Heparan Sulfate Regulates Self-renewal and Pluripotency of Embryonic Stem Cells*

Embryonic stem (ES) cell self-renewal and pluripotency are maintained by several signaling cascades and by expression of intrinsic factors, such as Oct3/4 and Nanog. The signaling cascades are activated by extrinsic factors, such as leukemia inhibitory factor, bone morphogenic protein, and Wnt. However, the mechanism that regulates extrinsic signaling in ES cells is unknown. Heparan sulfate (HS) chains are ubiquitously present as the cell surface proteoglycans and are known to play crucial roles in regulating several signaling pathways. Here we investigated whether HS chains on ES cells are involved in regulating signaling pathways that are important for the maintenance of ES cells. RNA interference-mediated knockdown of HS chain elongation inhibited mouse ES cell self-renewal and induced spontaneous differentiation of the cells into extraembryonic endoderm. Furthermore, autocrine/paracrine Wnt/β-catenin signaling through HS chains was found to be required for the regulation of Nanog expression. We propose that HS chains are important for the extrinsic signaling required for mouse ES cell self-renewal and pluripotency.

tent cell lines (1,2). ES cells retain the ability to differentiate into the representative cell types of all three germ layers of the developing mouse embryo. Human ES cell lines have been derived (3), and the pluripotency of these cells is a feature providing considerable potential for exploitation in the development of cell replacement therapies to treat human disease. However, the molecular mechanisms that control pluripotency and differentiation of ES cells are largely unknown. It will be essential to gain a better understanding of these mechanisms to exploit the potential of ES cells for therapeutic purposes.
A number of studies have investigated the factors controlling pluripotency of mouse ES (mES) cells (4). Self-renewal of mES cells is sustained by signals mediated by the cytokine, leukemia inhibitory factor (LIF) (5,6). LIF signals through the heteromeric receptor gp130 and the low affinity LIF receptor to induce activation of STAT3 (7)(8)(9)(10). However, exposure of cells to serum is also required for LIF-mediated maintenance of self-renewal (11).
Treatment of mES cells with the bone morphogenic proteins (BMPs) BMP2 and BMP4 or with growth differentiation factor 6 can substitute for serum. The ability of BMP/growth differentiation factor 6 to promote self-renewal requires co-stimulation with LIF (11). BMP induces the expression of inhibitor of differentiation (Id) genes through activation of Smad signaling, and the Id gene products suppress expression of genes involved in the induction of neural differentiation (11). Thus, BMP suppresses neural differentiation and, in combination with LIF, is sufficient to sustain self-renewal of mES cells without the need for feeder cells or serum factors.
Wnt signaling was shown to play a role in the regulation of self-renewal of both mouse and human ES cells independently of LIF/STAT3 signaling (12). Wnt proteins play roles in the regulation of gene expression, cell proliferation, and differentiation and in the maintenance of cell polarity (13). The binding of Wnt protein to its cognate receptor, Frizzled, results in the inhibition of glycogen synthase kinase-3. This in turn leads to the stabilization and nuclear accumulation of ␤-catenin and to changes in gene transcription. Signaling by this canonical Wnt pathway has been suggested to result in downstream activation of expression of the homeoprotein Nanog, a transcription factor that is essential for maintenance of the inner cell mass and of ES cell pluripotency (14,15). The activation of Nanog sustains ES cell self-renewal without the use of feeder cells or treatment with LIF (12).
Heparan sulfate (HS) proteoglycans are ubiquitously present in the extracellular matrix and on the cell surface. The HS polysaccharide chains of the proteoglycans are covalently attached to several core proteins (16). HS chains are synthesized in the Golgi apparatus by several enzymes, including members of the EXT protein family. The chains consist of repeating disaccharide units of D-glucuronic acid-N-acetyl-Dglucosamine that are modified differentially by epimerization and sulfation (16). A large number of physiologically important molecules can bind to specific sulfated regions of HS chains (17). Genetic studies have shown that HS chains regulate biological functions by interacting with various extracellular signaling molecules, such as members of the fibroblast growth factor (FGF) family, Wnt/Wingless (Wg), Hedgehog (Hh), and BMP (18). In Drosophila, for example, analyses of mutations of the EXT family genes tout-velu (ttv), sister of ttv (sotv), and brother of ttv (botv) have indicated that HS chains are required for signaling and distribution of Hh, Wg, and Decapentaplegic (the functional ortholog of mammalian BMP2 and BMP4) during embryogenesis and wing development (19 -22). In mammals, the importance of HS chains in development has been demonstrated by analyses of mutations of enzymes required for HS chain modification, and FGF and Indian hedgehog signaling through HS chains has been suggested to be required during development (18,(23)(24)(25)(26). Thus, there is evidence that HS chains have essential functions in development, however, it is not yet clear what role HS chains play in the regulation of early embryogenesis and in ES cells.
Our current understanding suggests that HS chains may contribute to the maintenance of ES cell self-renewal by regulating the activities of several signaling pathways, such as LIF/STAT3, BMP/Smad, and Wnt/␤-catenin. In the present study, we investigated the contribution of HS chains to the regulation of ES cell self-renewal and pluripotency. We used small interfering RNA (siRNA) to knockdown EXT1, which is required for HS chain elongation. Transfected mES cells grew more slowly than untreated control cells and differentiated into extraembryonic endoderm even in the presence of LIF and serum. This is the first demonstration of the importance of HS chains for the maintenance of self-renewal and pluripotency of mES cells.
For differentiation into embryoid bodies (EBs), the cells were transferred on day 3 to low-cell-binding 60 mm dishes (Nunc) and cultured in ES medium without LIF. The numbers of small noncystic EBs and large EBs filled with cystic cavity were counted by microscopic examination. EBs were fixed overnight in 4% paraformaldehyde at 4°C, dehydrated, embedded in paraffin and sectioned at 10 m. Sections were stained with hematoxylin and eosin (Merck).
For morphological observation and real time PCR analysis of differentiation markers, the cells were replated on day 2 and incubated with puromycin for 3 days.
FACS Analysis-Two days after transfection, mES cells were harvested and the cell suspension was incubated with mouse IgM negative isotype control (Chemicon), anti-HS antibody 10E4 (Seikagaku Corp.) or anti-chondroitin sulfate (CS) antibody 2H6 (Seikagaku Corp.) diluted in FACS buffer (0.5% bovine serum albumin (BSA) and 0.1% sodium azide in PBS). After washing, the cell suspension was incubated with fluorescein isothiocyanate-conjugated secondary antibody (Sigma) diluted in FACS buffer. Cell sorting and analysis were performed using a FACSAria Cell Sorter (BD Biosciences).
High Performance Liquid Chromatography (HPLC) Analysis of Unsaturated Disaccharides-Two days after transfection, mES cells were harvested and fluorometric post-column HPLC analysis of unsaturated disaccharides from HS chains was performed as reported previously (30).
Molecular Size Analysis of HS Chains-One day after transfection, mES cells were harvested, replated at 1.5 ϫ 10 6 cells per well in 6-well 0.2% gelatin-coated plates and incubated in sulfate-free ES medium with LIF, puromycin and 100 Ci/ml Na 2 35 SO 4 (ARC). After labeling for 24 h, the cells were washed twice with PBS and then treated with 1 mg/ml trypsin (WAKO) for 10 min at 37°C. The trypsin was neutralized with 2 mg/ml trypsin inhibitor (Roche Applied Science). After centrifugation, the supernatants were treated with 0.5 M NaOH at 4°C overnight and neutralized with 1 M acetic acid. The labeled galactosaminoglycans were eliminated by chondroitinase ABC (Seikagaku Corp.) digestion, desalted in a PD-10 column (GE Healthcare) and resistant HS chains were isolated by anion exchange chromatography on HiTrap DEAE FF (GE Healthcare) using sodium phosphate buffer (pH 6.0) containing 1.0 M NaCl as the eluent. The sizes of the purified HS chains were analyzed by gel chromatography on a Sephacryl S-300 column (GE Healthcare) (1 ϫ 44 cm) and eluted with 50 mM Tris-HCl, pH 7.4, containing 0.2 M NaCl. Fractions (1 ml/fraction) were collected and analyzed for radioactivity using a scintillation counter. The purity of the labeled HS was determined by sensitivity to enzyme digestion with 5 milliunits of heparitinase I and II (Seikagaku Corp.) and heparinase (Seikagaku Corp.). Estimations of molecular mass values were derived from fractionation of several Dextran molecular size markers (Sigma) by gel chromatography and staining with orsinol solution.
Proliferation Assay-Two days after transfection, mES cells were harvested and replated in triplicate at 0.8 ϫ 10 4 cells per well in 96-well 0.2% gelatin-coated plates in ES medium with LIF. Cell counting kit-8 (Dojindo) was added after 0 h, 24 h or 48 h and incubated further for 2 h. The soluble formazan product was measured at 450 nm.
Self-renewal Assay-Two days after transfection, mES cells were harvested and replated at 1 ϫ 10 4 cells per gelatin coated 60 mm tissue culture dish in ES medium with LIF. For detection of undifferentiated cells, cells were fixed and stained with 5-bromo-4-chloro-3-indolyl phosphate-nitro blue tetrazolium (Nacalai Tesque) 5 days after replating. Alkaline phosphatase (AP) positive colonies were counted by microscopic examination. Colonies of tightly packed and flattened AP-positive cells were counted as undifferentiated, and colonies of mixtures of unstained and stained cells and entirely unstained cells with flattened irregular morphology were considered differentiated.
Immunoblotting-Two days after transfection, the culture solution for the mES cells was replaced with serum-free ES medium without LIF for 4 h and the cells were stimulated for 20 min with one of the following: 15% FBS, 1000 units/ml LIF, 10 ng/ml BMP4 (R&D Systems), 40 ng/ml basic FGF (Upstate Biotechnology) or 15% FBS plus 1000 units/ml LIF. For depletion of HS chains, mES cells were incubated with 5 milliunits of heparitinase I and II and heparinase for 2 h before extrinsic stimulation. Cells were lysed with lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM Na 3 VO 4 , 10 mM NaF, protease inhibitors).
To isolate nuclear extracts, cells were first suspended in 100 l of buffer (10 mM Hepes, pH 7.4, 2 mM MgCl 2 ,1 mM EDTA, 10 mM KCl, 1 mM dithiothreitol, protease inhibitors). After incubation for 15 min on ice, 12.5 l of 5% Nonidet P-40 was added, the suspension was vortexed for 10 s, and incubated for a further 5 min on ice. The suspension was centrifuged at 13,000 rpm for 30 s. The supernatant was removed and the pellets, comprising the nuclear extracts, were washed with PBS and lysed with lysis buffer (25 mM Hepes, pH 7.4, 500 mM NaCl, 1 mM dithiothreitol, 10% glycerol, 0.2% Nonidet P-40, 5 mM MgCl 2 ). The purity of the cell fractionation (cytosol and nucleus) was confirmed in Western blot analysis using an anti-Yes monoclonal antibody (BD Biosciences) and an anti-Lamin B 1 antibody (Zymed Laboratories Inc.); these antibodies are specific for the cytosol and nucleus, respectively. Only low levels of cross-contamination were observed (Ͻ1%).
Luciferase Reporter Assay and Immunostaining-Transactivation of ␤-catenin on T-cell-specific factor (Tcf) was determined with a luciferase reporter assay. siRNA expression plasmid (2 g) was cotransfected with reporter plasmid such as, TOPFLASH (2 g, containing three Tcf binding sites, Upstate Biotechnology) or FOPFLASH (2 g, containing inactive Tcf binding sites, Upstate Biotechnology) and pCH110 (0.2 g, containing ␤-galactosidase, GE Healthcare) as control of transfection efficiency using Lipofectamine 2000 as described above. Cell lysates were prepared 3 days after transfection and luciferase activity was measured with Dual-Light System (Applied Biosystems). Luminescence was measured with a Lumat LB9501 luminometer (Berthold). Luciferase activity was normalized for transfection efficiency by ␤-galactosidase activity.
Relative luciferase activity is defined as the ratio of luciferase activity of TOPFLASH to that of FOPFLASH.
For immunostaining of ␤-catenin, mES cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% saponin 3 days after transfection. After washing and subsequent blocking, cells were stained with an anti-␤-catenin antibody. After washing, cells were stained with fluorescein isothiocyanate-conjugated secondary antibody (Chemicon) and counterstained with propidium iodide (BD Biosciences). Immunofluorescence images were taken using an LSM5Pascal confocal laser scanning microscope (Carl Zeiss) with ϫ40/1.3 objectives at room temperature.

125
I-Labeled Wnt3a Binding Assay-Two micrograms of recombinant mouse Wnt3a (R&D Systems) were iodinated with 100 Ci of 125 I-sodium (ARC) using iodogen-precoated reaction tubes (Pierce) according to manufacturer's instructions. Radiolabeled Wnt3a was separated from unincorporated 125 I-sodium on a PD-10 column. The specific activity of the radiolabeled Wnt3a was 1.85 ϫ 10 4 cpm/ng protein. For the binding assay, mES cells were harvested 2 days after transfection and replated in triplicate at 5 ϫ 10 5 cells per well in 24-well 0.2% gelatin-coated plates. The cells were allowed to attach for 3 h and then washed three times with ice-cold binding buffer (serum-free ES medium containing 1 mg/ml BSA and 0.1% sodium azide). After incubating with 80 ng/ml 125 I-labeled Wnt3a either alone or in the presence of 100 g/ml heparin (Sigma) in binding buffer for 3 h at 4°C, the cells were gently washed three times with ice-cold binding buffer and lysed with 0.2 N NaOH. The amount of radiolabeling in each extract was counted with a gamma counter (Aloka).
Surface Plasmon Resonance (SPR) Analysis-Heparin (Nacalai Tesque) was dialyzed against distilled water using an MWCO3500 membrane (SpectroPore) and lyophilized. Then the heparin was conjugated with a monovalent linker molecule to prepare the ligand conjugate for the immobilization of heparin on the gold-coated chip as previously described (31). The surface of the gold-coated chip (SUDx-Biotec) was oxidatively washed with UV ozone cleaner (Structure Probe Inc.) for 20 min. The chips were then immersed in 1 M of the ligandconjugate in 50% (v/v) methanol solution overnight at room temperature with gentle agitation to prepare Sugar Chips with immobilized heparin. The Sugar Chips were washed sequentially with water, PBS containing 0.05% Tween 20 and water, and dried at room temperature.
The Sugar Chip with immobilized heparin was set on a prism with refraction oil (n D ϭ 1.518, Cargill Laboratories Inc.) in an SPR apparatus (SPR670M, Moritex). The SPR measurements were performed at room temperature in accordance with the manufacturer's instructions and using Tris-buffered saline (20 mM Tris-HCl, 150 mM NaCl, pH 7.4) containing 0.05% Tween 20 and 0.1% BSA as the running buffer at a flow rate of 15 l/min. The kinetic binding parameters were calculated using the software of the manufacturer. We performed binding of BSA or von Willebrand factor (vWF) A1 to heparin, as negative and positive controls, respectively (supplemental data 1) (apparent association and dissociation rate constants k a (M Ϫ1 s Ϫ1 ) and k d (s Ϫ1 ) of vWF A1, 3.51 ϫ 10 3 , 4.38 ϫ 10 Ϫ3 , respec-tively; apparent equilibrium constant K D (nM) for binding of vWF A1 to heparin, 1248.1).
RT-PCR and Real Time PCR-Total RNA was isolated from mES cells by TRIzol Reagent (Invitrogen) and subsequently reverse transcribed using an oligo-dT primer (Invitrogen) and a SuperscriptII first strand synthesis kit (Invitrogen). Primer sets for PCR amplification are listed in supplemental data 2. All cDNAs were amplified in quantitative ranges, which were confirmed by examining various cycles for the samples giving maximum levels of signals in each primer sets. Primers sets and probes for real time PCR are listed in supplemental data 3. Real time PCR was performed using an ABI PRISM 7700 sequence detection system (Applied Biosystems). The relative amounts of each mRNA were normalized by ␤-actin mRNA in the same cDNA.

HS Chains Are Reduced by Knockdown of EXT1 mRNA-He-
reditary multiple exostosis is an autosomal dominant disorder characterized by the development of benign cartilage-capped tumors at the juxta-epiphyseal regions of long bones (32). It is associated with mutation of either EXT1 or EXT2, which encode glycosyltransferases possessing both D-glucuronic acid and GlcNAc-transferase activities that are necessary for HS chain elongation (33,34). In the Golgi apparatus, EXT1 and EXT2 form a hetero-oligomeric complex that exhibits higher glycosyltransferase activity than either enzyme alone (35,36). To examine the effect of reduced HS chain levels in ES cells, we knocked down expression of EXT1 mRNA using RNAi. We designed two constructs expressing different siRNAs targeting EXT1 (EXT1-1 and EXT1-2), as described previously (29), and siRNAs targeting EGFP as a negative control. We represent mES cells transfected with EGFP siRNA expression plasmids as "control cells" throughout this paper. Real time PCR analysis performed 2 days after transfection of the cells with EXT1 siRNA expression plasmids showed that the level of EXT1 mRNA was reduced to about 20% that of control cells. In contrast, the expression of other nontargeted genes was unaffected by this treatment in the two mES lines tested (R1 and E14TG2a) (Fig. 1A). The knockdown effect on EXT1 was maintained for at least 3 days following puromycin selection. We used E14TG2a to confirm the same knockdown effect on mES cells using one siRNA expression plasmid (EXT1-1) throughout all experiments.
Two days after transfection, we performed FACS analysis using an anti-HS antibody to determine whether HS chain expression was reduced by knockdown of EXT1 mRNA. As shown in Fig. 1B, HS chains were abundant in mES cells, whereas CS chains were not. We observed a significant reduction in HS chain expression on the cell surface of cells transfected with an EXT1 siRNA expression plasmid compared with control cells (Fig. 1B). The knockdown effect was higher using EXT1-1 siRNA compared with EXT1-2 siRNA. Reduction of HS chain expression was accompanied by a subtle increase in CS chain synthesis that correlated with the level of reduction of HS chains. Such an increase in CS chain expression was previously observed in EXT1-deficient ES cells derived from EXT1 knock-out mice (25). The reduction of HS chain expression in EXT1-deficient cells was also confirmed by HPLC analysis of unsaturated disaccharides from HS chains (data not shown). We also examined the molecular sizes of HS chains derived from EXT1-deficient cell surface by gel chromatography. As is shown in Fig. 1C, the lengths of HS chains in EXT1-deficient cells (ϳ35kDa) appeared to be reduced compared with control cells .
HS Chains Are Important for Self-renewal and Proliferation of mES Cells-We performed colony assays on EXT1-deficient cells to determine whether the resulting reduction in HS chain expression affected the frequency with which the cells remain in an undifferentiated state. The number of colonies derived from EXT1-deficient cells that remained in an undifferentiated state fell to about 50% that of control cells, even in the presence of LIF and serum ( Fig. 2A). We next examined proliferation of transfected mES cells. As shown in Fig. 2B, the rate of proliferation of EXT1-deficient cells decreased significantly compared with control cells. These results suggest that the reduction in HS chain expression inhibited the capacity for both self-renewal and proliferation of mES cells, presumably because of reduced levels of HS chain-dependent signaling.
The Reduction of HS Chain Expression Increases Spontaneous Differentiation of mES Cells into Extraembryonic Endoderm-We compared the morphologies of control and EXT1-deficient cells. Four days after transfection in the presence of LIF, control cells appeared to be undifferentiated cells with modest packed morphology in which the cells preferentially adhered to one another, but differentiated cells with a flattened morphology also existed at low levels because of the feeder-free culture conditions (Fig. 3A). In contrast, almost all of the EXT1-deficient cells exhibited a flattened, differentiated morphology that, in some cells, was reminiscent of the stellate morphology of the parietal endoderm (37) (Fig. 3A). Furthermore, expression of Oct3/4 and Nanog, which are markers of the undifferentiated state, was significantly decreased in EXT1-deficient cells compared with control cells (Fig. 3B), indicating that spontaneous differentiation of EXT1-deficient cells occurred more frequently than of control cells. Upon withdrawal of LIF, control cells exhibited a variety of flattened and differentiated morphologies (data not shown), suggesting that pluripotency had been maintained. However, most of the EXT1-deficient cells exhibited a parietal endoderm-like morphology (data not shown), as was the case in the presence of LIF.
To further characterize the transfected mES cells, we evaluated expression of several germ layer markers by real time PCR analysis of cells 4 days after transfection (Fig. 3, B-D). In the presence of LIF, we detected higher expression in EXT1-deficient cells of markers of the extraembryonic endoderm lineage (primitive endoderm, Gata4 and Gata6; parietal endoderm, Dab2 and LamininB1; and visceral endoderm, Bmp2 and Ihh) than control cells (Fig. 3C), whereas other lineage markers (such as the trophoblast marker, Cdx2; the primitive ectoderm marker, Fgf-5; the neuroectoderm marker, Isl1; and the mesoderm marker, Brachyury) were weakly expressed (Fig. 3D). These results reflect the morphologies of EXT1-deficient cells as shown in Fig. 3A. Following withdrawal of LIF, EXT1-deficient cells underwent further differentiation. We observed only induction of extraembryonic endoderm lineage markers in these cultures compared with control cells (Fig. 3, C and D). Upon induction of differentiation of control cells, various markers of differentiation, including Cdx2, Fgf-5, and Brachyury, and extraembryonic endoderm lineage markers exhibited a further increase in expression associated with a decrease of expression of Oct3/4 and Nanog, indicating that pluripotency had been maintained. Our analyses indicate that HS chains are important for the maintenance of the undifferentiated state and of pluripotency of mES cells and that signaling pathways mediated by HS chains may be involved in signaling pathways that control the differentiation of mES cells into the extraembryonic endoderm lineage.
We next examined the formation of EBs in suspension culture in control and EXT1-deficient cells. EB formation occurs during the in vitro differentiation of ES cells into the three germ layers, mesoderm, ectoderm, and endoderm. This differentiation process resembles the formation of the developing conceptus at the egg-cylinder stage. As shown in Fig. 3, E and F, almost all control cells formed large fluid-filled cystic EBs 6 days after EB formation, in which differentiation into the three germ layers had begun to proceed. In contrast, 53.02 Ϯ 5.83% of EXT1deficient cells formed noncystic and significantly small EBs. This suggests that the reduction in HS chain expression inhib-ited differentiation of EXT1-deficient cells into the three germ layers, supporting the results that differentiation of EXT1-deficient cells into nonextraembryonic endoderm lineage was inhibited in normal culture (Fig. 3, D and E). Similar results were obtained in cells transfected with the second siRNA expression plasmid targeting EXT1 (EXT1-2) and also in a second mES cell line (E14TG2a; data not shown).
The Reduction of HS Chains Down-regulates Specific Signaling-Several signaling molecules have been shown to be important for the maintenance of mES cell self-renewal, for example, LIF/STAT3, BMP/Smad, Wnt/␤-catenin, phosphoinositide 3-kinase (PI3K)/Akt, and members of the Src family (7-12, 38, 39). Therefore, we performed Western blot analysis of cell lysates prepared 2 days after transfection of control and EXT1-deficient cells to determine whether the activity of signaling molecules was affected by the reduction of HS chain expression. We observed a similar increase in the level of phosphorylation of Akt and Smad1 in control and EXT1-deficient cells following exposure to 15% FBS (supplemental data 4), suggesting that the serum-responsive signaling by PI3K and Smad was not affected by the reduced expression of HS chains. Moreover, we found that treatment of control or EXT1-deficient cells with LIF or BMP4 gave rise to similar increases in the phosphorylation of STAT3 and Smad1, respectively (Fig. 4, A and B). Heparitinase treatment was performed to examine the effect of HS chain depletion on BMP/Smad and LIF/STAT3 signaling. This treatment led to a reduction in BMP/Smad signaling but not of LIF/STAT3 signaling (Fig. 4, C and D), demonstrating that HS chains contribute to BMP signaling but not LIF signaling in mES cells. The level of phosphorylation of ERK in response to basic FGF treatment was reduced in EXT1-deficient cells compared with the level observed in control cells, suggesting that FGF signaling was reduced in EXT1-deficient cells (Fig. 4B) and more reduction was detected after heparitinase treatment (Fig. 4, C and D). Phosphorylation of the Src family member, cYes, which has been implicated in the maintenance of self-renewal of ES cells (38), was activated to a similar extent in both control and EXT1-deficient cells following treatment with LIF and FBS (Fig. 4, A and B). We observed comparable results in similar studies with E14TG2a cells (data not shown).
HS Chains Regulate Autocrine/Paracrine Wnt/␤-Catenin Signaling in mES Cells-In the absence of feeder cells, we observed a significantly higher level of phosphorylation of ␤-catenin in EXT1-deficient cells than control cells (Fig. 5A). This suggests that there was a decrease in autocrine/paracrine Wnt/␤-catenin signaling in EXT1-deficient cells. Next, we examined Wnt/␤-catenin signaling using a luciferase reporter system. We found a significant decrease in luciferase activity in EXT1-deficient cells compared with control cells under feederfree culture conditions (Fig. 5B). This is consistent with a reduction in autocrine/paracrine Wnt/␤-catenin signaling in EXT1-deficient cells. We subsequently examined the nuclear localization of ␤-catenin, an indicator of activation of the canonical Wnt pathway. Accumulation of ␤-catenin was significantly decreased in the nuclei of EXT1-deficient cells compared with control cells (supplemental data 5). We carried out a Western blot analysis and confirmed that nuclear accumula- tion of ␤-catenin was reduced in EXT1-deficient cells compared with control cells (Fig. 5C). Thus, activation of Wnt/␤catenin signaling appeared to be reduced in EXT1-deficient cells. Similar results were observed in analogous studies using E14TG2a cells (data not shown).
RT-PCR analysis showed that both R1 and E14TG2a cell lines expressed several Wnts, as did MEFs. Wnt2, Wnt4, and Wnt5a were expressed in all cells, but Wnt3a, Wnt7a, and Wnt10b were expressed only in the mES cells (Fig. 5D). We examined the effect of reduction of HS chains on the ability of Wnt3a to bind to the mES cell surface. EXT1-deficient cells exhibited a significantly lower specific 125 I-Wnt3a binding than control cells (Fig. 5E), indicating that HS chains contribute to the binding of Wnt3a to the mES cell surface. Furthermore, SPR analysis (supplemental data 6) confirmed binding of Wnt3a to heparin, the structural analogue of HS chains, indicating that Wnt3a does indeed bind to HS chains on mES cells (apparent association and dissociation rate constants k a (M Ϫ1 s Ϫ1 ) and k d (s Ϫ1 ), 2.22 ϫ 10 5 , 5.77 ϫ 10 Ϫ3 , respectively, apparent equilibrium constant K D (nM) for binding of Wnt3a to heparin, 26.0). These results demonstrate that HS chains regulate autocrine/ paracrine Wnt/␤-catenin signaling in mES cells.
HS Chain Regulation of Wnt/␤-Catenin Signaling Contributes to Self-renewal of mES Cells-To determine whether the regulation of Wnt/␤-catenin signaling by HS chains is necessary for the self-renewal and pluripotency of mES cells, we examined the effect of exogenous activation of Wnt/␤-catenin signaling in EXT1-deficient cells using BIO, a specific pharmacological inhibitor of glycogen synthase kinase-3. The level of luciferase activity of Wnt/␤-catenin signaling in BIO-treated    cells was 2-3-fold higher than in cells treated with MeBIO, an inactive analogue of BIO (data not shown). Two days after transfection, the level of Nanog mRNA was significantly upregulated in BIO-treated cells (Fig. 6A), indicating that Nanog expression is regulated by Wnt/␤-catenin signaling. The levels of Nanog and Oct3/4 mRNAs in MeBIO-treated EXT1-deficient cells were 30 -50% of those in MeBIO-treated control cells (Fig. 6A). Untreated cells showed similar results (data not   shown). This suggests that decreased signaling in EXT1-deficient cells affects the expression of Nanog and Oct3/4. BIO treatment rescued the level of Nanog mRNA in EXT1-deficient cells, but not of Oct3/4 (Fig. 6A). The signaling pathways that regulate Oct3/4 expression are therefore dependent upon HS chain expression but not upon Wnt. Furthermore, the expression patterns of Nanog and Oct3/4 proteins were correlated with mRNA levels (Fig. 6B). Recently, it has been demonstrated that the orphan nuclear receptor LRH-1 is required for maintenance of Oct3/4 expression in mES cells (40). The level of LRH-1 mRNA in EXT1-deficient cells treated with MeBIO was also reduced to 30 -50% that in control cells (Fig. 6A), similarly to Oct3/4 expression, but was unaffected by BIO treatment (Fig.  6A). This suggests that the reduction in Oct3/4 expression in EXT1-deficient cells is mediated by LRH-1. Overall, our analyses demonstrate that Wnt/␤-catenin signaling through HS chains regulates Nanog expression but not that of Oct3/4 in mES cells.
Next, we examined the ability of BIO to rescue self-renewal in EXT1-deficient cells. The number of AP-positive colonies derived from EXT1-deficient cells treated with MeBIO was significantly reduced compared with that of control cells (Fig. 6C).
Nearly all of the colonies derived from culture of EXT1-deficient cells in the presence of BIO exhibited a tightly packed and AP-positive morphology reminiscent of undifferentiated ES cells cultured on a feeder layer. BIO treatment therefore rescued defective self-renewal in EXT1-deficient cells (Fig. 6C). This rescue effect demonstrates that HS chain-dependent signaling by Wnt/␤-catenin contributes to self-renewal of mES cells. However, the total numbers of colonies scarcely differed between BIOtreated cells and MeBIO-treated cells (Fig. 6C), indicating that BIO treatment had no effect on cell proliferation. In a proliferation assay, we observed no effect of BIO on mES cell proliferation compared with untreated and MeBIO-treated cells (data not shown). These results demonstrate that HS chain-dependent signaling by autocrine/ paracrine Wnt/␤-catenin is required for the maintenance of self-renewal but not for the proliferation of mES cells.
Wnt/␤-Catenin Signaling Regulated by HS Chains Is Important for Pluripotency of mES Cells-Our next step was to examine the effect of BIO on EXT1-deficient cell pluripotency in the absence of LIF, in which ES cells spontaneously differentiated into several lineages (Fig. 3, B-D). The expression of differentiation markers was analyzed 4 days after transfection using real time PCR analysis (Fig. 6, D-F). The expression of various differentiation markers was examined in EXT1-deficient cells treated with either BIO or MeBIO. BIO induced expression of the markers, including Cdx2, Fgf-5, and Brachyury; expression of the differentiation markers was at a low level in the MeBIO-treated cells. In the latter treatment group, only markers of the extraembryonic endoderm lineage (Gata6, LamininB1, and BMP2) were detected at appreciable levels (similar to those described in Fig. 3, C and D). Therefore, BIO rescued the defective pluripotency of EXT1-deficient cells.
Nanog-deficient mES cells spontaneously differentiate into the extraembryonic endoderm lineage, implicating Nanog in the control of mES cell pluripotency (14,15). Nanog and Oct3/4 expression was reduced in EXT1-deficient cells treated with MeBIO, whereas treatment with BIO rescued the level of Nanog expression although only to the same level as in MeBIO-treated control cells (Fig. 6D). Thus Wnt/␤-catenin signaling sustains Nanog expression. In control cells, BIO treatment resulted in reduced expression of various differentiation markers associated with maintenance of Nanog and Oct3/4 expression com- pared with MeBIO treatment. This indicates that BIO inhibited differentiation of control cells. In turn, maintenance of Nanog expression by autocrine/paracrine Wnt/␤-catenin signaling through HS chains is important for the maintenance of pluripotency of mES cells.

DISCUSSION
In this study, we have demonstrated that EXT1-deficient mES cells, which express greatly reduced levels of HS chains, proliferate slowly and differentiate spontaneously into the extraembryonic endoderm lineage. Signaling by Wnt and FGF, but not BMP and LIF, was disrupted in EXT1-deficient cells. Furthermore, we demonstrated that autocrine/paracrine Wnt/␤-catenin signaling through HS chains was necessary for maintenance of Nanog expression, self-renewal, and pluripotency in mES cells even in the presence of LIF/ STAT3 and BMP/Smad signaling. Our results indicate that HS chains are important regulators of the extrinsic signaling  pathways contributing to self-renewal and pluripotency of mES cells.
We designed constructs expressing two different siRNAs targeting EXT1. Knockdown of EXT1 in response to EXT1-1 siRNA was maintained at ϳ20% of the level observed in control cells for at least 3 days following selection, even in the absence of puromycin. However, when EXT1-2 siRNA was used, the level of EXT1 mRNA recovered to 50 -60% of that of control cells 3 days after selection (data not shown). As a result, the effects of EXT1 knockdown were somewhat weaker in cells transfected with the EXT1-2 plasmid (Figs. 2-5), reflecting a response that correlated with EXT1 mRNA levels. These observations strongly support the conclusion that our results reflect the function of HS chains.
In this study, we used a transient knockdown system and found that the efficiency of knockdown decreased daily following the ending of selection. We therefore used puromycin to select for efficient knockdown in cells throughout the experiments described in Fig. 3 and Fig. 6, D-F. However, in the selfrenewal and proliferation assays ( Fig. 2 and Fig. 6C), the cells had to be replated at low densities, which rendered them susceptible to puromycin-induced cell death. Thus, puromycin could not be used after replating. These different conditions underlie the apparent experimental disparities such as that between Fig. 2A, in which undifferentiated colonies derived from EXT1-deficient cells were present at half of the level of control cells, and Fig. 3A, in which almost all EXT1-deficient cells differentiated.
Several signaling molecules, such as LIF/STAT3, BMP/ Smad, Wnt/␤-catenin, PI3K/Akt, and Src family members, are required for self-renewal of mES cells (7-12, 38, 39). Until now, it has been believed that extrinsic stimulation by both LIF and serum, including BMP, was sufficient to maintain mES cell selfrenewal (4). However, in this study, we showed that self-renewal of mES cells was significantly decreased in response to the reduction in HS chain expression, even when LIF-and serum-mediated signaling were not reduced ( Fig. 2 and Fig.  4). As shown in Fig. 5, several Wnts were expressed in mES cells, indicating that autocrine/paracrine Wnts function in mES cells. Indeed, autocrine/paracrine Wnt/␤-catenin signaling was actually observed in cells cultured under feederfree conditions. Furthermore, we showed that Wnt3a binds to the mES cell surface mediated by HS chains, and we demonstrated that autocrine/paracrine Wnt/␤-catenin signaling through HS chains is important for maintenance of self-renewal of mES cells (Fig. 5 and Fig. 6). Thus, we propose that maintenance of mES cell self-renewal requires not only LIF plus serum factors but also autocrine/paracrine Wnts.
Reduction of HS chain expression also led to decreased signaling by FGF (Fig. 4). It has been demonstrated that HS chains regulate mouse fibroblast cell proliferation through FGF signaling (41). FGF has been also reported to maintain self-renewal of human ES cells (42), although this effect has yet to be observed in mES cells. We speculate that FGF signaling through HS chains may contribute to the maintenance of self-renewal and proliferation of mES cells, if FGF is present in serum or the conditioned medium. This idea is supported by the fact that HS chains have been shown to play crucial roles in FGF signaling during development (18).
We did not observe a reduction in BMP/Smad signaling in response to reduced short HS chains in mES cells (Fig. 4B), although previous reports have indicated that HS chains contribute to signaling by BMP in Drosophila and Xenopus (18,43). The specific decrease in signaling by Wnt and FGF, but not by BMP, mediated by short HS chains in EXT1-deficient cells may be because of the effects of reduced HS chain length on the ability to accumulate sufficient ligands for their cognate receptors. This is supported by the result that HS chain depletion by heparitinase treatment led to not only further reduction in FGF signaling but also a reduction in BMP/Smad signaling (Fig. 4, C  and D).
IL-6 is a heparin/HS-binding cytokine, and HS chains may regulate paracrine IL-6 signaling (44). Although LIF is an IL-6 family member and might also bind to HS chains, we demonstrated that HS chains are not required for LIF/STAT3 signaling. Neither reduction of HS chains (Fig. 4, A and B) nor HS chain depletion by heparitinase treatment (Fig. 4, C and D) had an effect on LIF/STAT3 signaling. Taken together, HS chains contribute to the regulation of several signaling pathways mediated by Wnt, BMP, and FGF but not LIF in mES cells (Fig. 7).
As shown in Fig. 6C, the total numbers of colonies scarcely differed between BIO-treated cells and MeBIO-treated cells, and we observed no effect of BIO on mES cell proliferation (data not shown), indicating that BIO treatment had no effect on cell proliferation. As shown in Fig. 2B, HS chains contribute to mES cell proliferation. Thus, we suggest that HS chain-mediated signaling pathways that are not influenced by Wnt/␤catenin control mES cell proliferation.
Nanog expression was increased in response to BIO treatment in both control and EXT1-deficient cells (Fig. 6A). However, the difference in the extent of increase in the expression of Nanog mRNA following BIO treatment of control and EXT1deficient cells (Fig. 6A) indicates that other signaling pathways contribute to the control of its expression through HS chains.
Recently, it has been demonstrated that Nanog transcription is regulated by an interaction between Oct3/4 and Sox2 or a novel pluripotential cell-specific Sox element-binding protein (45,46). As such, the difference in the extent of the increase in the expression of Nanog mRNA following BIO treatment of control and EXT1-deficient cells may be due to a possible reduction in Oct3/4 and Sox2 or pluripotential cell-specific Sox elementbinding protein mediated by unknown signaling through HS chains.
Expression of Oct3/4 and LRH-1 mRNA was decreased in EXT1-deficient cells treated with MeBIO (Fig. 6A), and their expression was not affected by BIO treatment in either control or EXT1-deficient cells (Fig. 6A). Thus regulation of Oct3/4 and LRH-1 expression does not require Wnt/␤-catenin signaling through HS chains. LRH-1 has been shown to regulate the expression of Oct3/4 by binding to its proximal enhancer and promoter (40). LRH-1 has also been shown to play an important role in the regulation of cell proliferation (47). The signaling pathways controlling the expression of LRH-1 remain to be determined. The decrease in Oct3/4 expression observed in EXT1-deficient cells may be depend-ent upon the decrease in LRH-1 expression because of reduced signaling in these unidentified pathways in an HS chain-dependent manner. Although the signaling pathways activated downstream of HS chain expression remain to be determined, those required for LRH-1 expression might regulate the proliferation of mES cells.
In conclusion, we have demonstrated the importance of HS chains for the regulation of self-renewal and pluripotency of mES cells. These results suggest that it is possible to modify self-renewal and differentiation of ES cells by the use of RNAi targeted to genes encoding enzymes required for HS chain synthesis. Signaling by Wnt, FGF, and Notch have been shown to be involved in the regulation of self-renewal of hematopoietic and neural stem cells (48 -50). However, the contribution of glycochains to such signaling in these cells has not yet been determined. Modulation of the glycosylation pathways in stem cells will provide the opportunity to engineer stem cells to follow specific differentiation programs that may be of use for the development of cell replacement therapies. We demonstrated that autocrine/ paracrine Wnt/␤-catenin signaling mediated by HS chains regulates Nanog expression and sustains self-renewal by suppression of primitive endodermal differentiation. BMP/Smad signaling is regulated by HS chains (Fig. 4, C and D) and blocks neural differentiation by induction of Id (11). FGF signaling is also regulated by HS chains (Fig. 4), but the signaling pathway required for maintenance of self-renewal is unknown in mES cells. From the observations made in this study, we suggest that HS chain-dependent signaling by unknown factors regulates mES cell proliferation and Oct3/4 expression. LIF/STAT3 signaling is not regulated by HS chains (Fig. 4) and blocks non-neuronal differentiation by induction of Myc (51). *, BMP/Smad signaling is not decreased by short HS chains (ϳ35 kDa) but is decreased by depletion of HS chains.