Dehydrodiconiferyl Alcohol Isolated from Cucurbita moschata Shows Anti-adipogenic and Anti-lipogenic Effects in 3T3-L1 Cells and Primary Mouse Embryonic Fibroblasts*

Background: A water-soluble extract, prepared from Cucurbita moschata, has potent anti-obesity activities. Results: Dehydrodiconiferyl alcohol (DHCA), isolated from the extract, inhibited the mitotic clonal expansion by suppressing the DNA binding activity of C/EBPβ and directly inhibited the expression of the regulators of lipogenesis in 3T3-L1 and primary embryonic fibroblasts. Conclusion: DHCA may contain the anti-adipogenesis as well as anti-lipogenesis. Significance: DHCA might have potential as an anti-obesity therapeutic. A water-soluble extract from the stems of Cucurbita moschata, code named PG105, was previously found to contain strong anti-obesity activities in a high fat diet-induced obesity mouse model. One of its biological characteristics is that it inhibits 3T3-L1 adipocyte differentiation. To isolate the biologically active compound(s), conventional solvent fractionation was performed, and the various fractions were tested for anti-adipogenic activity using Oil Red O staining method. A single spot on thin layer chromatography of the chloroform fraction showed a potent anti-adipogenic activity. When purified, the structure of its major component was resolved as dehydrodiconiferyl alcohol (DHCA), a lignan, by NMR and mass spectrometry analysis. In 3T3-L1 cells, synthesized DHCA significantly reduced the expression of several adipocyte marker genes, including peroxisome proliferator-activated receptor γ (Pparg), CCAAT/enhancer-binding protein α (Cebpa), fatty acid-binding protein 4 (Fabp4), sterol response element-binding protein-1c (Srebp1c), and stearoyl-coenzyme A desaturase-1 (Scd), and decreased lipid accumulation without affecting cell viability. DHCA also suppressed the mitotic clonal expansion of preadipocytes (an early event of adipogenesis), probably by suppressing the DNA binding activity of C/EBPβ, and lowered the production level of cyclinA and cyclin-dependent kinase 2 (Cdk2), coinciding with the decrease in DNA synthesis and cell division. In addition, DHCA directly inhibited the expression of SREBP-1c and SCD-1. Similar observations were made, using primary mouse embryonic fibroblasts. Taken together, our data indicate that DHCA may contain dual activities, affecting both adipogenesis and lipogenesis.

A water-soluble extract from the stems of Cucurbita moschata, code named PG105, was previously found to contain strong anti-obesity activities in a high fat diet-induced obesity mouse model. One of its biological characteristics is that it inhibits 3T3-L1 adipocyte differentiation. To isolate the biologically active compound(s), conventional solvent fractionation was performed, and the various fractions were tested for antiadipogenic activity using Oil Red O staining method. A single spot on thin layer chromatography of the chloroform fraction showed a potent anti-adipogenic activity. When purified, the structure of its major component was resolved as dehydrodiconiferyl alcohol (DHCA), a lignan, by NMR and mass spectrometry analysis. In 3T3-L1 cells, synthesized DHCA significantly reduced the expression of several adipocyte marker genes, including peroxisome proliferator-activated receptor ␥ (Pparg), CCAAT/enhancer-binding protein ␣ (Cebpa), fatty acid-binding protein 4 (Fabp4), sterol response element-binding protein-1c (Srebp1c), and stearoyl-coenzyme A desaturase-1 (Scd), and decreased lipid accumulation without affecting cell viability. DHCA also suppressed the mitotic clonal expansion of preadipocytes (an early event of adipogenesis), probably by suppressing the DNA binding activity of C/EBP␤, and lowered the production level of cyclinA and cyclin-dependent kinase 2 (Cdk2), coinciding with the decrease in DNA synthesis and cell division. In addition, DHCA directly inhibited the expression of SREBP-1c and SCD-1. Similar observations were made, using primary mouse embryonic fibroblasts. Taken together, our data indicate that DHCA may contain dual activities, affecting both adipogenesis and lipogenesis.
Obesity, a major health concern of the 21st century, serves as a risk factor for various diseases, including diabetes, hypertension, arthritis, and coronary artery disease (1)(2)(3)(4). Obesity can be described as a state of excessive growth of adipose tissue mass. Adipocyte is a major cellular component in adipose tissue, and thus, the mechanism regulating adipocyte size and number has been an important target for obesity research (5)(6)(7)(8). Because excessive energy is stored in adipocytes as triglyceride, the expansion of adipocyte size (hypertrophy) is proportional to the amount of intracellular lipid accumulation, and it is dependent on adipocyte lipogenic activity (9,10). However, adipocytes cannot be enlarged and accumulate lipids indefinitely. Consequently, hypertrophy is an initial event of obesity development, whereas an increase in cell number (hyperplasia) becomes a more influential factor in the chronic state of obesity (1,9,10).
Adipocyte hyperplasia depends on adipogenesis, the generation of new adipocytes from precursor cells, and research on the area of adipogenesis has been facilitated by the establishment of the murine preadipocyte cell line 3T3-L1, which can be induced to differentiate into mature adipocytes in vitro (11)(12)(13). Preadipocyte differentiation program involves several stages (14 -16). Upon reaching confluence, proliferative preadipocytes become growth-arrested by contact inhibition (17,18). When exposed to the differentiation inducer consisting of insulin at a nonphysiologically high concentration, dexamethasone (a glucocorticoid) and 1-methyl-3-isobutylxanthine (MIX 4 ; a cAMP phosphodiesterase inhibitor that increases intracellular cAMP concentration) cells re-enter the cell cycle (18 -20). After this proliferation stage, referred to as a mitotic clonal expansion, the cells undergo terminal differentiation. Whether mitotic clonal expansion is required for adipocyte differentiation remains controversial (16,19,21); however, it is generally accepted that DNA unwinds, and remodeling during the mitotic clonal expansion stage is involved in the initiation of expression of various adipogenic genes (22,23).
Many reagents have been found or developed to modulate adipogenesis (24,25). For example, U0126, a MEK inhibitor, decreases the expression of cell cycle markers such as cyclinA and Cdk2, followed by inhibition of mitotic clonal expansion. It also suppresses the expression of major adipogenic genes such as Pparg and Cebpa and subsequently inhibits adipocyte differentiation (10,19). Roscovitine, a Cdk inhibitor, suppresses DNA replication and cell proliferation, which inhibits mitotic clonal expansion and consequently results in blocking the progression of the adipogenesis program (16,19). In addition to these agents targeting early event of adipogenesis, anti-adipogenic compounds inhibiting terminal differentiation, a late event of adipogenesis program, have also been reported. The protein-tyrosine phosphatase inhibitor vanadate is known to specifically inhibit terminal differentiation by decreasing expression of adipogenic genes and reducing the accumulation of cytoplasmic triglyceride (26,27).
Previously, we observed that PG105, a water-soluble extract from the stem parts of Cucurbita moschata, suppresses 3T3-L1 cell differentiation (28). In this study, we show that dehydrodiconiferyl alcohol (DHCA), isolated from the chloroform fraction of PG105, has potent anti-adipogenic and anti-lipogenic activities. DHCA is a type of lignan belonging to the phytoestrogen family (29,30). Although many lignans have been shown to possess a variety of biological activities and are described as anti-oxidant and anti-fungal compounds (31)(32)(33), their working mechanism is largely unknown. In this study, we found that DHCA inhibits adipocyte differentiation and intracellular lipid accumulation in the 3T3-L1 cell culture system and primary mouse embryonic fibroblasts (MEFs); furthermore, it affects the expression of various genes involved in adipogenesis or lipogenesis at the RNA level. DHCA also regulated the production of selective cell cycle markers affecting mitotic clonal expansion by suppressing the DNA binding activity of C/EBP␤. These data suggested that DHCA might have a potential to be developed as a reagent controlling the fat accumulation.

EXPERIMENTAL PROCEDURES
Solvent Extraction-Dried stems of C. moschata (2 kg) purchased from a farm (Jinju, Korea) were crushed and extracted with boiling water for 3 h three times. It was then concentrated and freeze-dried to obtain PG105. The yield of PG105 from the dried stems was estimated to be 14%. PG105 (200 g) was dissolved in sterile distilled water and extracted with n-hexane, chloroform, ethyl acetate, and n-butyl alcohol in a sequential manner. Each solvent-soluble fraction and the final aqueous residue were filtered, concentrated, freeze-dried, and dissolved in ethanol at a concentration of 100 mg/ml. All preparations were stored at Ϫ70°C until use. The isolation scheme and the yield of each fraction from the dried stems are shown in Fig. 1.
Isolation and Purification of Active Compound-Chloroform-soluble fraction was loaded onto a silica gel column (130 g, 3 ϫ 27 cm, Merck), and 11 fractions were eluted with a mixture of n-hexane, chloroform, and methanol (hexane/CHCl 3 / MeOH, 16:15:1 (v/v/v)). The purity of each fraction was checked by TLC. Two types of solvent systems (n-hexane/ CHCl 3 /MeOH, 10:9:1 (v/v/v), and n-hexane/CHCl 3 /MeOH, 4:3:1 (v/v/v)) were used. The spots were visualized by exposure to 10% H 2 SO 4 spray. Chloroform fraction number 9 was rechromatographed using the silica gel column and eluted with a mixture of chloroform and methanol (CHCl 3 /MeOH, 30:1 (v/v)), obtaining three different fractions. The third fraction with the highest bioactivity was further purified by semi-preparative HPLC using -Bondapak C18 column (7.8 ϫ 300 mm, Waters), and its final eluate was named CMC-9. The mobile phase consisted of 40% methanol in distilled water, and its flow rate was 2 ml/min. The purity of the major compound of CMC-9 was established by HPLC using Inertsil ODS-2 column (4.5 ϫ 150 mm, GL Sciences, Torrance, CA). The mobile phase consisted of 20% methyl cyanide in distilled water. The major peak was detected at 15.12 min by a 254-nm UV detector, and its purity was estimated to be 93% by peak area. The structure of active compound was elucidated as dehydrodiconiferyl alcohol by the analyses of NMR, mass, and IR spectral data. The molecular formula was deduced as C 20 H 22 O 6 by high resolution electron ionization mass data. The 1 H and 13 C NMR data showed characteristic signals of a lignan skeleton having 2 units of phenylpropanoid with two methoxy groups in the aromatic rings. The link between two aromatic rings was confirmed by HMBC spectra.
Cell Culture and Differentiation of 3T3-L1 Cells-Cell culture and differentiation method followed the protocol described previously (28). Briefly, 3T3-L1 preadipocytes obtained from American Type Culture Collection (Manassas, VA) were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% (v/v) calf serum. Cells in passages 3-9 were used in all studies. To induce differentiation, 2-day postconfluent 3T3-L1 (designated day 0) was incubated in differentiation medium consisting of MDI (5 g/ml insulin, 1 M dexamethasone, 0.2 mM MIX), and 10% FBS in DMEM for 2 days. The cells were then incubated in DMEM supplemented with 10% FBS and 1 g/ml insulin for another 2 days, and the medium was replenished every other day for an additional 4 days. To study the effects of various fractions of C. moschata on adipocyte differentiation, different concentrations of the fractions were added along with the culture media over the entire time of the experiment. Synthetic DHCA was prepared as ϫ1000 stocks in ethanol and added to cells. Troglitazone and SB203580 purchased from Calbiochem were used as a negative and positive controls in various experiments, respectively.
Preparation and Differentiation of MEFs-Primary MEFs were prepared and used for cell differentiation experiments as described previously (47). Briefly, embryos at day 14 post-coitum were obtained, at which point the brains and dark red organs were removed. Embryos were then finely minced and digested with trypsin/EDTA (37°C) at 250 l per embryo with gentle shaking for 30 min. The reaction was stopped by adding an equal volume of cold PBS with 50% FBS. The solution was filtered through a Falcon 40-m cell strainer and then collected by centrifugation (1500 rpm for 2 min). Cells were washed twice with culture media ((DMEM) containing 10% FBS (Cellgro)) and then plated with warm media. The medium was changed 3 h later to remove unattached cells. Remaining cells were cultured and frozen for later use. To induce differentiation, 2-day postconfluent MEFs (designated day 0) were incubated in differentiation medium containing MDI (5 g/ml insulin, 1 M dexamethasone, 0.5 mM MIX), and 10% FBS in DMEM for 3 days. The cells were then incubated in the same DMEM, but lacking dexamethasone and MIX, for another 2 days, and the medium was replenished every other day for an additional 4 days. DHCA was added along with the culture media over the entire period of differentiation.
Oil Red O Staining-After the induction of differentiation, cells were washed with phosphate-buffered saline (PBS) and fixed with 10% formalin in PBS for 1 h, then washed an additional three times with water, and finally air-dried. Cells were stained with Oil Red O (6 parts of saturated Oil Red O dye (0.6%) in isopropyl alcohol and 4 parts of water) for 15 min. Excess of stain was removed by washing with 70% ethanol, and stained cells were then washed with water. To quantify the intracellular lipids, spectrophotometric quantification of the stain was performed by dissolving the stained lipid droplets with 4% Nonidet P-40 in isopropyl alcohol for 5 min. The absorbance of extracted dye was then measured at 520 nm.
LDH Assay-Cytotoxicity of DHCA was evaluated by colorimetric assay based on the measurement of LDH activity. Briefly, after various types of cells were treated with DHCA or Triton X-100 for 2 days, an aliquot of medium was taken and centrifuged at 2000 rpm for 10 min. Supernatant (100 l) was added to the LDH detection reagent (Takara Bio, Shiga, Japan) to make a total volume of 200 l. Spectrophotometric analysis was performed at room temperature (20 -24°C) using an ELISA microplate reader measuring absorbance at 490 nm. LDH release value was determined by calculating the average absorbance value of the triplicates and subtracting from each of these the absorbance value obtained in the background control (media only). The equation used for IC 50 calculation was as follows: viability (%) ϭ (1 Ϫ (experiment value Ϫ background value)/(Triton X-100 value Ϫ background value)) ϫ 100.
cDNA probes were radiolabeled with [␣-32 P]dCTP (3000 Ci/mmol, PerkinElmer Life Sciences) by using the Megaprime DNA labeling system (Amersham Biosciences). Total RNA was isolated from 3T3-L1 cells using the guanidinium isothiocyanate method, and purified by ultracentrifugation on a cesium chloride gradient; however, TRIzol reagent (Invitrogen) was used in the case of MEFs. Aliquots of 20 g of RNA were resolved by electrophoresis in agarose-formaldehyde gel and transferred to a nylon filter by overnight capillary blotting. The filters were hybridized with the respective 32 P-labeled probes (10 6 cpm/ml) for 1 h at 68°C with Rapid-Hyb buffer (Amersham Biosciences), washed with 0.1% SDS added in 0.2ϫ SSC at 55°C for 30 min, and exposed to Reflection NEF 496 film (PerkinElmer Life Sciences) with an intensifying screen at Ϫ80°C.
Electrophoretic Mobility Shift Assay (EMSA)-The CRE oligonucleotide (Santa Cruz Biotechnology) was radiolabeled with 32 P-labeled ATP (25 Ci/ml, PerkinElmer Life Sciences) by incubating with T4 polynucleotide kinase (Takara Bio) for 10 min at 37°C. After inactivation for 10 min at 65°C, a 32 P-end-labeled CRE probe was prepared by removing free isotope using mini Quick Spin Columns (Roche Applied Science). Nuclear proteins were extracted from 3T3-L1 cells using a Nuc-Buster TM extraction kit (Novagen, San Diego), quantified by the Bradford method, and incubated with the radiolabeled CRE probes and reagents composed of an EMSA accessory kit (Novagen) for 30 min at 4°C, according to the manufacturer's instructions. The DNA-protein complex was resolved in a 7% native polyacrylamide gel, and the radioactive bands on the gels were visualized by autoradiography film. An unlabeled (cold) CRE probe and a CRE mutant oligonucleotide (Santa Cruz Biotechnology; 100-fold) were used to check the specificity of the probes. Oligomer sequences used to generate the radiolabeled probe and to check the specificity of the probe are as follows: CRE wild type oligonucleotide, 5Ј-AGAGATTGCCTGACGT-CAGAGAGCTAG-3Ј (underline indicates CRE); CRE mutant type oligonucleotide, 5Ј-AGAGATTGCCTGTGGTCAGAG-AGCTAG-3Ј (boldface indicates mutated bases).
Confocal Microscopy and Immunofluorescence Assay-For confocal microscopy, cells were grown on a coverslip. For immunofluorescence assay, cells were cultured in a 24-well microplate using the differentiation medium with or without DHCA for 16 h, fixed using 4% paraformaldehyde in PBS for 20 min at room temperature, and permeabilized by incubating in methanol for 7 min at Ϫ20°C. After washing three times in PBS and blocking in 10% BSA-containing PBS, cells were incubated for 1 h at room temperature with anti-cyclinA antibody (rabbit polyclonal, cross-reacts with mouse cyclinA, 1:100) or anti-C/ EBP␤ antibody (1:100) diluted in blocking buffer. For detection, FITC-conjugated anti-rabbit IgG or Alexa588-conjugated anti-rabbit IgG (1:200) was diluted in blocking buffer containing 1 ng/ml DAPI (4Ј, 6-diamidino-2-phenylindole; Molecular Probes, Eugene, OR) or Hoechst (Polyscience, Warrington, PA). For confocal microscopy, cells were incubated with secondary antibody mixture for 1 h at room temperature, washed three times in PBS, and covered with Permount mounting medium (Fisher). Slides were analyzed by confocal microscopy (MRC-1024 Laser Scanning Confocal Image System, Bio-Rad). For immunofluorescence assay, cells were incubated with secondary antibody mixture for 1 h at room temperature, washed three times in PBS, and observed by inverted immunofluorescence microscopy (Axiovert 200 M, Zeiss, Gottingen, Germany).
Chromatin Immunoprecipitation (ChIP) Assay-ChIPs were performed using the ChIP-IT TM express kit (Active Motif, Carlsbad, CA). Briefly, 3T3-L1 and primary MEFs were fixed by adding 37% formaldehyde to media and lysed according to the manufacturer's instructions. Chromatins sheared by sonication were precipitated with anti-C/EBP␤ antibody. PCR amplifica-tions were performed with input DNA or precipitated DNA and specific primers encompassing the C/EBP-binding site in the proximal promoter of C/EBP␣, PPAR␥, and aP2 genes. The amplified products were resolved by electrophoresis in agarose gel. The primer sequences used to amplify the specific fragment of the promoter are as follows: C/EBP␣, forward 5Ј-TCC CTA GTG TTG GCT GGA AG-3Ј and reverse 5Ј-CAG TAG GAT GGT GCC TGC TG-3Ј; PPAR␥, forward 5Ј-TTC AGA TGT GTG ATT AGG AG-3Ј and reverse 5Ј-AGA CTT GGT ACA TTA CAA GG-3Ј; and aP2, forward 5Ј-CCT CCA CAA TGA GGC AAA TC-3Ј and reverse 5Ј-CTG AAG TCC AGA TAG CTC.
Statistical Analysis-All values are expressed as means Ϯ S.E., and differences between values were analyzed by unpaired Student's t test by using SigmaPlot software (version 9.0; SYS-TAT Software Inc., Point Richmond, CA). Dose-response relations were evaluated by way of regression analysis. p values less than 0.05, which were calculated as one-tailed p values, were considered to be statistically significant.

Preparation of Anti-adipogenic Fractions from PG105-We
previously showed that PG105, a water-soluble extract from the stems of C. moschata, contains strong anti-obesity activities in a high fat diet-induced obesity mouse model (28). One of its biological characteristics is the inhibition of the differentiation of 3T3-L1 cells to mature adipocytes. Using this property, we attempted to identify the biologically active fraction and compound(s) responsible for this activity. Initially, a total watersoluble extract (PG105) was prepared from the dried stems of C. moschata and sequentially extracted with n-hexane, chloroform, ethyl acetate, and n-butanol (Fig. 1). 3T3-L1 cells were treated with 100 g/ml of one of four obtained fractions, together with MDI, and then the extent of adipocyte differentiation was determined by the Oil Red O staining method. Among four fractions, chloroform fraction showing a maximum activity ( Fig. 1) was chosen for further purification.
The chloroform fraction was subjected to the column chromatography using silica gel, and 11 subfractions were eluted by a mixture of n-hexane/chloroform/methanol (16:15:1). Each fraction was concentrated by heat evaporation and freeze-drying, dissolved in ethanol, and added to 3T3-L1 cell culture media at the concentration of 25 g/ml and simultaneously with MDI. After 8 days of culture, the cells were stained with Oil Red O dye to observe the presence of lipid droplets. Chloroform fraction number 9 produced the highest activity among the 11 fractions in terms of the inhibition of adipocyte differentiation (Fig. 2).
Purification and Identification of DHCA-Various column fractions were concentrated under reduced pressure, and the homogeneity of each fraction was determined by TLC. Chloroform fraction number 9 showing a single spot on TLC plate (supplemental Fig. S1A) was subjected to silica gel column chromatography again followed by eluting with a mixture of chloroform/methanol (30:1). The third among the obtained three fractions was most effective in inhibiting adipocyte differentiation at a concentration of 2.5 g/ml (Fig. 2), indicating that anti-adipogenic activities might have been enriched in this frac-tion, because the treatment concentration was lower by 10-fold as compared with 25 g/ml used in Fig. 2. This fraction was further purified by semi-preparative HPLC using -Bondapak C18 column. This final fraction was named CMC-9.
The purity of CMC-9 was demonstrated by HPLC using Inertsil ODS-2 column to reveal a single peak (supplemental Fig. S1B), and the molecular weight of the compound was determined by electron impact mass spectrum as 358 (supplemental Fig. S1C). The molecular formula was also deduced as C 20 H 22 O 6 from the exact molecular weight of 358.14166 that was obtained from the high resolution electron impact mass spectrum (supplemental Fig. S1D). Based on the data by 1 H, 13 C, and 1 H-13 C COSY NMR spectroscopy and mass spectrometry (supplemental Fig. S1, E-L), its structure was identified as DHCA, a lignan (Fig. 3).
Effects of Synthetic DHCA on Differentiation of 3T3-L1 and Primary MEFs-DHCA was synthesized on a laboratory scale as described previously (35). 3T3-L1 cells were induced to differentiate with MDI, and various concentrations of synthetic DHCA or extracted CMC-9 were added simultaneously. The concentration of synthetic DHCA utilized in this study was chosen based on the fact that 2.5 g/ml of extracted CMC-9 from the above tests was comparable with 7 M of synthetic DHCA. Cells were treated with 0.25, 2.5, and 25 g/ml extracted CMC-9 or 0.7, 7, and 70 M of synthetic DHCA, fol-lowed by Oil Red O staining. Lipid accumulation was effectively inhibited, and the number of mature adipocytes was decreased by both CMC-9 and DHCA in a dose-dependent manner (Fig.  4A). Similar results were obtained, using primary MEFs (prepared from E14 mice) treated with MDI and DHCA (data not shown). These data strongly suggested that DHCA is a major component of extracted CMC-9 and responsible for the antiadipogenic activity of CMC-9.
To determine the relative amounts of lipids accumulated in DHCA-treated and untreated adipocyte populations, Oil Red O stain bound to neutral lipids in adipocytes was extracted with isopropyl alcohol, and its absorbance was read at 520 nm. Through this quantitation method, we estimated the EC 50 value of DHCA to be 36 and 31 M for 3T3-L1 and primary MEFs, respectively (Fig. 4B).
The cytotoxic effect of DHCA on various cell types was examined by LDH assay. The IC 50 value of DHCA varied depending on cell types. In 3T3-L1, it was ϳ770 M, and the survival rate was higher in MDI-stimulated 3T3-L1 cells as compared with unstimulated ones (Fig. 4C). Primary MEFs were more sensitive to DHCA with its IC 50 value being 500 and 530 M, depending on MDI stimulation (Fig. 4C). Other cell lines like HepG2 (human hepatocytes) and ECV304 (human umbilical vein endothelial cells) were more resistant to DHCA (Fig. 4C). These data suggested that the inhibitory effect of DHCA on adipocyte differentiation did not result from cytotoxic effects, because DHCA concentrations used in the Oil Red O staining experiment were far lower than IC 50 .
Effects of Synthetic DHCA on RNA Levels of Genes Involved in Adipocyte Differentiation-Because adipocyte differentiation accompanies the changes in expression of various adipogenic and lipogenic genes (18,19,23,25), effects of DHCA were examined on the mRNA levels of genes involved in adipocyte differentiation and lipid synthesis in 3T3-L1 cells by Northern blot analysis. Confluently cultured 3T3-L1 preadipocytes were stimulated with MDI and treated with 0.1% ethanol (vehicle control) or three different concentrations of DHCA (0.7, 7, and 70 M) for 8 days. As shown in Fig. 5A, the mRNA level of PPAR␥ and C/EBP␣ was decreased by DHCA treatment. The amount of C/EBP␣ RNA was decreased by DHCA treatment in a dose-dependent manner and was almost undetectable at 70 M. The expression of preadipocyte marker, Pref-1, was decreased by MDI stimulation (Fig. 5A, compare lanes 1 and 2) but slightly increased upon treatment with DHCA (Fig. 5A,  lanes 3-5). The mRNA level of sterol regulatory element-binding protein-1c (SREBP-1c) and stearoyl-CoA desaturase-1 (SCD-1), the major regulators of lipogenesis, was markedly increased by MDI (Fig. 5A, compare lanes 1 and 2). However, DHCA treatment virtually shut down their gene expression even at the lowest concentration, 0.7 M (Fig. 5A, lanes 3-5). These data suggested that lipogenic gene expression is more sensitive to DHCA, as compared with adipogenic marker genes expression. These results are consistent with the observation that the accumulation of cytoplasmic triglyceride was prevented by DHCA treatment as shown in Fig. 4.
Next, we examined the time course of DHCA effects on the expression of various genes involved in adipogenesis. 3T3-L1 cells were cultured in MDI-containing media in the presence or FIGURE 1. Solvent fractionation. Various fractions were separated by sequential solvent fractionation from PG105, a water-soluble extract from the stems of C. moschata. 3T3-L1 cells were treated with each fraction at a concentration of 100 g/ml for 8 days and microscopically analyzed as shown. The PPAR␥ agonist, troglitazone, and p38 inhibitor, SB203580, were used as control drugs. The magnification is ϫ40.
absence of 70 M DHCA. Total RNAs were prepared at various time points as indicated in Fig. 5B. All genes tested, except for Pref-1, were highly activated by MDI, and their RNA levels were significantly increased as the MDI treatment time increased (Fig. 5B, compare lanes 1, 3, 5, 7, and 9), but DHCA effectively suppressed the expression of these genes to the background or an undetectable level at all time points (for example, compare lanes 7 and 8 in Fig. 5B). Pref-1 was the exceptional case; the expression level of this preadipocyte marker gene was already high before MDI stimulation, and its level gradually decreased to an undetectable level on day 11 (Fig. 5B, lane 9). It is interesting to note that the RNA level of Pref-1 was a lot higher at day 11 in DHCA-treated cells as compared with untreated control cells (Fig. 5B, compare lanes 9 and 10), suggesting that this lignan molecule might influence the production or stability of Pref-1 RNA. Whatever the detailed molecular action mecha-nism is, these results suggested that DHCA exerts suppressive effects on genes involved in adipogenesis.
DHCA also decreased the RNA levels of adipogenic and lipogenic genes in primary MEFs. MEFs were cultured, stimulated with MDI, and treated with various concentrations of DHCA (7, 35, and 70 M) for 8 days. After differentiation, total RNAs were isolated and examined by Northern blot analysis. As shown in Fig. 5C, the RNA levels of PPAR␥, C/EBP␣, and SREBP-1c were increased by MDI stimulation, whereas DHCA significantly down-regulated these expression in a dose-dependent manner.

Effects of Synthetic DHCA on Mitotic Clonal Expansion and Expression of Cell Cycle
Markers-Adipogenesis is divided into two stages, early and late, each characterized by mitotic clonal expansion and terminal differentiation, respectively (18,23). To study the working mechanism underlying inhibitory effects of DHCA on adipocyte differentiation as shown above, we examined its influence on mitotic clonal expansion. Cells were stimulated with MDI, simultaneously treated with different concentrations of DHCA, and analyzed for the cell number and DNA replication on days 2 and 4. DHCA suppressed cell division and DNA synthesis in a dose-dependent manner as shown in Fig. 6, A and B. The cell number and the level of 3 H-labeled thymidine incorporation were ϳ2-fold lower in 70 M DHCA-treated cells than the untreated control, indicating the delayed progression of the FIGURE 2. Isolation of CMC-9 from chloroform fraction. Chloroform fraction obtained from Fig. 1 was separated into 11 fractions by silica gel column chromatography. The mixture of n-hexane, chloroform, and methanol (n-hexane/CHCl 3 /MeOH, 16:15:1) was used as an eluent, and to each fraction was added 3T3-L1 cells at a concentration of 25 g/ml. Chloroform fraction number 9 showing the highest anti-adipogenic activity was subjected to the silica gel column chromatography again, producing three different fractions. Note that the third fraction most effectively inhibited 3T3-L1 cells differentiation at a concentration of 2.5 g/ml. This active fraction was further purified by semi-preparative HPLC using -Bondapak C18 column, and the eluted fraction was named CMC-9. After 8 days of treatment with each fraction, 3T3-L1 cells were stained with Oil Red O and microscopically analyzed. As a vehicle control, 0.1% ethanol-treated cells were used. Troglitazone and SB203580 were used as positive and negative control drugs, respectively. The magnification is ϫ40. cell cycle. This anti-proliferative effect seems to be specific for MDI-stimulated cells, as DHCA alone, without MDI stimulation, minimally influenced the cell division and DNA synthesis in the cells (Fig. 6, A and B).
To test whether such anti-proliferative effect of DHCA is restricted to 3T3-L1 cells, 3 different cell lines (Raw264.7, HepG2, and ECV304 cells) were treated with diverse concentrations of DHCA, and after 2 and 4 days, the cell number and 3 H-labeled thymidine incorporation were measured in the same way. As shown in Fig. 6, C and D, the proliferation rate varied between different cell lines, but at a given cell line, there was no significant difference between DHCAtreated and untreated cells (Fig. 6, C and D). These data suggested that the inhibitory activity of DHCA on the cell cycle might be specific to the differentiated 3T3-L1 preadipocytes.
To confirm the inhibitory effect of DHCA on mitotic clonal expansion at the molecular level, the changes in cell cycle marker gene expression were examined. MDI-stimulated 3T3-L1 cells were cultured in the presence or absence of 70 M DHCA. Total cell lysates were prepared from the cells every 8 h after MDI stimulation, followed by Western blot analysis using antibodies against cyclinA and Cdk2, which are required for the progression from G 1 to S in most cell types (36). As shown in Fig. 7A, the level of cyclinA was increased by MDI stimulation until 24 h, at which point it then decreased (compare lanes 1, 2, 4, 6, and 8). The expression of Cdk2 was also activated as differentiation progressed, reaching a peak at 24 -32 h (Fig. 7A, lanes 6 and 8). However, DHCA treatment lowered the level of cyclinA and Cdk2 (for example, compare lanes 6 and 7 in Fig. 7A), indicating that the progression of the 3T3-L1 cell cycle might have been blocked by DHCA treatment.
Similar results were obtained using primary MEFs. MEFs were cultured and stimulated with MDI in the presence or absence of DHCA for 24 h. Total cell lysates were prepared, followed by Western blot analysis, using antibodies to cyclinA and Cdk2. The expression of cyclinA and Cdk2 was up-regulated by MDI, but DHCA significantly lowered their levels, indicating that MDI-mediated proliferation of primary MEFs could also be effectively inhibited by this lignan molecule (Fig. 7C).
These findings are supported by the cells immunostained with antibody to cyclinA at 16 h during the adipocyte differentiation. The signal intensity of cyclinA in DHCA-treated 3T3-L1 cells was The stain bound to intracellular lipids was extracted with isopropyl alcohol, and its absorbance was analyzed spectrophotometrically. Results are expressed as a percentage of the value measured in the control adipocytes. The EC 50 values of DHCA on 3T3-L1 and primary MEFs were 36 and 31 M, respectively. C, cytotoxicity of DHCA. Survival rate of four types of cells treated with various concentrations of DHCA was measured by LDH assay. In the case of NIH 3T3-L1 and primary MEFs, both MDI-unstimulated (Ϫ) and stimulated (ϩ) cells were tested. Various concentrations of DHCA and MDI were simultaneously added to the cells. Triton X-100 was added at a final concentration of 1% for the determination of maximum release of LDH activity. Forty eight hours later, an aliquot of medium was taken and added to the LDH detection reagent. The absorbance of DHCA-untreated cells culture medium gave a background value. Absorbance was measured at 490 nm using the ELISA microplate reader. Results are presented as concentration-response curves, and data were fit using a nonlinear regression. Equation used for LDH assay was as follows: survival rate (%) ϭ (1 Ϫ (experiment value Ϫ background value)/(Triton X-100 value Ϫ background value)) ϫ 100. The IC 50 values of DHCA in HepG2, ECV304, MDI-unstimulated (Ϫ), and stimulated (ϩ) 3T3-L1 and primary MEFs were 4840, 1380, 770, 1060, 500, and 530 M, respectively.
visibly lower than that in untreated cells (Fig. 7B). In DAPI staining analysis of the same cells, no cytotoxic signs were observed such as an apoptotic body in these DAPI-stained nuclei (Fig. 7B). These results suggested that DHCA might inhibit mitotic clonal expansion by regulating the expression of cell cycle marker genes, rather than simply damaging the cells. Eight days after treatment, total cellular RNA was extracted, and 20 g of RNA was used for Northern blot analysis. 28 S rRNA was used as a loading control. B, time course of DHCA effects. 70 M DHCA was added into differentiation medium at day 0, and total cellular RNA was extracted at appropriate time points followed by Northern blot analysis. Cell viability, as determined by direct cell counting, was unaffected during the whole experimental period. C, effect of DHCA on the RNA level of adipogenic or lipogenic genes in primary MEFs. MEFs were cultured in differentiation medium in the presence of various concentrations of DHCA as indicated (7,35, and 70 M). Eight days after treatment, total RNAs were extracted, and 20 g of RNA was used for Northern blot analysis. GAPDH was used as a loading control.

Effects of Synthetic DHCA on Signaling Molecules and Transcription Factors for Adipocyte Differentiation-
The inhibitory activity of DHCA on mitotic clonal expansion indicated that this lignan molecule might suppress specific signals involved in the re-entry of growth-arrested preadipocytes into the cell cycle. To identify the underlying molecular mechanism, the effect of DHCA was examined on various proteins responsible for the initiation of the cell cycle during the early stage of adipocyte differentiation. 3T3-L1 cells were stimulated with MDI for 30 min in the absence or presence of DHCA, and the total cell lysates were prepared, followed by Western blot analysis to determine the status of insulin-like growth factor-1 receptor (IGF-1R) and insulin receptor substrate (IRS)-1, which are involved in the initial signaling cascade induced by insulin (37)(38)(39). As shown in Fig. 8A, MDI and DHCA did not affect the protein expression of IGF-1R but MDI dramatically increased the phosphorylation of IGF-1R and IRS-1, as compared with that of unstimulated control cells (compare lanes 1 and 2 in Fig.  8A). However, treatment with DHCA did not have any significant effect on MDI-mediated phosphorylation of IGF-1R and IRS-1 (Fig. 8A, compare lanes 2 and 3).
We also investigated the time course of DHCA effect on the status of ERK1 and ERK2, both playing key roles in one of the important downstream cascades under the insulin signaling pathway (40,41). 3T3-L1 cells were stimulated with MDI, followed by treatment with 70 M DHCA, and then total lysates were extracted at various time points and subjected to Western blot analysis with antibodies against ERK1/2 and phosphorylated ERK1/2 (Fig. 8B). The total amount of ERK1/2 was not affected by MDI stimulation as well as by the treatment with DHCA (Fig. 8B). When the culture medium was supplemented with FBS as a negative control, the amount of phosphorylated ERK1/2 was slightly increased at 30 min, but thereafter, it gradually decreased to an undetectable level (Fig. 8B, compare lanes  1, 2, 5, 8, and 11). The phosphorylation was dramatically increased in MDI-treated cells, with a kinetics similar to that observed in negative control cells, and the presence of DHCA had no effect on it (for example, compare lanes 5-7 in Fig. 8B).
Insulin signaling has been also reported to increase the activity of the cAMP-response element-binding protein (CREB) through the PI3K pathway (8,42). To test whether DHCA affected the DNA binding activity of CREB, 3T3-L1 cells were stimulated with MDI and simultaneously treated with 7 or 70 M of DHCA for 1 h, and nuclear proteins were analyzed by EMSA with CREB-specific binding oligonucleotide. The result showed that the DNA binding activity of CREB was induced by the treatment with MDI (Fig. 8C, compare lanes 1 and 2), and DHCA slightly increased the amount of the specific DNA-protein complex (Fig. 8C, compare lanes 2-4).
Among the downstream targets of CREB, C/EBP␤ is one of the critical transcription factors involved in the initiation of mitotic clonal expansion and the induction of the expression of pleiotropic transcription factors, such as C/EBP␣ and PPAR␥ (16,18,(43)(44)(45). We investigated the effect of DHCA on the expression of C/EBP␤, using the same cell lysates employed to detect ERK1/2. The level of C/EBP␤ protein was increased as the adipocyte differentiation progressed, reaching a maximum level at 3 h (Fig. 8D, compare lanes 1, 3, 6, 9, and 12). However, there was little difference between DHCA-treated and untreated control cells (for example, compare lanes 9 and 10 in Fig. 8D).
It had been reported that the expression level of C/EBP␤ increases immediately upon the induction of differentiation, whereas its DNA binding activity is observed at later times, for example between 12 and 16 h after MDI stimulation, concurrent with the entry of preadipocytes into S phase at the onset of mitotic clonal expansion (16,41,45). To examine whether the DNA binding activity of C/EBP␤ was affected by DHCA, 3T3-L1 and primary MEFs were stimulated with MDI in the absence or presence of DHCA for 16 h and then treated with Hoechst stain and antibody against C/EBP␤. In this immunofluorescence assay, the C/EBP␤ protein was localized in the nucleus, and in particular, the "punctate" pattern of C/EBP␤ was observed when cells were treated with MDI only (Fig. 8E, compare large boxes). The punctate pattern indicates that C/EBP␤ has acquired DNA binding activity and become localized to the centromeres (43,45). In the presence of DHCA, however, the C/EBP␤ protein was dispersed in the nucleus (Fig. 8E).
To confirm this result, the ChIP experiment was performed using cell lysates taken for the immunofluorescence assay. The data from these experiments showed that MDI treatment induced the binding of C/EBP␤ to chromatins associated with the C/EBP regulatory element present in the promoters for C/EBP␣, PPAR␥, and aP2 (Fig. 8F, compare lanes 4 and 5), whereas these interactions were disrupted in DHCA-treated cells (Fig. 8F, compare lanes 5 and 6).
It was previously reported that sequential phosphorylation of C/EBP␤, initially by MAPK (on Thr-188) and later by glycogen synthase kinase 3␤ (GSK3␤) (on Ser-184 or Thr-179), is necessary for its DNA binding activity (41,46). To verify whether DHCA affects the status of phosphorylated C/EBP␤, cell lysates were prepared in the same way as ChIP assay, subjected to twodimensional gel electrophoresis, and then immunoblotted using anti-C/EBP␤ antibody. The results showed that 16 h after MDI treatment, C/EBP␤ exhibited lower isoelectric points as compared with that from unstimulated cells (Fig. 8G). However, DHCA treatment highly inhibited MDI-mediated conversion of C/EBP␤ to more acidic (pI) forms, resulting in a large amount of the protein with more basic pI values. This result indicated that DHCA blocked the phosphorylation of C/EBP␤, consistent with the data showing the DHCA-mediated decrease in DNA binding activity of C/EBP␤ as measured by above for immunofluorescence and ChIP assays.
Taken together, DHCA could suppress the mitotic clonal expansion by blocking the phosphorylation as well as the DNA binding activity of C/EBP␤, resulting in the inhibition of adipo-genic differentiation without affecting the expression of C/EBP␤ and its upstream signaling pathways.
Effects of Synthetic DHCA Added at Different Times of MDI Treatment-The above data indicated that DHCA blocks adipogenesis by suppressing the clonal expansion of preadipocytes at the early stage of adipogenesis. To test whether the effects of DHCA are restricted to the early stage of adipocyte differentiation, cells were treated with DHCA at various time points during adipogenesis. Seventy M DHCA was added to the cells, simultaneously with MDI or 1-4 days after MDI stimulation, and the RNA level of various genes was determined by Northern blot analysis. Consistent with the data shown in Fig. 5, simultaneous treatment with MDI and DHCA strongly inhibited the adipogenic gene expression (Fig. 9, lane 2). However, when cells were treated with DHCA 1 day after MDI stimulation, the expression of PPAR␥ and C/EBP␣ could not be completely suppressed (Fig. 9, lane 3). Such a reduced effect became clearer when the DHCA treatment time was more delayed (Fig.  9, compare from lane 2 to lane 6). On the contrary, the delayed treatment of DHCA was still effective in decreasing the RNA level of SREBP-1c and SCD-1, even 4 days after MDI stimulation (Fig. 9, lanes 3-6). The data indicated that DHCA could regulate the late stage of adipogenesis by directly inhibiting the expression of lipogenic genes, even after the adipogenic genes were sufficiently induced.

DISCUSSION
We previously found that PG105, a water-soluble extract from stem parts of C. moschata, not only inhibits the adipocyte  D). An aliquot of cell lysate (20 g) was subjected to SDS-PAGE and immunoblotted with antibodies against IGF-1R, phosphorylated IGF-1R, phosphorylated-IRS-1, ERK1/2, phosphorylated-ERK1/2, C/EBP␤, or ␤-actin. The ␤-actin protein was used as a loading control. C, analysis of DNA binding activity of CREB. Seven or 70 M DHCA was added to differentiation medium, and nuclear proteins were prepared after 30 min, followed by EMSA. E, immunofluorescence analysis of the cells immunostained with antibodies against C/EBP␤. 3T3-L1 and primary MEFs were incubated in differentiation medium with or without 70 M DHCA for 16 h. Nuclei were detected with Hoechst staining method. F, ChIP analysis on the interaction between C/EBP␤ and the C/EBP-binding site in the promoters of C/EBP␣, PPAR␥, or aP2. 3T3-L1 and MEFs under the same conditions with immunofluorescence analysis were used. G, two-dimensional gel electrophoresis and Western blotting of C/EBP␤. Cell lysates were prepared under the same conditions as those for the above ChIP analysis, migrated in the immobilized pH gradient strips, and then subjected to SDS-PAGE, followed by Western blot analysis using antibody against C/EBP␤. differentiation in vitro but also reduces the adiposity of high fat diet-fed obese mice (28). In an effort to identify the biologically active compound(s), the conventional solvent fractionation was performed using PG105, and a specific subfraction containing a potent inhibitory effect on adipocyte differentiation was isolated from the chloroform fraction. The structural analysis revealed that a major component of the active fraction is DHCA.
To test whether DHCA actually has an anti-adipogenic activity, DHCA was chemically synthesized, and its biological activities were tested using 3T3-L1 and primary MEFs. DHCA suppressed the adipocyte differentiation and the accumulation of intracellular triglyceride in a dose-dependent manner, coinciding with the decreased expression of major adipogenic transcription factors, PPAR␥ and C/EBP␣, as well as a lipogenic transcription factor, SREBP-1c. These data indicated that DHCA regulates adipocyte differentiation and related biochemical factors by controlling the key transcription factors involved in the pathways.
The mitotic clonal expansion is known to occur prior to the activation of various adipocyte-specific genes (16,18,21). Our data indicated that DHCA might suppress mitotic clonal expansion by inhibiting the proliferation and DNA replication of preadipocytes through the regulation of the production and stability of cell cycle markers such as cyclinA and Cdk2. It is noteworthy that this anti-proliferative effect of DHCA was restricted to MDI-stimulated 3T3-L1 cells or primary MEFs because the cell number and DNA synthesis of unstimulated 3T3-L1 cells as well as Raw264.7, ECV304, and HepG2 cells were little affected by DHCA treatment even at the highest concentration of 70 M. These results suggested that DHCA might interfere with the signals prior to mitotic clonal expansion induced by MDI, which is essential for the initiation of the adipogenic program.
It has previously been reported that C/EBP␤ is expressed early in the adipocyte differentiation, and its Thr 188 site is immediately phosphorylated by ERK1/2, which is required but is not sufficient for its DNA binding activity. Following a period of lengthy delay, GSK3␤ translocates from the cytoplasm into the nucleus and additionally phosphorylates Ser 184 or Thr 179 of C/EBP␤, resulting in the conformational change and the acquirement of DNA binding activity. At this time, dually phosphorylated C/EBP␤ becomes localized to C/EBP consensusbinding sites in centromeric satellite DNA, concurrent with synchronous re-entry of preadipocytes into the cell cycle (43,45). Our data showed that DHCA could inhibit the phosphorylation, DNA binding, and centromeric localization of C/EBP␤. Although it still remains to be clarified how DHCA affects the phosphorylation of C/EBP␤, it is clear that C/EBP␤ is a key target molecule of DHCA for its anti-adipogenic activity in 3T3-L1 and primary MEFs.
Preadipocytes that have passed the stage of mitotic clonal expansion are known to enter terminal differentiation and become characteristic mature adipocytes (16,25,27). To investigate the roles of DHCA in adipocyte terminal differentiation, 3T3-L1 cells were treated with DHCA after the mitotic clonal expansion was fully initiated. Under this experimental condition, adipocyte differentiation or the expression of PPAR␥ and C/EBP␣ was not affected by DHCA, although the expression of SREBP-1c and SCD-1 was effectively suppressed. These data indicated that the observed effects of DHCA on adipogenesis may be the combined results of its actions on the mitotic clonal expansion and lipogenic genes.
DHCA contains the dual functions controlling both adipogenesis and lipogenesis by regulating key transcription factors. Taken together, our data indicate that DHCA might be a useful and safe reagent that can control the adipose tissue mass, and thus have potential as an anti-obesity therapeutic.