All procedures related to animal care and use were approved by the Institutional Animal Care and Use Committee of Brigham Young University. Male and female Noble rats were purchased from Charles River Laboratories (Wilmington, MA) at thirty-five to forty-two days of age for use as breeders. This rodent strain has a low spontaneous incidence of grossly recognizable adenocarcinomas of the dorsal lobe of the prostate 32 but has been used in numerous studies of hormonally-induced prostate cancer 33 and has shown to be prone to steroid hormone reproductive tumors 34. Four males and eight females were assigned to each dietary treatment. The two basal diets used were the Rodent Phytoestrogen Reduced I formulation of Zeigler Bros. (Zeigler Bros. Inc., Gardners, PA) which provides approximately 10 mg isoflavones/kg diet 35 and 0.45 mg Se/kg, and the Harlan-Teklad 8406 diet (Harlan-Teklad, Madison, WI) which provides roughly 600 mg isoflavones/kg and 0.33 mg Se/kg. The detailed composition of these two diets, as previously reported 36, is shown in Table 1. For 36 of the 43 nutrients listed, the difference in concentration is less than 50%, and for only one nutrient (Vitamin D) is the difference more than 2-fold. All nutrients were provided in both diets at levels that met or exceeded the minimum recommendations of the American Institute of Nutrition 37. These two diets were fed with or without a supplement of 3 mg Se/kg diet as Se-methylselenocysteine, a naturally occurring food form of this mineral. The Se-methylselenocysteine used was a generous gift of Kelatron Corporation (Kelatron Corporation, Ogden, UT) and was incorporated into the diets by the manufacturers during formulation. For this chemical form of Se, the maximum tolerable dose, defined as "the dose that produced the first indication of a significant suppression in body weight", was previously reported to be 5.0 ppm 38. The Se content of the high Se diets (3.33–3.45 ppm) used in this study was less than 70% of that value.

These diets provided Se and isoflavones in food forms and combinations naturally occurring in foodstuffs, rather than as semipurified diets with supplements of a single isoflavone and/or Se compound. The use of these diets, tested in pups whose exposure to the diets was begun at conception, was intended to imitate dietary conditions in Asian countries where prostate cancer rates are very low. While there were minor differences between basal diets in vitamin and mineral contents, the difference in isoflavone content was 60-fold. Likewise, the Se-supplemented diets provided a concentration of Se that was roughly 8–10 fold higher than the unsupplemented basal diets. Thus, it is highly likely that differences between diets in gene expression and other parameters were the result of differences in Se and isoflavone dietary concentrations, rather than minimal differences between diets in other nutrients.

Breeders consumed their respective diets for 30 days prior to mating. This ensured that pups used as subjects in this study would be exposed to their respective dietary treatments from conception. Eight offspring in each dietary group were weighed and sacrificed at 200 days of age. Blood was collected and dorsolateral prostate lobes and liver were dissected and weighed, then frozen in liquid nitrogen and stored at -80°C until analyzed. Many previous studies looking at androgen metabolism in the rodent have focused on the ventral prostate, or have not specified which lobes were examined. Much less attention has been given to the dorsolateral prostate lobes. However, the rodent ventral prostate has no clear homologous counterpart in the human prostate. The rodent dorsolateral prostate is the only prostate lobe comparable to the peripheral zone of the human prostate, from which most human tumors develop 39.

Analysis of serum isoflavone content was performed as described previously 40. Briefly, phytoestrogen concentrations were analyzed for each dietary treatment group via gas chromatography/mass spectrometry (GC/MS). This was performed by liquid-solid extraction and liquid gel chromatographic techniques to isolate the phytoestrogen fractions using standards (internal controls) to validate the assay.

Due to limiting amounts of prostate tissue, activity of 5alpha-reductase in dorsal prostate lobes was assayed in just two dietary groups. We chose the two groups most likely to show a difference due to diet – rats fed the adequate Se/low isoflavone diet, and rats fed the high Se/high isoflavone diet. Analysis was performed as described in detail elsewhere 41. Briefly, sections from dorsal prostate lobes were analyzed by TLC using [³H 1,2,6,7]testosterone as the substrate. Radioactivity was then measured in the reduced steroids [5alpha-androstane-3β, 17-dione, 5alpha-dihydrotestosterone and 5alpha-androstane-3alpha, 17 β-diol (3alpha-diol)] and activity rates were calculated. As a measure of Se status, activity of cellular Se-dependent glutathione peroxidase 1 (EC 1.11.1.9; GPX1) was assayed in cytosolic fractions from livers of all animals in each of the four diet treatments as previously described 42.

To select genes for examination in this study we reviewed the results of the data mining exercise of Zhang et al. 43. By microarray analysis, they identified 422 AR-regulated genes in LNCaP human prostate cancer cells, and over 1000 Se-regulated genes in the same cell line. Comparison of the two lists revealed 92 genes regulated by both Se and androgen, of which 37 were reciprocally regulated. These authors also compared differences found in three independently published microarray analyses of gene expression in human prostate tumors compared to normal human prostate tissue. Over 1000 genes dysregulated in prostate cancer appeared in all three reports. Of the 37 genes reciprocally regulated by androgen and Se in LNCaP cells, 6 were among the genes shown by comparison of microarray analyses to be dysregulated in human prostate tumors. Thus, the genes selected for analysis in this study met three criteria: 1) they are dysregulated in human prostate cancer, 2) they are AR-regulated and 3) the androgen effect is opposed by Se. These genes included Abcc4 (ATP-binding cassette sub-family C member 4), Dhcr24 (24-dehydrocholesterol reductase), Facl3 (fatty acid CoA ligase 3), Gucy1a3 (guanylate cyclase alpha 3), and human kallikreins 2 and 3. Kallikreins 2 and 3 have no homologs in rats and were therefore not examined in this work. In addition to the four AR- and Se-regulated genes relevant in prostate cancer, for which homologs exist in rats, we examined expression of the AR itself and of Akr1c9 – the gene for the enzyme which catalyzes the reduction of dihydrotestosterone to its corresponding, less potent 5alpha-androstane-3alpha, 17beta-diol (commonly referred to as 3alpha-diol).

Total RNA was isolated from rat dorsolateral prostates using TRIzol reagent (Invitrogen, Carlsbad, CA). Concentration and purity of RNA were determined spectrophotometrically, and RNA integrity was verified by gel electrophoresis. Equal quantities from each of three individual total RNA isolations within each dietary group were combined to form a total RNA pool for that group. Total RNA pools were reverse transcribed using random hexamers as primers. First strand cDNA was used as a template in quantitative PCR analysis (LightCycler, Roche, Mannheim, Germany) as we have previously described 44. PCR Primers (Table 2) were designed using Accelrys DS gene 1.5 software (San Diego, CA). Optimum temperatures for primer annealing were determined experimentally for each primer pair using a range of annealing temperatures (RoboCycler, Stratagene, La Jolla, CA) followed by gel electrophoresis to confirm amplification of a single band of the expected size. For each gene, at least 3 LightCycler runs were performed. Each run included 3 replicates for each diet. Steady state mRNA levels for 18S rRNA were also quantified and used for normalization.

Proteins were isolated from dorsolateral prostate tissue from three rats in each dietary group. Equal quantities from each of three individual protein isolations within each dietary group were combined to form a protein pool for that group. Proteins (20 μg) were electrophoresed through denaturing SDS-PAGE gels (NuPAGE Novex 10% Bis-Tris gels, Invitrogen, Carlsbad, CA) and blotted to nitrocellulose membranes (Hybond ECL, GE Healthcare Life Sciences, Piscataway, NJ) as previously described 45. Membranes were probed with antibodies to actin, AR, Facl3 (Acsl3), Abcc4 (Mrp4), (Novus Biologicals, Littleton, CO), Dhcr24 (Proteintech Group Inc., Chicago, IL), and Gucy1a3 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). A commercial source for Akr1c9 antibody was not found. Secondary antibodies were conjugated to infrared dyes (LI-COR Biosciences, Lincoln, NE) and detection was by direct infrared fluorescence (Odyssey Infrared Imaging System, LI-COR Biosciences, Lincoln, NE).

Concentrations of amplified products for each replicate of each gene were calculated by the LightCycler software. To correct for possible minor differences in pipetting, initial cDNA concentrations, etc., the final concentration calculated by the LightCycler software for each replicate of a gene of interest in a dietary group was normalized by dividing by the calculated mean for all 18S rRNA replicates in that dietary group.

For each gene, statistical analysis was performed on the 9–12 normalized replicates for each dietary treatment using analysis of variance (ANOVA), followed by Fisher's pairwise comparison to determine significance of differences between dietary groups (Minitab, State College, PA). In the process, a normalized mean value was calculated for each dietary group. Finally, to compare relative expression among the four dietary groups for each gene, the 18S-normalized mean for each of the other dietary groups was divided by the 18S-normalized mean of the adequate Se/low isoflavone group. These are the values shown in Figures 1 and 2.

Analysis of 5alpha-reductase data was done using a two-sample t-test to compare mean values for the adequate Se/low isoflavone diet to the high Se/high isoflavone diet (NCSS, Kaysville, UT). Data for GPX1 activity were also analyzed using ANOVA, followed by Fisher's pairwise comparison as described above.

There were no statistically significant effects of Se on body weight. This confirms the previous report 38 that the level of Se fed in the high Se diets in this study was not toxic to growing rats.

As expected, serum isoflavone levels were significantly higher in rats fed the high isoflavone diets. Serum concentrations averaged 1254 ± 141 (mean ± SD) and 1659 ± 197 ng/mL in rats consuming high isoflavone diets, with and without a 3.0 mg/kg Se supplement, respectively (p < 0.05). In contrast, serum concentrations in rats fed the low isoflavone diets averaged 22.0 ± 2.3 and 24.6 ± 0.1 ng/mL in animals with and without the Se supplement, respectively. The total serum isoflavone concentrations in rats fed 600 mg/kg were similar to those of adults living in Asia while the levels in rats fed only 10 mg/kg were comparable to those in persons consuming typical Western diets 2446.

There were no significant differences due to diet in the activity of hepatic GPX1. This was expected since all diets provided a concentration of Se higher than needed to maximize activity of this enzyme in rat liver. Mean activities in the four dietary groups ranged from 453–567 mU/mg protein, which are comparable to previously published values for GPX1 activity in Se-adequate rat liver 4247. Figure 3 shows that rats fed the high Se-high isoflavone diet had a lower total 5alpha-reductase activity (p = 0.001) when compared to rats fed diets low in Se and isoflavones.

Figures 1 and 2 show genes for whose expression the main effect of Se (Figure 1) or of isoflavones (Figure 2) was statistically significant. Above each bar graph depicting steady state mRNA levels is shown an inset of a Western blot probed with an antibody to that protein. Results of densitometric scanning of band intensities are shown above each inset. As noted above, we were unable to find a commercial source for an antibody to rat Akr1c9. Stripping each blot and probing with an antibody to actin showed no significant differences in gel loading.

In agreement with findings in cultured cells, high Se intake by rats in this study dramatically reduced expression of the AR in the dorsolateral prostate (Figure 1A). Somewhat surprisingly, the main effect of isoflavones was not statistically significant.

The Dhcr24 gene codes for an antiapoptotic protein which inhibits caspase 3. Rats consuming high Se diets had significantly lower expression of this gene (p < 0.001) than animals fed diets without the Se supplement (Figure 1B). There was no significant effect of isoflavones on Dhcr24 gene expression (p = 0.438).

Abcc4 (also called Mrp4) is a member of the multidrug resistance protein family. Its down-regulation may enhance the efficacy of anticancer drugs. The main effect of high supplemental Se intake in reducing Abcc4 gene expression was statistically significant (p = 0.025) (Figure 1C). High isoflavone intake also modestly decreased gene expression, but that main effect was not statistically significant (p = 0.055). This was the only gene for which the combination of high Se and high isoflavones had a more significant effect than either compound alone, or any other combination.

High isoflavone intake significantly decreased the expression of the Gucy1a3 gene (p < 0.001) (Figure 2A). The product of this gene catalyzes conversion of GTP to the second messenger cyclic GMP, which regulates the activity of a variety of protein kinases, phosphodiesterases, and ion channels. The main effect of Se on Gucy1a3 gene expression was not significant (p = 0.472).

Facl3 (fatty acid CoA ligase 3, also called acyl CoA synthetase 3, Ascl3) is associated with fatty acid-induced apoptosis. Ironically, in the cell culture work of Zhang et al., androgen increased expression of this potentially protective gene while Se decreased its expression. In this work, in which rats were rats fed different levels of Se and isoflavones, high isoflavones significantly down-regulated Facl3 (p = 0.001), but Se had no significant effect (p = 0.995) (Figure 2B).

Expression of the Akr1c9 gene – the rat homolog of the human gene whose product reduces dihydrotestosterone to its 3alpha-diol – is decreased in prostate cancer. Somewhat surprisingly, expression of this gene was also decreased by high isoflavone intake (Figure 2C). The main effect of treatment with 3.0 mg/kg Se on expression of Akr1c9 was not statistically significant, although expression in the low isoflavone/high Se group was significantly greater than all other dietary groups.