Lipid rafts play an important role for maintenance of embryonic stem cell self-renewal.

Lipid rafts are cholesterol-rich microdomains of cell membranes that have a variety of roles in cellular processes including receptor-mediated signal transduction. Lipid rafts also occur in embryonic stem (ES) cells, but their role in ES cells is largely unknown. Therefore, we investigated the role of lipid rafts in the maintenance of ES cell self-renewal. In the present study, we observed that the presence of lipid rafts/caveolae. The results from sucrose gradient fractionation showed that the expression of glycoprotein 130 (gp130) and leukemia inhibitory factor receptor β (LIFRβ) was decreased by treatment with methyl-β-cyclodextrin (Mβ-CD) but, interestingly, was not affected by caveolin-1 small interfering RNA (siRNA). In addition, LIF increased phosphorylation of signal transducer and activator of transcription 3 (STAT3) and Akt, and the expression level of c-Myc, which were attenuated by the pretreatment with Mβ-CD. However, caveolin-1 siRNA did not influence LIF-induced phosphorylation of STAT3 and Akt, and expression of c-Myc. Treatment with Mβ-CD and caveolin-1 siRNA decreased expression levels of Oct4 protein and Oct4, Sox2, FoxD3, and Rex1 mRNAs in normal culture conditions. Additionally, Mβ-CD and caveolin-1 siRNA decreased the expression levels of cyclin D1 and cyclin E, and the proliferation index [(S + G2/M)/(G0/G1 + S + G2/M)] of ES cells.

media every day for 5 days. Mouse ES cells were collected after transfection with 50 nM caveolin-1 siRNA or control siRNA for 24 h and then cultured with normal culture media for 4 days. Total RNA was extracted from mouse ES cells using a RNA extraction kit (Qiagen, Valencia, CA). Reverse transcription was performed on 3 g of RNA using an AccuPower RT PreMix reverse-transcription system kit (Bioneer, Daejeon, Korea) with oligo (dT) 18 primers. The real-time quantifi cation of RNA targets was performed using a Rotor-Gene6000 real-time thermal cycling system (Corbett Research, New South Wales, Australia) using a QuantiTect SYBR Green RT-PCR kit (Qiagen). The primers used are described in Table 1 . The 20 l reaction mixture contained 200 ng cDNA, 0.5 M of each primer, enzymes, and fl uorescent dyes. The data was collected during the extension step and was analyzed using the manufacturer's software. To verify the specifi city and identity of the PCR products, the amplifi cation cycles were followed by a high-resolution melting cycle from 65°C to 99°C at a rate of 0.1°C/2 s. When the melting temperature (Tm) is reached, double stranded DNA is denatured and the SYBR is released, which causes a dramatic decrease in fl uorescence intensity. The rate of this change was determined by plotting the derivative of the fl uorescence relative to the temperature (dF/dT) versus temperature by data analysis software of the real-time PCR instrument. The temperature at which a peak occurs on the plot corresponds to the Tm of the DNA duplex. ␤ -actin of control group was the endogenous control used for calibration and normalization.

Detergent-free purifi cation of caveolin-rich membrane fraction
Caveolin-enriched membrane fractions were prepared as described previously ( 29 ). Cells were washed twice with ice-cold PBS, scraped into 2 ml of 500 mM sodium carbonate (pH 11.0), transferred to a plastic tube, and homogenized with a Sonicator 250 apparatus (Branson Ultrasonic, Danbury, CT, USA) using three 20 s bursts. The homogenate was adjusted to 45% sucrose by the addition of 2 ml 90% sucrose prepared in MES-buffered solution consisting of 25 mM MES (pH 6.5) and 0.15 M NaCl, and placed at the bottom of an ultracentrifuge tube. A 5-35% discontinuous sucrose gradient was formed above (4 ml each of 5% and 35% sucrose, both in MBS containing 250 mM sodium carbonate) and centrifuged at 40,000 rpm for 20 h in a SW 41 rotor (Beckman Coulter, Fullerton, CA). Twelve 1 ml fractions were collected and analyzed by 8-12% SDS-PAGE.

Immunofl uorescence staining
Cells were fi xed with 3.5% paraformaldehyde in PBS and permeabilized for 10 min with 0.1% (v/v) Triton X-100 and washed three times with PBS, for 10 min each wash. Cells were preincu-may exert their functional roles following induction by insulin ( 15 ), transforming growth factor-␤ (TGF-␤ ) ( 16 ), bone morphogenetic protein (BMP) ( 17 ), platelet-derived growth factor (PDGF) ( 18 ), and receptors in nonraft regions, resulting in cellular physiological responses ( 19 ). Therefore, we hypothesized that lipid rafts/caveolae are associated with ES cell self-renewal because of the presence of receptor initiating signals and localization of signal molecules in these specifi c plasma membrane microdomains.
It is well-known that the self-renewal capacity of ES cells is maintained by various extracellular cues, such as cytokines, growth factors, hormones, fetal animal serum, and serum extract (20)(21)(22)(23)(24)(25)(26). The propagation of pluripotent mouse ES cells especially depends upon a cytokine known as leukemia inhibitory factor (LIF) ( 27 ). Presently, we attempted to determine the role of lipid rafts in ES cell self-renewal through the examination of involvement of lipid rafts in LIF-induced signaling as one of the factors to maintain ES cell selfrenewal. We demonstrate that lipid rafts are present in mouse ES cells and suggest a functional model in which lipid rafts play an important role in maintenance of ES cell self-renewal as a mediator of extracellular cues.

Materials
Mouse ES cells (ES-E14TG2a) were obtained from the American Type Culture Collection (Manassas, VA). FBS was purchased from Biowhittaker (Walkersville, MD). Fluorescein (FITC)-conjugated goat-anti rabbit IgM anti ␤ -actin antibodies were acquired from Sigma-Aldrich (St. Louis, MO). Anti-cyclin D1, cyclin E, c-myc , caveolin-1, gp130, LIFR and phospho-signal transducer and activator of transcription 3 (STAT3), horseradish peroxidase (HRP)-conjugated goat anti-rabbit and rabbit anti-mouse antibodies, and LIF were acquired from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-phospho-Akt 308 and phospho-Akt 473 antibodies were supplied by New England Biolabs (Ipswich, MA). Methyl-␤cyclodextrin (M ␤ -CD) was purchased from Sigma-Aldrich (St. Louis, MO). To exclude the possibility of nonspecifi c cytotoxicity of M ␤ -CD, we prepared the cholesterol-loaded M ␤ -CD (CLM) as described by Purdy and Graham ( 28 ). The treatment of CLM has no effect on the stem cell markers and cell proliferation markers examined in present study (supplementary Fig. I). In addition, we confi rmed that M ␤ -CD itself did not affect plasma membrane functions that do not depend on lipid raft microdomains through 2-DG uptake (supplementary Fig. II). All reagents were purchased commercially and were of the highest available purity.

ES cell culture
Mouse ES cells were cultured for 5 days in DMEM (Gibco-BRL, Gaithersburg, MD) supplemented with 3.7 g/l of sodium bicarbonate, 1% penicillin and streptomycin, 1.7 mM L-glutamine, 0.1 mM ␤ -mercaptoethanol, 5 ng/ml mouse LIF, and 15% FBS without a feeder layer. Cells were grown on gelatinized 12-well plates or 60 mm-diameter culture dishes in a 37°C incubator in an atmosphere of 5% CO 2 . The medium was replaced by serum-free DMEM for 12 h before the experiments to investigate the effect of LIF.

RNA isolation and real-time RT-PCR
Mouse ES cells were harvested after treatment with 10 mM M ␤ -CD for 1 h, and then the culture media was changed with M ␤ -CD free ing of glycoprotein 130 (gp130) and the LIF receptor (also referred as to LIFR ␤ ) ( 32 ). To visualize gp130, caveolin-1, and LIFR ␤ expression in the plasma membrane of mouse ES cells, immunofl uorescence staining was carried out. The staining revealed expression of gp130, caveolin-1, and LIFR ␤ mainly in the plasma membrane fraction ( Fig. 1A ). In addition, it was observed that gp130 is colocalized with LIFR ␤ ( Fig. 1B ). To assess whether LIFR ␤ and gp130 colocalized with the lipid raft fraction, the latter fractions were prepared by detergent-free purifi cation using discontinuous sucrose density gradient centrifugation. The plasma membrane lipid raft fraction was found to reside mainly in the light buoyant membranes ( Fig. 1C , fractions 4, 5, 6). Western blot analysis for LIFR ␤ , gp130, and caveolin-1 demonstrated localization of gp130 and LIFR ␤ in the lipid raft fraction ( Fig. 1C ). Control siRNA did not effect caveolin-1, gp130, or LIFR ␤ in lipid raft fraction in control condition ( Fig. 1D ). Importantly, gp130 and LIFR ␤ expression in the lipid raft fraction was signifi cantly decreased by methyl-␤ -cyclodextrin (M ␤ -CD) but was not affected by caveolin-1 siRNA or control siRNA. The results of immunofl uorescent staining showed that distribution of gp130 and LIFR ␤ in the plasma membrane was altered by M ␤ -CD but not by caveolin-1siRNA ( Fig. 1E ).

Relationship between lipid rafts and LIF-induced signaling
It is well-known that LIF induces the activation of STAT3 and PI3K/Akt signaling, which converge in c-Myc expression through the gp130 protein in mouse ES cells ( 4,33,34 ). We investigated the effect of M ␤ -CD and caveolin-1 siRNA in LIFinduced signals. LIF signifi cantly increased the phosphorylation of STAT3 ( Fig. 2A ) and Akt ( Fig. 2B ). c-Myc protein expression was also increased by LIF treatment. However, the increase of c-Myc expression level was attenuated by a 1 h pretreatment with 10 mM M ␤ -CD ( Fig. 2C ).
Next, we investigated the involvement of caveolin-1 in LIF-induced signaling. Caveolin-1 siRNA or control siRNA did not affect phosphorylation of STAT3 and Akt, or protein expression of c-Myc in normal culture conditions ( Fig.  2D, E ). We investigated the effect of M ␤ -CD and caveolin-1 siRNA in LIF-induced downstream signals. LIF-induced phosphorylation of STAT3 ( Fig. 3A ) and Akt ( Fig. 3B ), as well as c-Myc expression ( Fig. 3C ), were signifi cantly attenuated by a 1 h pretreatment with 10 mM M ␤ -CD. However, intriguingly, caveolin-1 siRNA did not abrogate bated with 10% BSA (Sigma-Aldrich) in PBS for 20 min to decrease nonspecifi c antibody binding. Cells were then incubated for 60 min with primary antibody in a solution containing 1% (v/v) BSA in PBS, and washed with PBS as above. Cells were incubated with 1% (v/v) BSA for 5 min and then incubated for 60 min with FITCconjugated secondary antibody in PBS containing 1% (v/v) BSA, and washed with PBS as above. Samples were mounted on slides and visualized with a FluoView 300 confocal microscope (Olympus, Tokyo, Japan) equipped with a 400× objective lens.

Western blot analysis
Cell homogenates containing 20 g protein were separated by 10%-12% SDS-PAGE and transferred to polyvinylidene fl uoride transfer membranes (Pall, Gelman Laboratory, Ann Arbor, MI). Each membrane was washed with TBST [10 mM Tris-HCl (pH 7.6), 150 mM NaCl, 0.05% Tween-20] and blocked with 5% skimmed milk for 1 h prior to incubation with a 1:1,000 dilution of the appropriate primary antibody. Each membrane was washed, and primary antibodies were detected with a 1:10,000 dilution of HRP-conjugated goat anti-rabbit IgG or goat anti-mouse IgG. The reactive bands were visualized with enhanced chemiluminescence (Amersham Pharmacia Biotech, Franklin Lakes, NJ) on KODAK chemiluminescence BioMax fi lm (Carestream Health Inc., Rochester, NY).

FACS analysis
Cells were dissociated in trypsin/EDTA, pelleted by centrifugation, and resuspended at approximately 10 6 cells/ml in PBS containing 0.1% BSA. The cells were then fi xed with 70% ice-cold ethanol for 30 min at 4°C, followed by incubation in a freshly prepared nuclei staining buffer consisting of 250 g/ml propidium iodide (PI) and 100 g/ml RNase for 30 min at 37°C. Cell cycle histograms were generated after analyzing the PI-stained cells by fl uorescenceactivated cell sorting (FACS) (Beckman Coulter). The samples were analyzed using CXP software (Beckman Coulter) and the proliferation indices [(S + G2/M)/(G0/G1 + S + G2/M)] were calculated.

Statistical analysis
The results are reported as the mean ± SE, and all experiments were analyzed by ANOVA. Analysis was followed by a comparison of the treatment means with the control using the Bonferroni-Dunn test. Statistical signifi cance was defi ned as P < 0.05.

Presence of lipid raft/caveolae and localization of LIF receptor in lipid rafts
Mouse ES cells can self-renew with the addition of recombinant LIF protein to the culture medium ( 30,31 ). The receptor for LIF is a heterodimeric complex consist-

Oct4
Sense These results suggest that lipid rafts/caveolae play an important role in the maintenance of ES cell pluripotency. Next, we examined the relationship between the proliferative capacity of mouse ES cells and lipid rafts/caveolae. As shown in Fig. 5A , M ␤ -CD signifi cantly decreased the cyclin E and cyclin D1 expression levels in normal culture conditions. In addition, the expression levels of these proteins were decreased by caveolin-1 siRNA in a dosedependent manner ( Fig. 5B, C ). Furthermore, M ␤ -CD and caveolin-1 siRNA signifi cantly decreased the proliferation index compared with control (control: 68.56 ± 7.63%; M ␤ CD: 51.86 ± 4.1%; caveolin-1 siRNA: 55.50 ± 3.15%) ( Fig.  5D ). Taken together, these results suggest that lipid rafts/ caveolae have a functional role in ES cell self-renewal.

DISCUSSION
In the present study, we demonstrated the presence of lipid rafts/caveolae and its role in self-renewal of mouse ES cells. Of the various functions of lipid rafts, evidence has accumulated for their role as a cell membrane signaling platform where multiple signals are initiated ( 35,36 ). For example, various cytokine and growth factor receptors LIF-induced phosphorylation of STAT3 ( Fig. 3D ), Akt ( Fig. 3E ), or c-Myc expression ( Fig. 3F ). These results suggest that LIF-induced downstream signals are not mediated by caveolae or caveolin-1.

Role of lipid rafts/caveolae in ES cell self-renewal
To examine the role of lipid rafts/caveolae in maintenance of pluripotency and proliferation of mouse ES cells, mouse ES cells were incubated with or without 10 mM M ␤ -CD for 1 h everyday for 5 days, followed by replacement with media lacking M ␤ -CD, and then Oct4 protein expression was examined. As shown in Fig. 4A , M ␤ -CD signifi cantly decreased Oct4 protein expression level. Next, the effect of caveolin-1 knockdown on Oct4 expression was assessed. Oct4 protein expression was decreased by caveolin-1 siRNA in a dose-dependent manner ( Fig. 4B ), but Oct4 protein expression was not infl uenced by control siRNA ( Fig. 4C ). In addition, the level of Oct4 protein expression was decreased by withdrawal of LIF in a timedependent manner ( Fig. 4D ). To confi rm these results, the expression of pluripotency marker mRNAs was examined using real-time RT-PCR. As shown in Fig. 4E-H , mRNA expression levels of Oct4, Sox2, FoxD3, and Rex1 were signifi cantly decreased by M ␤ -CD and caveolin-1 siRNA. In general, mouse ES cells can be maintained undifferentiated by LIF and FBS ( 30,31 ). This is due to the regulated expression of transcription factors and the signals for differentiation, at least in part, by LIF supplementation or by several extrinsic factors found in FBS that include BMPs and Wnt. Appropriately, we investigated the involvement of lipid rafts/caveolae in LIF-induced signals as one of the cytokines essential for maintenance of self-renewal of mouse ES cells. At fi rst, we observed that lipid rafts/caveolae reside in the plasma membrane of mouse ES cells along with the coexpression of the gp130 LIF receptor subunit. In addition, gp130 was detected in caveolin-enriched membrane fractions. These observations suggested that gp130 could interact with caveolin-1, with the LIF-induced signal being mediated by caveolin-1. However, gp130 expression and distribution in the lipid raft fraction were changed by 10 mM M ␤ -CD, which are constitutively localized in lipid rafts or are translocated into lipid rafts upon ligand binding, and lipid raft localization is essential for downstream signaling ( 37,38 ). Indeed, localization of BMP receptors in lipid raft microdomains and this membrane region infl uences BMP signaling ( 17 ), and TGF-␤ receptors can form a complex in lipid rafts as well as in nonraft regions ( 16 ). Moreover, the PDGF receptor may be sequestered in a raft compartment at the developmental stage, with the rafts functioning in defi ning the downstream signaling response to PDGF in oligodendrocytes ( 39 ). Therefore, we hypothesized that signaling by a variety of extacellular cues for self-renewal of ES cells might be mediated by lipid rafts/caveolae. The present results are, to our knowledge, the fi rst report of the presence and potential function of lipid rafts/caveolae in mouse ES cells. is known to disrupt lipid raft structure and inhibit lipid raft clustering in response to cytokine stimulation by binding to and sequestering cholesterol from the plasma membrane ( 11,40,41 ); however, knockdown of caveolin-1 had no effect. In addition, LIF-induced phosphorylation of STAT3, Akt, and c-Myc expression was attenuated by M ␤ -CD but not by caveolin-1 siRNA. These results indicate that the LIFR is located in noncaveolae lipid rafts or at least that LIFR-initiated signals are not mediated by caveolin-1.  (D) Cells were cultured with or without LIF for 5 and 10 days, and Oct4 protein expression levels were detected by Western blot. Each example is representative of three experiments. The graphs denote the mean ± SE of three experiments for each condition determined from densitometry relative to ␤ -actin. (E-H) Mouse ES cells were harvested after culture in normal culture condition for 5 days, cultured with 10 mM M ␤ -CD for 1 h every day for 5 days in normal culture condition, and cultured in normal culture condition for 4 days after transfection with 50 nM caveolin-1 or nontarget control siRNA for 24 h, respectively. Real-time RT-PCR quantifi cation of Oct4, Sox2, FoxD3, and Rex1 was carried out. The data is reported as the mean ± SE of three independent experiments, each conducted in triplicate. * P < 0.05 versus control. ES cell, embryonic stem cell; LIF, leukemia inhibitory factor; M ␤ -CD, methyl-␤ -cyclodextrin; ROD, relative optical density; siRNA, small interfering RNA; STAT3, signal transducer and activator of transcription 3. renewal of mouse ES cells, and when inhibited, mouse ES cells did not keep the self-renewal markers expression level of undifferentiated state. We also show that disruption of lipid rafts and caveolin-1 siRNA resulted in declined expression of the cell cycle regulatory proteins cyclin D1 and cyclin E and decreased ES cell proliferation. The most obvious feature of the pluripotent cell cycle, apart from it being unusually rapid, is its short G1 phase. A contributing factor to this is likely to be the activity of cyclin D1/E and perhaps associated kinases. This is supported by a large body of evidence demonstrating cyclin E to be a rate-limiting factor for the G1-to-S transition (42)(43)(44)(45). One important point of these data is that the self-renewal marker was decreased not only by M ␤ -CD but also by the knockdown of caveolin-1. In general, a variety of factors are needed for the expression of transcription factors and signals for ES cell self-renewal, except for LIF. In other words, these results indicate the importance of lipid rafts/caveolae for the function of other factors necessary for ES cell self-renewal.

CONCLUSION
The present study demonstrates the importance of lipid rafts/caveolae in maintaining self-renewal of mouse ES cells. We propose that lipid rafts are required for the efficient signal transduction to effectively maintain selfrenewal of mouse ES cells. Further investigations should reveal the nature of the participation of lipid rafts/caveolae with other factors implicated in ES cell self-renewal or differentiation. It will also be interesting to evaluate if lipid rafts are important for self-renewal of human ES cells.
Next, we determined the effect of M ␤ -CD and caveolin-1 siRNA to prove the role of lipid rafts/caveolae in the maintenance of self-renewal of mouse ES cells. Destruction of lipid rafts by M ␤ -CD and caveolin-1 siRNA downregulated the expression of Oct4, a pluripotency marker protein of ES cells, as well as the expression of caveolin-1. A similar result was obtained upon withdrawal of LIF. Correspondingly, M ␤ -CD and caveolin-1 siRNA signifi cantly decreased mRNA expression of Oct4, Sox2, Rex1, and FoxD3, which are related to ES cell pluripotency. We assume that lipid rafts/ caveolae are important for maintenance of effi cient self-