Mouse 10T1/2 cells grown at 37˚C in DMEM with 10% FBS, 4 mM glutamine, and 1x antibiotic-antimycotic (GM) were treated with 20 μM 5-azacytidine (aza) in GM. After 3 days, cells were split 1:2 and were allowed to recover in GM for 2 days. Cells were seeded at 70% confluence in 6-well plates or chamber slides and grown with test agents for 0 to14 days [17] .

Mouse 3T3-L1 preadipocytes (American Type Culture Collection, Manassas, VA) were maintained in growth medium (GM) containing DMEM supplemented with 10% FBS as described previously [21] . Differentiation was induced in 100% confluent 3T3-L1 cells by incubating them in adipogenic medium (AM) (GM with 0.5 mM isobutylmethylxanthine, 1 μM dexamethasone and 10 μg/ml insulin) for three days with or without androgens. This was followed by changing the medium to GM containing 10 μg/ml insulin with or without androgens at various concentrations for another 2 days and then to GM plus treatments for up to 10 days.

For immunochemical analyses, cells grown in chamber slides were fixed in 2% paraformaldehyde for 20 min, quenched with H₂O₂, blocked with normal horse serum and incubated with anti-myosin heavy chain II (MHCII) antibody [20] . Detection was based on a secondary biotinylated antibody, followed by the addition of streptavidin-horseradish peroxidase ABC complex (1:100) (Vectastain Elite ABC System, Novocastra Laboratories, Newcastle upon Tyne, UK) and 3,3-diaminobenzidine. The cells were counter-stained with Meyer’s hematoxyline.

Cells were fixed in 2% paraformaldehyde after treatments and stained with 0.3% Oil Red O (Sigma Chemical Co., Saint Louis, MO) for 15 min as described [22] . For quantitative analysis of Oil Red O retention in these cells, stained adipocytes were extracted with 1 ml of 4% Igepal CA-630 (Sigma) in isopropanol and absorbance was measured by spectrophotometry at 520 nm.

Cell lysates (equal aliquots of 50 - 100 μg) in lysis buffer (20 mM Tris-HCl, pH 7.8, 0.5% SDS containing protease inhibitors) were subjected to Western blot analyses [20] by 7.5% - 12% SDS-PAGE, transferred to PVDF membrane by electroblotting and analyzed by immunodetection using anti-myosin heavy chain II (MHCII) antibody (1:300 dilution) (Santa Cruz Biotechnology, Santa Cruz, CA). Washed filters were incubated with 1:2000 dilution of secondary antibody of goat anti-rabbit IgG linked to horseradish peroxidase. Immuno-reactive bands were visualized using the ECL detection system (Amersham Biosciences, Piscataway, NJ).

Total RNA was extracted from cell cultures by using Trizol-reagent (Invitrogen, Carlsbad, CA) and was purified and characterized by measuring at A₂₆₀. Two micrograms aliquot of total RNA was reversely transcribed and the resulting cDNA was amplified for 35 PCR cycles by melting at 94˚C for 30 sec, primer annealing at 58˚C for 30 sec and extension at 72˚C for 1 min. PCR products were analyzed by electrophoresis in 1.5% agarose gel using ethidium bromide staining (20). The locations of DNA sequences for the forward/reverse PCR primers are as follows: C/EBPα (225 bp), 843-864/1067-1047 on NM_007678; and GAPDH (152 bp) 606-626/758-738 on BC023196.

Data are presented as mean ± SEM. Pair-wise comparisons between groups were performed using Student’s T-test. All comparisons were two-tailed and P value less than 0.05 were considered statistically significant. The experiments were repeated a minimum of two times for consistency and representative cell pictures from experiments are shown.

Initially the effect of AD on adipogenesis was studied using 10T1/2 cells induced to differentiate with 5-azacy- tidine and then treated with or without DHT or with alcohol vehicle which served as control. AD was added in a dose-dependent manner to 10T1/2 cells during differentiation of adipocytes. With increasing AD concentrations, a dose-dependent decrease in adipocyte numbers were seen in 10 nM, 100 nM and 1 μM similar to the inhibition observed with 10 nM DHT. However, at high concentration of 10 μM, suppression of adipogenesis was relieved back to the control level, as shown in the representative photomicrographs (Figure 1(a)). This biphasic effect on adipogenesis was also correlated quantitatively by counting Oil Red O stained adipocytes in wells (Figure 1(b)) and also by RT-PCR for C/EBPα expression. Analysis of mRNA for the key adipogenic transcription factor C/EBPα by RT-PCR again showed inhibition of gene expression at 1 μM AD, but relief from inhibition at the higher 10 μM concentration of AD (Figure 1(c)).

Biphasic effect on adipogenesis suggested that the action of AD could be mediated through androgen receptor (AR). So, adipogenic experiment was repeated with AD along with androgen receptor antagonist bicalutamide. When cells were incubated with 300 nM AD along with increasing concentrations of bicalutamide (Bic) up to 1 μM, there was a dose dependent increase in fat cells as shown in the representative stained fat cell pictures (Figure 2(a)). This increase in fat cell numbers was correlated by counting Oil Red O stained adipocytes in wells (Figure 2(b)) at the cellular level and by RT-PCR for key adipogenic marker C/EBPα expression

(Figure 2(c)) at the molecular level. Experiments indicated AD biphasic effect on adipogenesis was mediated through AR. So, lack of suppression of adipogenesis seen at 10 μM concentration of AD had to do with AD binding to AR. In fact, Chen et al. [23] showed a partial agonist/antagonist nature of AD binding to AR, based on cell transcription assay.

Since 10 μM concentration of AD did not inhibit adipogenesis as expected for an androgen, we checked the effect of AD on myogenesis with increasing concentrations up to the high 10 μM concentration of AD. Myogenesis was increased in a dose dependent manner even at the highest (10 μM) concentration of AD. This was confirmed by immunostaining for the key myotube protein myosin heavy chain II (MHCII) at the cellular level and by Western blot for MHCII protein at the molecular level (Figure 3(a)). This experiment was repeated with the addition of increasing concentrations of bicalutamide along with 300 nM AD. Bicalutamide addition decreased myogenesis drastically as confirmed by immunostaining for MHCII protein at the cellular and by Western blot at the protein level (Figure 3(b)). Androgen receptor antagonist bicalutamide substantially blocked the formation of myotubes at 1 μM concentration.

Above observation led to the comparison of myogenesis and adipogenesis at 10 μM concentration of AD. At 10 μM concentration of AD, myogenesis was increased, but adipogenesis was not suppressed as expected for an androgen (Figure 4(a)). But, if androstenedione (300 nM) was incubated with bicalutamide (1 μM), myogenesis was suppressed, whereas adipogenesis was not suppressed (Figure 4(b)). This stands to reason, because AD action was mediated through AR and when AR was blocked, myogenesis was suppressed and adipogenesis was not suppressed again pointing to the role of AD binding to AR. We resorted to colorimetric method for quantitation of fat cells not only to remove any error crept in due to manual counting of stained fat cells, but also to compare quantitatively adipogenesis by another weak androgen and also adipogenesis in another adipogenic cell line. For colorimetric quantitation, Oil Red O stained fat cells were extracted with isopropanol containing Igepal and absorption was read at 520 nm. There was suppression of adipogenesis at 10 nM, 100 nM and 1 μM concentrations of AD, but not at 10 μM concentration of AD (Figure 4(c)). We repeated adipogenic experiment in 10T1/2 cells with another weak androgen DHEA to check whether this biphasic effect on adipogenesis was due to a spurious effect of AD or due to a common effect of weak androgens. Colorimetric quantitation of fat cells showed the same pattern of adipogenesis with DHEA. Inhibition of adipogenesis at 10 nM, 100 nM and 1 μM of DHEA and a reversal back to the control level of adipogenesis at 10 μM concentration of DHEA (Figure 4(c)), indicating it was a common effect of weak androgen on adipogenesis in C3H10T1/2 cells. As a follow-up of the above experiment, we tested the effect of AD on adipogenesis in another adipogenic cell line, the well-studied preadipocyte 3T3-L1 cells [21] . Adipogenesis was quantitated by colorimetric measurement. There was suppression of adipogenesis at 1 μM of AD and non-suppression at 10 μM of AD (Figure 4(d)). In 3T3-L1 cell line also a reversal back to the control level was reproducibly observed at high concentration (10 μM) of AD. So AD was unable to suppress adipogenesis at high concentration in another adipogenic (3T3-L1) cell line, which was shown to have AR [21] . Thus, indicating that the biphasic effect was an inherent biological effect of weak androgen AD on adipogenesis.

Since AD was a metabolic precursor to T, it was expected that T could increase myogenesis and decrease adipogenesis. However, T can also be converted to estrogen (E₂) by the enzyme aromatase. In the clinical experiment treating humans with AD, there were increases in estrone and estradiol levels [17] . So we reasoned that at 10 μM concentration of AD, the increase in estrone and estradiol might be contributing to the increase in adipogenesis in C3H10T1/2 cells. Hence, we used aromatase enzyme inhibitor (AI) to block endogenous estrogen production in the cell. We carried out adipogenic experiments in the presence of AD and AD + AI. There was no significant difference in adipogenesis between cells treated with AD or with AD + AI (Figure 5(a)). So estrogen

had no action on adipogenesis in 10T1/2 cells. In fact when estrogen was added directly to 10T1/2 cells, there was no increase in adipogenesis compared to untreated control cells. In order to make sure that the lack of increase inadipogenesis in the treated cells was not due to an increase in myogenesis mediated by AI, we also checked myogenesis by immunohistochemical staining for MHCII protein in myotubes. Again there were no appreciable differences in myogenesis between AD treated cells and AD + AI treated cells (Figure 5(b)). By

these control experiments, the role of estrogen contributing to adipogenesis or AI contributing to myogenesis was ruled out. So, the increase back to the control level of adipogenesis at 10 μM of AD was not due to estrogen, but due to AD only.