by selenite and to a lesser extent by MSeC exposure (Fig 3). Consistent with this acute exposure model, chronic feeding of MSeC to MNU-treated rats led to the increased expression of P27 in intraductal proliferation (IDP) lesions which represent an early stage of mammary carcinogenesis progression (38). Because cyclin D1 and P27 are respective positive and negative regulators of CDK activities, the observed induction of P27 and reduction of cyclin D1 expression by acute Se exposure suggest a potential in vivo mechanism for Se to arrest cell cycle progression through modulating the activities of CDK4, leading to growth arrest (Fig. 6C). In addition, the observed in vivo pro-apoptotic effect of Se may be facilitated by perturbed survival/apoptosis signals through stress-activated protein kinase pathways. The observed JNK activation in some Se-treated mammary carcinomas (Fig. 4) lends credence to this postulate (Fig. 6C).

In summary, the lack of up-regulation by Se treatments on the in vivo expression of gadd genes in two preclinical animal cancer models cast doubt on their role as potential mediators of the anti-proliferative and pro-apoptotic activity of Se. The observed reduction of cyclin D1 expression, induction of P27Kip1 and the activation of JNK in a majority of the Se-exposed mammary carcinomas implicated their potential involvement in the Se actions in vivo.

The abbreviations

Se

selenium

MSeC

Se-methylselenocysteine

MSeA

methylseleninic acid

CDK

cyclin-dependent kinase(s)

CKI

cyclin-dependent kinase inhibitor(s)

MNU

1-methyl-1-nitrosourea

gadd

growth arrest and DNA damage inducible genes

DR

diet restriction

OVX

ovariectomy

BrdU

5-Bromo-2′,3′-dideoxyuridine

RT-PCR

reverse transcription-polymerase chain reaction

IHC

immunohistochemistry

MAPK

mitogen activated protein kinases

JNK

Jun N-terminal kinase