Peroxisome Proliferator-activated Receptor γ (PPARγ) as a Molecular Target for the Soy Phytoestrogen Genistein

The principal soy phytoestrogen genistein has an array of biological actions. It binds to estrogen receptor (ER) α and β and has ER-mediated estrogenic effects. In addition, it has antiestrogenic effects as well as non-ER-mediated effects such as inhibition of tyrosine kinase. Because of its complex biological actions, the molecular mechanisms of action of genistein are poorly understood. Here we show that genistein dose-dependently increases estrogenic transcriptional activity in mesenchymal progenitor cells, but its biological effects on osteogenesis and adipogenesis are different. At low concentrations (≤1 μm), genistein acts as estrogen, stimulating osteogenesis and inhibiting adipogenesis. At high concentrations (>1 μm), however, genistein acts as a ligand of PPARγ, leading to up-regulation of adipogenesis and down-regulation of osteogenesis. Transfection experiments show that activation of PPARγ by genistein at the micromolar concentrations down-regulates its estrogenic transcriptional activity, while activation of ERα or ERβ by genistein down-regulates PPARγ transcriptional activity. Genistein concurrently activates two different transcriptional factors, ERs and PPARγ, which have opposite effects on osteogenesis or adipogenesis. As a result, the balance between activated ERs and PPARγ determines the biological effects of genistein on osteogenesis and adipogenesis. Our findings may explain distinct effects of genistein in different tissues.

In recent years, soy phytoestrogens have attracted wide attention due to their potential beneficial effects on some common medical disorders (1)(2)(3). Genistein, the principal soy phytoestrogen, has an array of biological actions and is widely available in herbal tablets (3)(4)(5). It binds to estrogen receptors (ERs), 1 ER␣ and ER␤, and has ER-mediated effects (6,7). In addition, it has antiestrogenic effects, but the underlying mechanism is still unknown (1,2,4). Non-ER mediated genistein actions such as an inhibition of protein tyrosine kinase, DNA topoisomerases I and II and ribosomal S6 kinase have also been reported (8 -10). These actions are most likely mediated through transcriptional processes rather than via direct effects on enzyme activity (11,12).
Peroxisome proliferator-activated receptor-␥ (PPAR␥), one of the subtypes of PPARs, is a ligand-dependent transcription factor of the nuclear hormone receptor superfamily (13). PPAR␥ is most highly expressed in adipose tissue and is involved in critical physiological functions such as adipogenesis and glucose and cholesterol metabolism (14). It is a target for therapeutic intervention in cardiovascular diseases, various cancers, and diabetes (15).
PPAR␥ is the essential transcriptional factor for adipogenesis (16 -18). Adipocytes and the bone-forming cells, the osteoblasts, arise from the same bone marrow mesenchymal precursor cells (19,20). The osteoprogenitor KS483 cells, which are cloned from mouse calvaria (21,22), have been shown to differentiate into both osteoblasts and adipocytes. Using this cell line, we recently showed that 17␤-estradiol (E 2 ) stimulates osteogenesis and concurrently inhibits adipogenesis in these precursor cells (23). Whether the phytoestrogen genistein has similar effects is unknown.
In the present study, we examined the effects of genistein on osteogenesis and adipogenesis and explored its molecular mechanisms of action. Our results show that genistein, in addition to its estrogenic activity, activates PPAR␥, resulting in a down-regulation of osteogenesis and an up-regulation of adipogenesis. This action is concentration-dependent. Our data show that the balance between activated ERs and PPAR␥ determines the biological effects of genistein.

EXPERIMENTAL PROCEDURES
Cell Cultures and Assays-The methods for cell culture have been described before (23). In brief, KS483 cells and mouse bone marrow cells were cultured in phenol red-free ␣-minimum essential medium (␣-MEM) supplemented with 10% fetal bovine serum (Invitrogen) or 15% fetal bovine serum (for mouse bone marrow), 50 g/ml ascorbic acid, 10 mM ␤-glycerophosphate, and 10 Ϫ8 M dexamethasone (only for mouse bone marrow). Cells were continuously exposed to genistein 1 day after plating until the end of the experiment at day 21. Assays for ALP activity and DNA content, mRNA expression by RT-PCR, and Oil-Red-O staining for adipocytes were performed as described previously (23).
Membrane-bound PPAR␥ Binding Assay-Binding assays, using a human full-length PPAR␥ construct expressed in bacteria, were performed in 96-well plates (24). Binding buffer consisted of 10 mM Tris/ HCl, pH 8.2, containing 50 mM KCl and 1 mM dithiothreitol. Membrane preparations (5 g/ml) were incubated for 180 min at 4°C in the presence of [ 3 H]rosiglitazone (BRL49653, Amersham Biosciences) (10 nM) and the tested compounds. Nonspecific binding was defined using an excess of unlabelled rosiglitazone (10 M). Incubation was terminated by the addition of ice-cold 50 mM Tris/HCl buffer, pH 7.4, followed by rapid filtration under reduced pressure through Whatman GF/C filter plates presoaked with ice-cold buffer, followed by three successive * 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.
Transient Gene Expression Assays in KS483 Cells-The estrogenresponsive reporter gene construct (2XERE-TATA-luc), which contains two copies of a consensus estrogen response element, and the empty control TATA-luc plasmids were kindly provided by Dr. E. Kalkhoven and Dr. M. G. Parker. The peroxisome proliferator-responsive element (3XPPRE-tk-luc) containing three copies of a consensus peroxisome proliferator-responsive element and the human PPAR␥2 constructs were kind gifts from Dr. J. Auwerx. The luciferase reporter construct (5XPPRE-TATA-luc) contained five copies of a consensus PPRE and a TATA box and were provided by Dr. M. Karperien. The pT-109 FARE PPRE construct was a kind gift from Dr. K. van der Lee and Dr. M. van Bilsen. The ACO-luc PPRE construct was kindly supplied by Dr. K. W. Kinzler and Dr. B. Vogelstein. The human ER␣ construct was kindly provided by Dr. G. Kuiper. KS483 cells were seeded into 24-well plates. After 24 h, they were transfected using a lipid-based FuGENE 6 transfection reagent according to the manufacturer (Roche Molecular Biochemicals). For each triplicate of sample, 100 ng of luciferase reporter and 500 ng of ␤-galactosidase expression vector were applied. The transfection medium was changed after 16 h into the different medium as indicated. After 48 h, cells were washed twice with PBS, lysed in PBS containing 1% Triton X-100 and sonicated. Luciferase activity was measured and expressed as fold induction Ϯ S.E., which was corrected for transfection efficiency using ␤-galactosidase activity.
Statistics-Data are presented as means Ϯ S.E. Differences between groups were accepted at p Ͻ 0.05, which were assessed by one-way analysis of variance or related test using software Instat. Fig. 1, genistein added to cultures of KS483 cells had a clear biphasic effect on osteogenesis, similar to that of E 2 (23). At concentrations from 0.1 to 10 M, genistein stimulated ALP activity, nodule formation, and calcium deposition, with a maximal effect at 1 M. In contrast, at concentrations of 25 M or higher, genistein inhibited ALP activity, nodule formation and Ca 2ϩ deposition. These changes were paralleled by mRNA expression of the osteoblastic markers, Cbfa1, osteocalcin, and PTH/PTHrP receptor that, relative to control, were increased by 1 M genistein and decreased by 25 M (Fig. 1). Similar stimulatory and inhibitory effects of genistein on bone formation were also observed in mouse bone marrow cell cultures (Fig. 2). In those cultures, genistein stimulated ALP activity and Ca 2ϩ deposition at concentrations between 0.1 and 10 M, whereas it inhibited osteogenesis at concentrations of 25 M or higher. These data demonstrate that genistein affects osteogenesis of progenitor cells in a biphasic way; namely, it increases osteogenesis at low concentrations and inhibits osteogenesis at high concentrations.

FIG. 2. Effects of genistein on osteogenesis in mouse bone marrow cells.
Mouse bone marrow cells were cultured in 12-well plates in the medium containing charcoal-stripped serum and continuously exposed to different concentrations of genistein for 21 days. Cellular ALP activity (A) and calcium deposition (B) were quantified. Each value is the mean Ϯ S.E. of the results from three different wells and is representative of results from at least five different experiments. Significant differences (* Ͻ 0.05) were indicated.

FIG. 1. Effects of genistein on osteogenesis in KS483 cells.
KS483 cells were cultured in 12-well plates in the medium containing charcoal-stripped serum and continuously exposed to different concentrations of genistein for 18 days. Cellular ALP activity (A), the number of nodules (B), calcium deposition (C), and mRNA expression of osteogenic markers Cbfa1, osteocalcin, and PTHrP-R (D) were quantified. Each value is the mean Ϯ S.E. of the results from three different wells and is representative of results from at least five different experiments. Significant differences (* Ͻ 0.05) were indicated.
Adipogenesis-Genistein had also a biphasic effect on adipogenesis, which was, however, different to that of E 2 (23). At low concentrations between 0.1 and 1 M, it decreased adipocyte numbers, while at higher concentrations (Ͼ10 M) it stimulated adipogenesis (Fig. 3A). The effects of genistein on adipogenesis were paralleled by changes in mRNA expression of the adipocyte markers, PPAR␥2, aP2, and lipoprotein lipase (Fig.  3B). Adipogenic responses of mouse bone marrow cells to different doses of genistein are shown in Fig. 3C. Mouse bone marrow cultures treated with genistein concentrations of 25 M or higher did not reach confluence, and there were no adipocytes during the cultures. However, compared with control an increase in adipocyte numbers was observed at the concentration of 10 M, whereas a decrease in adipocyte numbers was found at the concentrations of 0.1 and 1 M. These data show that genistein affects adipogenesis of progenitor cells in a biphasic way, i.e. an inhibition of adipogenesis at low concentrations and a stimulation of adipogenesis at high concentrations.
ER-dependent and ER-independent Effects of Genistein-Both ER-dependent and ER-independent effects were observed in KS483 cells treated with different concentrations of genistein (Fig. 4). At a concentration of 1 M, the effects were mediated by ERs because stimulation of ALP activity and inhibition of adipogenesis were both blocked by 1 M ICI 164,382, a specific antiestrogen. In contrast, at higher concentrations of genistein the effects observed were ER-independent because ICI 164.382 at concentrations from 0.01 to 100 M did not affect the action of genistein on osteogenesis or adipogenesis. In addition, E 2 (10 Ϫ10 M to 10 Ϫ5 M) did not reverse the effects of genistein at 25 M on osteogenesis or adipogenesis. These data suggest that the action of genistein at low concentrations is likely ER-mediated, whereas its effects at high concentrations are not ER-mediated.
Activation of PPAR␥-We transiently transfected KS483 cells with a luciferase reporter construct containing five copies of a consensus PPRE inserted in front of a TATA box together with expression plasmids encoding human PPAR␥2. PPRE-luc reporter activity was measured after incubation of transfected cell cultures with different doses of genistein. As shown in Fig 5A, genistein in the micromolar range increased PPRE-luc reporter activity dose-dependently. Furthermore, in the same concentration range, genistein increased PPRE-luc reporter activity in ER-positive and ER-negative breast cancer cell lines, T47D and MDA-MD-231, respectively (not shown). These results were confirmed with three other reporter constructs including the PPAR␥ response element ACO-luc (25). Thus, genistein transcriptionally activates PPRE-luc reporter activity independent of the cell lines and constructs used.
To determine whether genistein activates PPAR␥ through direct interaction with this receptor, we performed a membrane-bound PPAR␥ binding assay. Genistein had a measurable K i of 5.7 M (Fig. 5B), which is comparable to that of the known PPAR␥ ligands (24). We have checked whether genistein bound competitively with [ 3 H]rosiglitazone to the same PPAR␥ site. Indeed the dissociation constant (K d ) of [ 3 H]rosiglitazone in saturation experiments in the presence of a high dose of genistein was significantly reduced as compared with that in the absence of genistein. The maximal number of sites labeled was not altered. These data demonstrate that both Balance between Activated ERs and PPAR␥-As both the antiestrogenic effects and the activation of PPAR␥ were increased by micromolar concentrations of genistein, we tested whether activation of PPAR␥ is involved in the antiestrogenic action. When KS483 cells were treated for 18 days either with a specific PPAR␥ agonist ciglitazone, genistein, E 2 , or a combination of genistein and E 2 , a decrease in ALP activity was observed with all treatments, except for E 2 alone that increased ALP activity (Fig. 6A). When, however, we transiently transfected KS483 cells with a luciferase reporter construct containing two copies of a consensus ERE inserted in front of a TATA box and exposed these cells to different concentrations of genistein, a dose-related increase of ERE-luc reporter activity was observed at a concentration between 0.1 and 50 M (Fig.  6B). Furthermore, the estrogenic potency of genistein at the micromolar range was greater than that of E 2 (10 Ϫ8 M), and anti-estrogenic effects of genistein were not observed (Fig. 6C). The lack of an antiestrogenic effect in the gene reporter assays could be due to low amount of endogenous PPAR␥2 in KS483 cells during the first 5 days (23), the period in which the gene reporter assays were performed. To investigate this further, we transiently transfected vectors expressing human PPAR␥2 or empty vector along with a ERE-luc construct and exposed KS483 cells to genistein. The transient co-transfection of PPAR␥2 resulted in a decrease of ERE-luc reporter activity at high genistein concentrations. Taken together, our data show that the antiestrogenic effects of genistein are due to an activation of PPAR␥2, leading to down-regulation of ER-mediated transcriptional activity and osteogenesis.
The question that arises is whether activation of ERs by genistein can also alter the transcriptional regulation of PPAR␥. To investigate this, we transiently transfected vectors expressing human ER␣ or ER␤ or empty vector along with a PPAR␥2 construct and a PPRE-luc construct. Co-transfection of ER␣ (Fig. 7A) or ER␤ (Fig. 7B) decreased PPRE-luc reporter activity in KS483 cells treated with different concentrations of genistein. Interestingly, genistein at 1 M suppressed PPREluc reporter activity to a level lower than that of controls, while at 10 M it suppressed it to the control level in the presence of sufficient levels of ERs. In contrast, PPRE-luc reporter activity was higher than the control levels at 50 M genistein of and was not influenced by the levels of ER␣ or ER␤ (Fig. 7). DISCUSSION We show here that PPAR␥ is a molecular target for genistein. At the micromolar range, genistein binds to and transactivates PPAR␥, leading to a decrease of osteogenesis and an increase in adipogenesis. In addition, genistein dose-dependently transactivates ERs, resulting in an up-regulation of osteogenesis and a down-regulation of adipogenesis. Moreover, activation of ERs by genistein could down-regulate PPAR␥ transcriptional activity and vice versa. The balance between the activation of ERs and PPAR␥ is concentration-related. As a result, the biological effects, i.e. osteogenesis and adipogenesis, vary according to the concentrations of genistein (Fig. 8). Our findings can explain the previously reported diverse actions of genistein in different tissues.
At low concentrations (Յ1 M), genistein has ER-dependent effects on osteogenesis and adipogenesis; the effects are similar to those of E 2 (23). At high concentrations (Ͼ1 M), however, genistein has antiestrogenic actions, namely, it down-regulates osteogenesis, which is opposite to E 2 -induced effects. Antiestrogenic effects of genistein have been reported in many cell types and animal models, but the mechanism responsible for this is still not known (1,2,4,26). We show here that the antiestrogenic effects are not due to a decrease of estrogenic activity of genistein. Instead, genistein at micromolar concentrations dose-dependently increased estrogenic transcriptional activity, and the levels were even higher than those induced by E 2 . These results are in line with reports using different cell lines or assays (6,27,28). Moreover, antiestrogenic effects of genistein could not be restored or blocked by E 2 or by the antiestrogen compound ICI164,382. Together, our results implicate that antiestrogenic effects of genistein are elicited via pathways other than the ER pathway.
Different from E 2 , genistein binds to and transactivates PPAR␥, leading to adipogenesis. Moreover, activation of PPAR␥ may also be due to an inhibition of the MAPK pathway. It is well known that the A/B domain of PPAR␥ contains a consensus MAPK site (29 -31). Inhibition of PPAR␥ phosphorylation by the specific MAPK inhibitor PD98059 stimulates adipogenesis (32). Genistein inhibits p42/44 MAPKs in KS483 cells. 2 It is therefore possible that an inhibition of p42/44 MAPKs contributes to an activation of PPAR␥. By using a pure PPAR␥ ligand, ciglitazone, we showed that activation of PPAR␥ down-regulates osteogenesis in KS483 cells. These results are consistent with observations in MC3T3-E1 cells and in U33 cells (33,34). It has been shown that PPAR␥2 plays a dominant role in the determination of the fate of mesenchymal progenitor cells (35). An increase in adipogenesis and a decrease of osteogenesis by genistein at concentrations of 25 M or higher indicate that PPAR␥ actions dominate at higher genistein concentrations.
Genistein tional factors, ERs and PPAR␥. These two transcriptional factors have opposite effects on osteogenesis or adipogenesis. We showed that activation of PPAR␥ by genistein at the micromolar concentrations down-regulates its estrogenic transcriptional activity, while activation of ER␣ or ER␤ down-regulates PPAR␥ transcriptional activity. It is plausible that genistein at certain concentrations activates ERs and PPAR␥ to a different extent. The balance between activated ERs and PPAR␥ determines the biological effects of genistein, i.e. osteogenesis and adipogenesis, which are fully concentration-dependent.
Our findings provide the molecular basis of the mechanism of action of genistein and may have wide implications. Diverse effects of genistein in different tissues have been explained by the high binding affinity for ER␤ because ER␤ can act as a dominant negative regulator of estrogenic activity. These dominant negative effects were only observed below the micromolar concentrations of genistein (36). However, the distinct genistein effects in different tissues are often observed at the micromolar concentrations (1,2,4,37). We show that the balance between activated ERs and PPAR␥ determines the biological effects of genistein, which might explain its diverse biological effects in different organs. Therefore, the biological effects of genistein in certain tissues strongly depend on the concentration of genistein present and the levels of ERs and PPAR␥ within that particular tissue. There is accumulating evidence that health benefits occur only when phytoestrogens are consumed in sufficient quantities (1,2,4). It has been reported that plasma concentration of genistein is relatively low and generally less than 40 nM in humans consuming diets without soy, whereas it can reach 4 M in the plasma of Japanese who consume high amount of soy products (1, 2, 4). Our findings FIG. 6. Antiestrogenic effects of genistein in KS483 cells. KS483 cells were cultured in 12-well plates in the medium containing charcoal-stripped serum and continuously exposed to genistein of 25 M, E 2 (10 Ϫ8 M), and ciglitazone of 25 M as well as combination of genistein and E 2 for 18 days. ALP activity (A) was measured, which shows that genistein exerts antiestrogenic effects on osteogenesis. KS483 cells containing an integrated ERE-luc reporter gene were exposed to various concentrations of genistein for 48 h. Genistein dose-dependently increased ERE-luc activity (B). When these cells were exposed to E 2 or genistein of 25 M or in combination of both for 48 h, antiestrogenic effects of genistein were not observed in this gene reporter assay (C). When KS483 cells were cotransfected with ERE-luc reporter gene together with constructs expressing PPAR␥2 and exposed to different concentrations of genistein, down-regulation of ERE-luc activity was observed (D). C, control; E, E 2 ; G, genistein; Ci, ciglitazone.
FIG. 7. Effects of ER␣ and ER␤ on PPAR␥ transcriptional activity. KS483 cells were seeded into 24-well plates and transiently transfected with a luciferase reporter construct containing five copies of a consensus PPRE inserted in front of TATA box together with expression plasmids encoding human PPAR␥2 and in combination with ER␣ or ER␤. PPRE-luc reporter activity was measured after incubation of transfected cell cultures with different doses of genistein for 48 h. Cotransfection of ER␣ (A) or ER␤ (B) down-regulates PPAR␥ transcriptional activity. Each value is the mean Ϯ S.E. of the results from three different wells and is representative of results from at least three different experiments. Significant differences (* Ͻ 0.05) were indicated. might explain why genistein functions only at a certain level. For example, genistein at the micromolar concentration range inhibits growth of ER-positive breast cancer cells like MCF7 and T47 D as well as ER-negative breast cancer cells like MDA-MD-231 cells (38). Since it is now well established that ligand activation of PPAR␥ inhibits cell growth and induces apoptosis in these cancer cells (39 -41), it is plausible that only when PPAR␥ is activated, genistein at certain levels could inhibit the growth of cancer cells.