Heparan Sulfate Facilitates FGF and BMP Signaling to Drive Mesoderm Differentiation of Mouse Embryonic Stem Cells*

Background: HS has been implicated in regulating ESC differentiation. Results: Mouse ESCs lacking EXT1 fail to differentiate into mesoderm. Restoration of the FGF and BMP signaling each partially rescued the mesoderm differentiation defect. Conclusion: HS facilitates FGF and BMP signaling to drive mesoderm differentiation. Significance: This study shows that HS essentially promotes mesoderm differentiation of mouse ESCs. Heparan sulfate (HS) has been implicated in regulating cell fate decisions during differentiation of embryonic stem cells (ESCs) into advanced cell types. However, the necessity and the underlying molecular mechanisms of HS in early cell lineage differentiation are still largely unknown. In this study, we examined the potential of EXT1−/− mouse ESCs (mESCs), that are deficient in HS, to differentiate into primary germ layer cells. We observed that EXT1−/− mESCs lost their differentiation competence and failed to differentiate into Pax6+-neural precursor cells and mesodermal cells. More detailed analyses highlighted the importance of HS for the induction of Brachyury+ pan-mesoderm as well as normal gene expression associated with the dorso-ventral patterning of mesoderm. Examination of developmental cell signaling revealed that EXT1 ablation diminished FGF and BMP but not Wnt signaling. Furthermore, restoration of FGF and BMP signaling each partially rescued mesoderm differentiation defects. We further show that BMP4 is more prone to degradation in EXT1−/− mESCs culture medium compared with that of wild type cells. Therefore, our data reveal that HS stabilizes BMP ligand and thereby maintains the BMP signaling output required for normal mesoderm differentiation. In summary, our study demonstrates that HS is required for ESC pluripotency, in particular lineage specification into mesoderm through facilitation of FGF and BMP signaling.


Heparan sulfate (HS) has been implicated in regulating cell fate decisions during differentiation of embryonic stem cells (ESCs) into advanced cell types.
However, the necessity and the underlying molecular mechanisms of HS in early cell lineage differentiation are still largely unknown. In this study, we examined the potential of EXT1 ؊/؊ mouse ESCs (mESCs), that are deficient in HS, to differentiate into primary germ layer cells. We observed that EXT1 ؊/؊ mESCs lost their differentiation competence and failed to differentiate into Pax6 ؉ -neural precursor cells and mesodermal cells. More detailed analyses highlighted the importance of HS for the induction of Brachyury ؉ pan-mesoderm as well as normal gene expression associated with the dorso-ventral patterning of mesoderm. Examination of developmental cell signaling revealed that EXT1 ablation diminished FGF and BMP but not Wnt signaling. Furthermore, restoration of FGF and BMP signaling each partially rescued mesoderm differentiation defects. We further show that BMP4 is more prone to degradation in EXT1 ؊/؊ mESCs culture medium compared with that of wild type cells. Therefore, our data reveal that HS stabilizes BMP ligand and thereby maintains the BMP signaling output required for normal mesoderm differentiation. In summary, our study demonstrates that HS is required for ESC pluripotency, in particular lineage specification into mesoderm through facilitation of FGF and BMP signaling.
Embryonic stem cells (ESCs) 2 are derived from the inner cell mass of blastocysts, and retain the ability to differentiate into the three primary germ layers and subsequently to all cell types of the adult body (pluripotency) (1,2). The capacity to differentiate into desired lineages and cell types endow ESCs with great promise for the development of cell replacement therapies for degenerative disease and injury. Understanding the mechanisms that control ESC self-renewal and differentiation is central to materialize this potential. Meanwhile, ESCs have been increasingly recognized as a versatile system for developmental biologists to study lineage specification and organogenesis (2). The regulatory network and molecular requirements for the maintenance of self-renewal and lineage specification of stem cells have been under intense investigation and are increasingly defined (3).
It has been clear that ESC fate decisions are influenced by their interactions with the components of their microenvironment, including soluble and immobilized factors and the extracellular matrix (3). For example, a number of growth factors and morphogens, including bone morphogenetic proteins (BMPs), Wnt/Wingless, and members of the fibroblast growth factor (FGF) family, have been elucidated as master regulators of lineage specification in ESCs. Upon differentiation, both BMP and Wnt have been shown to stimulate gene expression associated with mesoderm and to inhibit neuroectoderm differentiation (4 -6). In contrast, FGF signaling stimulates neuroectoderm differentiation and is also required for mesoderm differentiation (7,8).
Heparan sulfate (HS) is a linear polysaccharide with sulfation modification and belongs to the family of glycosaminoglycans. The biosynthesis of HS primarily takes place in the Golgi apparatus by the co-polymerases EXT1 and EXT2, which alternately adds N-acetylglucosamine and glucuronic acid residues to form HS precursors. Following chain elongation, N-deacetylase/Nsulfotransferases (NDST) act on discrete regions of the HS precursors, replacing N-acetyl groups with N-sulfate, creating appropriate substrates for further modification reactions including epimerization and O-sulfation. The modification reactions are incomplete, resulting in highly heterogeneous sulfation patterns. These clustered modification reactions generate highly sulfated domains dispersed with low/non-sulfated regions leading to a highly heterogeneous structure in mature HS (9). The heterogeneity possesses a structural basis for HS to interact with a large variety of protein ligands to modulate a wide range of biological functions (9). In tissues, the HS chains covalently attach to core proteins to form HS proteoglycans (HSPG), such as syndecans, glypicans and perlecan, and are present abundantly on the cell surface and in the extracellular matrix where HSPGs interact via HS chains with growth factors, growth factor binding proteins, extracellular proteases, protease inhibitors, chemokines, morphogens, and adhesive proteins to modulate diverse biological functions (9 -11). The essential role of HS in development has been manifested by genetic studies targeting HS biosynthetic enzymes (10). For example, mice deficient in Ext1 or Ext2 fail to undergo gastrulation, lack organized mesoderm and extraembryonic tissues, and die during early development (12,13), illustrating that HS critically regulates early embryogenesis.
HS is abundantly expressed by undifferentiated and differentiating ESCs, and becomes increasingly sulfated as ESCs undergo differentiation (14,15), suggesting that HS may critically modulate ESC differentiation. This presumption has been supported by the examination of NDST1/2 Ϫ/Ϫ or EXT1 Ϫ/Ϫ mouse ESCs (mESCs). NDST1/2 Ϫ/Ϫ mESCs, devoid of N-and 2-O-sulfation, failed to differentiate into endothelial cells, neuronal cells and adipocytes (16 -18). ESCs derived from EXT1 deficient mice (conventional EXT1 Ϫ/Ϫ mESCs) lacked HS and could not differentiate into ␤3-Tubulin-positive neuron, hemangioblasts, hematopoietic, and endothelial cells (15,19). These studies demonstrate that HS critically participates in ESC differentiation into mid-and terminally differentiated cell types. Very recently, Forsberg et al. observed that NDST1/2 Ϫ/Ϫ mESCs could take the initial step toward differentiation into all three primary germ layers: the endoderm, ectoderm and mesoderm (18). The NDST1/2 Ϫ/Ϫ mESCs express lower sulfated HS and may still allow for partial functionality of the HS chain, therefore the necessity of HS for mESC differentiation into the three germ layers remains unclear. Furthermore, the underlying molecular mechanism has not been explored if HS appeared to be essential for primary germ layer differentiation.
We previously derived a mESC line from conditionally targeted EXT1 mice and then generated EXT1 Ϫ/Ϫ mESC daughter lines by Cre recombinase treatment (20). The EXT1 deletion resulted in a block of lineage commitment due to impaired FGF signaling and the failure to down-regulate pluripotency genes (20). In this study, we bypassed the requirement of HS for lineage commitment by rescuing FGF signaling and then examined the potential of the cells to differentiate into the three primary germ layers. We observed that the EXT1 Ϫ/Ϫ mESCs differentiated into endoderm and Nestin ϩ -neural precursor cells (NPC), but failed to differentiate into Pax6 ϩ -NPCs and mesoderm. Examination of mesoderm differentiation-related signaling observed that FGF and BMP, but not Wnt, were disrupted in the EXT1 Ϫ/Ϫ mESCs, and restoration of the FGF and BMP signaling each partially rescued the mesoderm differentiation defect. We also observed that BMP4 was more prone to be degraded in conditioned medium of EXT1 Ϫ/Ϫ ESC differentiation culture than in the wild-type control, revealing that HS stabilizes BMP4 to promote BMP signaling. Therefore our study demonstrates that HS facilitates FGF and BMP signaling to drive mesoderm differentiation of mESCs.
Quantitative PCR (qRT-PCR) Analysis-Total RNA was isolated using the RNeasy kit (Qiagen) and cDNA was made from 1 g of total RNA using the SuperScript III First-Strand Synthesis System (Invitrogen). Sybr-Green RT-PCR was performed as described previously (14). Primers used are listed in supplemental Table S1.
Immunostaining-Cells were fixed with 4% PFA for 10 min at room temperature, washed three times with PBS, then incubated for 1 h in blocking buffer (3% BSA, 0.1% Triton X-100 in PBS). The anti-Brachyury antibody (Santa Cruz Biotechnology) was diluted in blocking buffer at 1:100 and applied overnight at 4°C. Secondary antibodies conjugated to Alexa fluorophores (Molecular Probes) were diluted at 1:500 in blocking buffer and applied for 1 h at room temperature. Cells were washed twice and incubated with DAPI (10 g/ml) before viewing. Fluorescent images were visualized using a fluorescence microscope (Nikon Eclipse, TE2000-S) with a 20ϫ/0.40 objective at RT and captured using a Qimaging (Retiga 1300i Fast) camera and Qcapture software version 2.90.1.
FGF Signaling-mESCs cultured in differentiation medium with or without supplement of FGF-2 at 200 ng/ml were lysed with RIPA buffer containing protease and phosphatase inhibitors (Sigma), and 30 g of whole cell lysates were resolved on 10% SDS-PAGE gels, transferred onto PVDF membranes, and blotted with anti-phospho-Erk1/2 (Cell Signaling Technology) and anti-Erk1/2 (Cell Signaling Technology) antibodies at 1:1000 (20).
Wnt Signaling-Wnt/␤-catenin activity was examined using a TCF responsive luciferase reporter vector (TOPflash, M50) containing TCF/LEF binding sites and a control vector (FOPflash, M51) that contains mutant TCF/LEF binding sites (21). Cells were transfected with 0.5 g of M50/M51 using Fugene (Roche). Transfection of a TK-Renilla construct was used as an internal control. Twenty-four hours after transfection, the medium was changed to serum-free medium containing DMEM ϩ 1X N2 (Gemini Bio-products) ϩ 1ϫ B27 (Gemini Bio-products) with or without 100 ng/ml human Wnt3 (R&D Systems). Six hours later, cells were lysed, and luciferase activity was measured.
BMP Signaling-The cells were serum starved for 6 h and then stimulated with 25 ng/ml of BMP-4 (R&D Systems) for 0 -60 min, or alternatively, the cells were cultured in differentiation medium for 0 -96 h without or with BMP-4 supplemented at 5 ng/ml. Thereafter, cells were lysed with RIPA buffer and 30 g of whole cell lysates were resolved on 10% SDS-PAGE gels, transferred onto nitrocellulose membranes and blotted with anti-phospho-SMAD1/5/8 (Cell Signaling Technology) and anti-actin (Santa Cruz Biotechnology) antibodies at 1:1000.
Western Blot Analysis of HSPGs-mESCs were cultured in ESC medium until confluency and then changed to serum-free media. After an overnight culture, the conditioned media were collected. The collected media were concentrated 10-fold using 0.5 ml 3 kDa cutoff filters (Amicon Ultra Centrifugal Filter, Millipore). A portion of the concentrated media was then subjected to heparin lyase I-III digestion for 1 h at 37°C. The heparin lyase-treated and untreated media were then resolved on a 6% SDS-polyacrylamide gel, transferred onto PVDF membrane and blotted with anti-HS antibody (10E4, Seikagaku Corporation, Tokyo, Japan) at 1:1000.
Quantification of BMP-4-mESCs were cultured for 48 h in ESC medium, after which medium was changed to DMEM containing 2.5% FBS, 1000 units/ml LIF, 1ϫ NEAA, 0.1 mM ␤-mercaptoethanol, 4 mM L-glutamine, and 10 ng/ml human BMP-4 with or without heparin (10 g/ml). After 12, 24, and 36 h, the media were collected, and the amount of human BMP-4 in the culture media was quantified using an enzyme-linked immunosorbent assay kit (Quantikine, R&D Systems). To ensure that the media were collected from wells with equal cell numbers, cells were lysed and protein concentrations were determined.

Rescue of FGF Signaling Commits EXT1 Ϫ/Ϫ ESCs to Lineage
Differentiation-EXT1 is part of a co-polymerase complex with EXT2 that catalyzes the alternate addition of glucuronic acid and N-acetylglucosamine residues to polymerize the HS chain, which subsequently becomes heavily sulfated. We previously generated EXT1 Ϫ/Ϫ mESC lines by ablating conditionally targeted EXT1 allele with Cre recombinase (20). The EXT1 Ϫ/Ϫ mESCs produce no HS (supplemental Fig. S1) and fail to commit to lineage differentiation due to a defect in FGF signaling (20). The disruption of FGF signaling retains an elevated expression of the pluripotency gene Nanog, thereby preventing cell fate commitment of EXT1 Ϫ/Ϫ mESCs (20). To restore the competence of the EXT1 Ϫ/Ϫ mESCs to commit to lineage differentiation, we strategized to supplement a high dose of FGF-2 (200 ng/ml) to our differentiation medium for the initial 2 days after withdrawal of LIF in order to bypass the requirement of HS in FGF signaling ( Fig. 1A) (18,22). Indeed, the FGF-2 addition restored MAPK signaling and down-regulation of Nanog expression in EXT1 Ϫ/Ϫ mESCs to levels comparable with wild FIGURE 1. Rescue of FGF signaling restores the competence of EXT1 ؊/؊ mESCs to commit to cell lineage differentiation. A, scheme of FGF2 treatment during mESC differentiation. B, rescue of FGF signaling in EXT1 Ϫ/Ϫ mESCs. mESCs were seeded and cultured in DMEM ϩ 10% FBS, 10% KSR, and 1000 units/ml LIF for 2 days. Following, the medium was replaced with differentiation medium (DMEM ϩ 15% FBS Ϯ FGF-2 at 200 ng/ml) and Erk phosphorylation in the cells was determined 8 h after the medium change by Western blot analysis. C, rescue of FGF signaling down-regulated Nanog expression in EXT1 Ϫ/Ϫ ESCs. mESCs were cultured in serum-containing differentiation medium Ϯ 200 ng/ml FGF-2 for 2 days and then Nanog transcript levels were determined by qRT-PCR. Error bars indicate S.E. from triplicates of the same experiment. Statistical analysis was carried out by Student's t test. *, p Ͻ 0.001. D, EXT1 Ϫ/Ϫ cells displayed representative differentiated morphology after initial two-days treatment with high FGF-2. mESCs were cultured in differentiation medium Ϯ 200 ng/ml FGF-2 for 2 days after which cells were cultured in differentiation medium only until day 6. At this condition, EXT1 Ϫ/Ϫ mESCs transited from compact dome-shaped colony morphology to flattening monolayers, similar to the morphology changes occurring in EXT1 ϩ/ϩ mESCs. The cell morphologies were documented by phase contrast micrographs. Bar, 150 m. type cells (Fig. 1, B & C). Consistent with the reduced expression of Nanog, the EXT1 Ϫ/Ϫ mESCs exhibited cell morphology alterations that are commonly associated with differentiation commitment of pluripotent cells: compact, dome-shaped colonies began to spread into flat single cell layers and showed that EXT1 Ϫ/Ϫ mESCs committed fully to cell differentiation (Fig.  1D). In contrast to the full rescue effect in EXT1 Ϫ/Ϫ mESCs, the high FGF-2 supplement did not substantially alter FGF signaling and Nanog expression in EXT1 ϩ/ϩ ESCs (Fig. 1, B & C). Taken together, these observations show that high FGF-2 supplement successfully committed EXT1 Ϫ/Ϫ mESCs to differentiation without inducing substantial changes in the differentiation dynamics of wildtype ESCs. From here onward, we used this differentiation condition as our experimental system to examine the role of HS in early differentiation of mESCs.
EXT1 Ϫ/Ϫ mESCs Exhibit Neuroectoderm and Mesoderm Differentiation Defects-To determine the role of HS in early cell differentiation, EXT1 Ϫ/Ϫ and EXT1 ϩ/ϩ mESCs were differentiated in serum-containing medium with FGF2 (200 ng/ml) supplemented for the initial 2 days of the 6-day differentiation scheme (Fig. 1A). The expression of pluripotency genes and early cell lineage-associated genes was determined by qRT-PCR analysis over the time course of differentiation. In agreement with the afore described observations, the expression of pluripotency genes Nanog ( Fig. 2A) and Rex1 (data not shown) in EXT1 Ϫ/Ϫ mESCs declined over the time course. In contrast, the expressions of endoderm-associated markers (Sox17, Foxa2) were up-regulated (Fig. 2, B & C), indicating that EXT1 ablation stimulated mESC fate to endoderm. The neuroectoderm marker Nestin was induced late and Pax6 was expressed at lower levels altogether in EXT1 Ϫ/Ϫ mESCs, showing that HS is required for differentiation into NPCs (Fig. 2, D & E). Strikingly, examination of mesoderm-associated genes observed that the expression of the pan-mesoderm marker Brachyury, ventralposterior (V-P) mesoderm markers Evx1 and Mesp1, and anterior mesoderm marker Goosecoid all failed to become up-regulated (Fig. 2, F-I), indicating that EXT1 Ϫ/Ϫ mESCs were unable to differentiate into mesoderm lineage. To determine whether the abnormal mesoderm differentiation of EXT1 Ϫ/Ϫ mESCs was truly due to HS deficiency, we supplemented the cell differentiation culture with heparin, a structural analog of HS, and examined Brachyury expression. We observed that heparin supplement at 10 g/ml fully restored Brachyury expression in EXT1 Ϫ/Ϫ mESCs (Fig. 3A). Consistent with this gene expression profile, immunostaining showed that heparin efficiently restored Brachyury expression in the differentiating EXT1 Ϫ/Ϫ mESC populations during the differentiation course (Fig. 3, A &  B). These observations demonstrate that HS is required for mESCs to differentiate into mesoderm lineage.

HS Facilitates FGF Signaling to Promote Mesoderm
Differentiation-FGF signaling is known to promote ESCs to differentiate into mesoderm (8,(23)(24)(25)(26)(27). To commit the EXT1 Ϫ/Ϫ mESCs to cell differentiation, we rescued the FGF signaling only for the first 2 days of differentiation and thereafter the FGF signaling was disrupted again. It is possible that the following FGF signaling disruption may cause the subsequent mesoderm differentiation defect of the EXT1 Ϫ/Ϫ mESCs. To test this idea, we continued the FGF supplement through the entire course of differentiation. As shown in Fig. 3A, restoration of FGF signaling partially rescued Brachyury expression in EXT1 Ϫ/Ϫ mESCs, showing that HS facilitates FGF signaling to promote mesoderm differentiation. The partial rescue effect also suggested that other signaling pathways required for mesoderm differentiation were also disrupted in EXT1 Ϫ/Ϫ mESCs.
Wnt Signaling Alteration Does Not Contribute to the Mesoderm Differentiation Defect of EXT1 Ϫ/Ϫ mESCs-Wnts are developmentally regulated molecules that are secreted and expressed by mESCs (28,29). Canonical Wnt signaling is essentially required for development of mesoderm during mESC differentiation (4 -6). HS has been shown to function as a co-receptor, to stabilize and to maintain Wnt gradients to critically modulate Wnt signaling during embryogenesis (30 -33), suggesting a possibility that HS may facilitate Wnt signaling to drive mesoderm differentiation of mESCs. To test this hypoth-esis we carried out luciferase reporter assay to measure Wntreporter activity in both EXT1 ϩ/ϩ and EXT1 Ϫ/Ϫ mESCs. Surprisingly, luciferase activity was consistently higher in EXT1 Ϫ/Ϫ ESCs than in EXT1 ϩ/ϩ ESCs in response to Wnt3 (Fig. 3C), reflecting that HS absence enhanced Wnt signaling during mESC differentiation. The elevated Wnt signaling did not correlate with the mesoderm differentiation defect in EXT1 Ϫ/Ϫ mESCs, indicating that the Wnt signaling alteration was not the cause of the mesoderm differentiation defect in EXT1 Ϫ/Ϫ mESCs.
HS Facilitates BMP Signaling to Promote Mesoderm Differentiation-BMPs modulate numerous aspects of organogenesis and early embryogenesis by regulating cell proliferation, survival and cell fate (4, 6, 34 -36). Smad1/5/8 are direct targets of BMP signaling. During mESC differentiation, BMP/ Smad signaling promotes mesoderm formation by acting on the family of Id (Inhibitors of DNA binding) transcription factors (35). HS has been shown to facilitate BMP and Decapentaplegic (the functional ortholog of mammalian BMP2 and BMP4) signaling to promote organogenesis in mice and Drosophila (37)(38)(39), suggesting that EXT1 deficiency may disrupt BMP signaling to lead to the mesoderm differentiation defect in EXT1 Ϫ/Ϫ mESCs. To test this hypothesis, we examined activation of Smads as well as induction of Id transcription upon stimulation by BMP-4. We found that phosphorylation of Smad 1/5/8 was

. Heparin and FGF2 supplementation each rescues the differentiation of EXT1 ؊/؊ mESCs into mesoderm, and Wnt signaling is up-regulated in EXT1 ؊/؊ mESCs.
A, transcript levels of Brachyury were examined by qRT-PCR. EXT1 ϩ/ϩ and EXT1 Ϫ/Ϫ mESCs were cultured in differentiation medium Ϯ 200 ng/ml FGF-2 for 2 days, and following, the cells were cultured in differentiation medium Ϯ heparin at 10 g/ml or with the same concentration of FGF-2 supplementation during differentiation day 3-8 (D3-D8). The transcript levels of pan-mesoderm marker Brachyury were examined over the time course of the differentiation. The data presented are summarized from three independent experiments in triplicate. Error bars indicate S.E. generated from triplicates of the same experiment. B, immunostaining of Brachyury expressed by mESCs during differentiation Ϯ heparin at 10 g/ml. Nuclei were stained with DAPI. Bar, 20 m. Abbreviation: HEP, heparin. C, analysis of Wnt signaling by luciferase reporter assay. mESCs transfected with reporter constructs (TOPflash or FOPflash) were stimulated with recombinant Wnt3 (50 ng/ml) and with bovine serum albumin (BSA) as negative control, and then luciferase activity was determined. Statistical analysis was carried out by Student's t test. *, p Ͻ 0.01; **, p Ͻ 0.001. reduced in response to BMP-4 in EXT1 Ϫ/Ϫ mESCs compared with BMP-4 in the EXT1 ϩ/ϩ mESC control, and the signaling attenuation was analogous to EXT1 ϩ/ϩ mESCs treated with Noggin, a specific BMP signaling inhibitor (Fig. 4, A & B). Furthermore, phosphorylation of Smad1/5/8 at a steady state of differentiation culture was also significantly reduced in EXT1 Ϫ/Ϫ mESCs (Fig. 4B), reflecting that BMP signaling was consistently attenuated during differentiation. Consistent with these observations, the transcript levels of Id1 and Id3 were lower in EXT1 Ϫ/Ϫ and Noggin-treated EXT1 ϩ/ϩ mESCs than in untreated EXT1 ϩ/ϩ mESCs under both examining conditions: shortly after BMP-4 stimulation of serum-starved mESCs, and steady state in serum-free medium containing BMP-4 with or without Noggin addition (Fig. 4, C & D). Heparin addition rescued Smad1/5/8 phosphorylation and partially restored Id1 expression in a dose-dependent manner at concentrations between 1-10 g/ml (Fig. 4, E & F), supporting our hypothesis that the impaired BMP signaling in EXT1 Ϫ/Ϫ mESCs was due to HS deficiency. Taken together, our results indicate that efficient BMP signaling in mESCs requires HS.
To determine whether EXT1 ablation disrupts BMP signaling to result in the aberrant mesoderm differentiation of EXT1 Ϫ/Ϫ mESCs, we explored whether restoration of BMP signaling would rescue the mesoderm differentiation defect. We first tested to see whether the addition of recombinant BMP-4 could restore BMP signaling/Id1 expression in EXT1 Ϫ/Ϫ mESCs. Indeed, supplementing the differentiation medium with BMP-4 at 2.5-5 ng/ml rescued Id1 expression in EXT1 Ϫ/Ϫ mESCs comparable to that of EXT1 ϩ/ϩ mESCs (Fig. 5A). Next, we examined whether the BMP-4 addition would restore mesoderm differentiation in EXT1 Ϫ/Ϫ mESCs by measuring expression of mesoderm-associated genes that are directly dependent on BMP signaling. We examined two mesodermassociated genes that have previously been shown to be activated by BMP signaling, Evx1 and Mesp1, and two genes that have been shown to be suppressed by BMP signaling, Hoxb1 and Foxa2, during ESC differentiation (4). Consistent with the previously published results, inhibition of BMP signaling in EXT1 ϩ/ϩ mESCs with Noggin resulted in lower expression of posterior genes Evx1 and Mesp1 and higher expression of anterior genes Hoxb1 and Foxa2 (Fig. 5, B--E). Differentiation of EXT1 Ϫ/Ϫ mESCs in the presence of supplemented BMP-4 at 5 ng/ml elevated Evx1 and Mesp1 expression and suppressed Foxa2 and Hoxb1 expression with mRNA levels approximating

. EXT1 ؊/؊ mESCs show attenuated BMP/Id signaling during differentiation and the attenuation is rescued by heparin.
A, Phosphorylation of Smad1/5/8 upon BMP-4 stimulation. mESCs were initially cultured in serum-containing differentiation medium with FGF-2 supplement at 200 ng/ml for 2 days and then were changed to serum-free medium for 6 h. The cells were then stimulated with BMP-4 and phosphorylation of Smad1/5/8 was examined by Western blot analysis. B, phosphorylation of Smad1/5/8 at steady-state differentiation culture. The ESCs were initially cultured in differentiation medium with FGF-2 supplement at 200 ng/ml for 2 days. Following, the cells were cultured in differentiation medium with or without supplement of Noggin and examined for levels of Smad1/5/8 phosphorylation at steady-state culture. Actin levels were used as internal controls. C, Id1 and Id3 expression upon BMP4 stimulation. The mESCs were treated as described in A. Transcript levels of Id1 and Id3 were examined after stimulation of serum-starved differentiating mESCs with BMP-4 in the absence of Noggin. Error bars indicate S.E. generated from triplicates of the same experiment. D, Id1 and Id3 expression at steady-state with or without BMP-4 supplement. The mESCs were initially cultured in differentiation medium with FGF-2 supplement at 200 ng/ml for 2 days. Following, the cells were cultured in serum-free medium in the absence or presence of 25 ng/ml BMP-4 Ϯ 100 ng/ml Noggin (NOG), and were examined for Id expression by real-time RT-PCR analysis. Error bars indicate S.E. generated from triplicates of the same experiment. E and F, heparin rescues BMP4-Smad1/5/8 signaling and Id expression. mESCs were differentiated for 2 days with FGF-2 supplement at 200 ng/ml, and then the cells were cultured in differentiation medium Ϯ heparin at 10 g/ml. Smad 1/5/8 phosphorylation and Id1 mRNA levels were examined. Arrows indicate the pSmad1/5/8 bands which showed difference between EXT1 ϩ/ϩ and EXT1 Ϫ/Ϫ samples. Error bars indicate S.E. generated from triplicates of the same experiment. Statistical analysis was carried out by Student's t test. *, p Ͻ 0.001. those in EXT1 ϩ/ϩ mESCs, illustrating directly that the compromised BMP signaling in EXT1 Ϫ/Ϫ mESCs leads to their mesoderm differentiation defect (Fig. 5, B--E). In combination with our aforementioned finding that efficient BMP signaling in mESCs requires HS, these results demonstrate that loss of HS leads to aberrant expression of mesodermal genes as a direct result of defective BMP signaling and reveals that HS derives mesoderm differentiation via facilitation of BMP signaling.
HS Stabilizes BMP-4 to Facilitate BMP Signaling-Our short-term BMP4-stimulaton experiment observed that Samd1/5/8 phosphorylation was significantly attenuated in EXT1 Ϫ/Ϫ mESCs compared with EXT1 ϩ/ϩ mESCs (Fig. 4A), suggesting a co-receptor function for HS in facilitation of BMP signaling, analogous to the well-recognized regulatory mode of HS on FGF signaling (40). This notion was also supported by our transcript analysis showing that mESCs abundantly expressed cell surface HSPGs, including both syndecans and glypicans (Fig. 6A). Meanwhile, we also realized that our mESC differentiation analysis was a long-term experiment that lasted up to 8 days. Under this experimental condition, the stability of BMP4 in culture was a critical requirement for the signaling (41). Extracellular HS has also been suggested to function to stabilize ligand in order to modulate extrinsic signaling (10,11). Cell surface HSPGs can be shed from the cell surface to extracellular matrix by proteolytic cleavage (42,43). Meanwhile, our transcript analysis observed that mESCs also highly expressed extracellular HSPGs, including perlecan and agrin (Fig. 6A). The presence of extracellular HSPGs in EXT1 ϩ/ϩ mESCs conditioned medium was confirmed by Western blot analysis probing with anti-HS antibody (Fig. 6B). As predicted, extracellular HSPGs in EXT1 Ϫ/Ϫ mESC culture did not carry any HS (Fig. 6B). It is possible that the HS deficiency in conditioned medium may destabilize BMP4 in culture, thereby attenuating BMP signaling in EXT1 Ϫ/Ϫ mESCs. This hypothesis was tested initially by examining whether BMP stability was compromised in the EXT1 Ϫ/Ϫ mESC culture. Human BMP-4 was supplemented in mESC differentiation culture and the remaining BMP-4 in medium was determined over a time course of 12, 24, and 36 h. We found that BMP-4 levels significantly lower in conditioned EXT1 Ϫ/Ϫ mESC culture media than in EXT1 ϩ/ϩ controls at all the three time points examined (Fig. 6C). Secondly, we tested whether heparin could substitute endogenous HS to restore the stability of BMP-4 in EXT1 Ϫ/Ϫ mESC culture by supplementing BMP-4 with heparin. Heparin at 10 g/ml significantly enhanced the stability of BMP-4 in EXT1 Ϫ/Ϫ mESC culture (Fig. 6D). Together, these observations suggest that HS stabilizes BMP-4 to facilitate BMP signaling in mESCs during differentiation.

DISCUSSION
The role of HS in mESC differentiation has been examined by targeting various HS biosynthetic genes using both gene knockout and knockdown approaches (15, 17-20, 29, 44). In this study, we used conditionally targeted EXT1 Ϫ/Ϫ mESCs, which produce no HS (20). We observed previously that EXT1 Ϫ/Ϫ mESCs sustained self-renewal even in the absence of LIF and demonstrated that HS is required for mESCs to exit from the self-renewal state and to commit to differentiation (20). Other studies have supported this idea and have shown that HS-deficient mESCs remain undifferentiated or fail to commit past the primitive ectoderm stage (15,17). Here, we induced mESC differentiation by bypassing the requirement for HS in initial FGF- mediated lineage commitment and thus were able to examine cell fate decisions once mESCs were fully committed. We observed that loss of HS lead to a restricted differentiation potential of ectoderm as manifested in failure or delay to express NPC genes including PAX6 and Nestin, respectively. Similar observations were made by John et al. and Forsberg et al. that used neural differentiation protocols and found similar restrictions in neural development of EXT1 Ϫ/Ϫ and Ndst1/ 2 Ϫ/Ϫ mESCs, respectively (15,18).
We also observed that EXT1 Ϫ/Ϫ mESCs failed to efficiently give rise to mesodermal precursor cells, revealing the necessity of HS for mesoderm differentiation. In line with our observation, other reports have shown the requirement of HS to generate mesoderm derivatives including hemangioblast and hematopoietic cells (19) and adipocytes (18). Our study suggests that failure to form mesoderm-derivative and terminally differentiated cell types in HS mutant cells may arise from an early block in transitioning from stem cell stage to mesodermal precursors after the cells committed to cell differentiation.
Several extrinsic signaling pathways, including FGF, BMP, and Wnt, function to promote mesoderm differentiation of mESCs (5,6,8,26,27,(45)(46)(47). We observed that EXT1 ablation attenuated FGF and BMP, but not the Wnt signaling. The attenuation of FGF and BMP signaling correlated positively with the disrupted mesoderm differentiation phenotype, and more directly, we observed that restoration of FGF and BMP4 signaling each partially rescued the mesoderm differentiation defect, demonstrating that disruption of FGF and BMP4 signaling were the major reason for the aberrant mesoderm differentiation of EXT1 Ϫ/Ϫ mESCs. These observations led us to conclude that HS facilitates FGF and BMP signaling to drive mesoderm differentiation of mESCs.
It is well-known that endogenous HS functions as a co-receptor to facilitate FGF signaling, but the regulatory role of endogenous HS on BMP signaling appears to be more complex and depends on cell/tissue context. Our study demonstrated that HS facilitates BMP signaling in the context of mESCs. This observation is in agreement with previous reports showing that HS depletion by heparinase treatment or by knockdown of the PAPST, the transporter that transports PAPS from cytosol to Golgi for sulfation modification at HS biosynthesis, both attenuated BMP signaling in mESCs (29,44). Interestingly, Manton et al. observed that HS depletion by heparinase treatment enhanced BMP signaling in human mesenchymal stem cells during differentiation into osteoblasts (48), indicating that HS functions to inhibit BMP signaling in the context of human mesenchymal stem cells. The positive and negative regulatory roles of HS in BMP signaling have been also observed during embryogenesis in EXT1-or NDST1-deficient mice, respectively (37,49). Several mechanisms have been suggested for HS to FIGURE 6. HS/heparin stabilizes BMP-4 in mESC culture. A, HSPG expression. The HSPGs expressed by mESCs were analyzed by qRT-PCR. Error bars indicate S.E. generated from triplicates of the same experiment. B, mESCs produced soluble HSPGs in culture. The conditioned medium of serum-free mESC culture was collected and concentrated 10-fold using 0.5 ml 3 kDa cut-off filter. Following, the samples treated with or without heparinase I-III (hep'ase) were resolved by 6% SDS-PAGE gel, transferred to nitrocellulose membrane and then probed with anti-HS antibody (10E4). C, BMP-4 level in mESC culture. EXT1 ϩ/ϩ and EXT1 Ϫ/Ϫ mESC cultures were supplemented with 10 ng/ml human BMP-4. Human BMP-4 levels in conditioned media were determined by quantitative ELISA at 12, 24, and 36 h of the culture. Error bars indicate S.E. generated from triplicates of the same experiment. D, heparin stabilizes BMP-4 in EXT1 Ϫ/Ϫ culture. EXT1 Ϫ/Ϫ mESC culture was supplemented with 10 ng/ml human BMP-4 Ϯ 10 g/ml heparin. BMP-4 levels in conditioned media were determined by quantitative ELISA at 12, 24, and 36 h of the culture. Abbreviations: HEP, heparin. Statistical analysis was carried out by Student's t test. *, p Ͻ 0.001. modulate BMP signaling, including co-reception, stabilization and maintaining gradient of the ligand, and mediation of BMP internalization (39,50). In our study, BMP signaling was attenuated in EXT1 Ϫ/Ϫ mESCs upon BMP stimulation in short-term experiments, suggesting that HS may function as a co-receptor for BMP signaling. We also observed that BMP is more prone to degradation in EXT1 Ϫ/Ϫ mESC culture and that this degradation could be inhibited by heparin. This observation revealed that HS stabilizes BMP during mESC differentiation. Therefore, our observations suggest HS functions through both coreceptor and ligand stabilization modes to facilitate BMP signaling during mESC differentiation.
We observed that Wnt signaling was up-regulated in differentiating EXT1 Ϫ/Ϫ mESCs, indicating that HS negatively modulates Wnt signaling in the context of differentiating mESCs. A similar effect was also observed by Manton et al. during differentiation of human mesenchymal stem cells into osteroblasts (48). Since the up-regulation of Wnt signaling did not correlate with the disrupted mesoderm differentiation phenotype of EXT1 Ϫ/Ϫ mESCs, we postulated that the Wnt signaling alteration was not the molecular mechanism underlying the mesoderm differentiation abnormality. Previous studies by Sasaki et al. also examined the effect of HS deficiency on Wnt signaling in mESCs, and observed that knockdown of EXT1 or PAPSTs, both attenuated Wnt signaling, showing that HS positively regulates Wnt signaling in mESCs (29,44). Cell surface HS has been suggested to function as a co-receptor to facilitate Wnt signaling as implicated in the EXT1-KD and PAPST-KD studies (29,44) or in a "catch" mode to negatively modulate Wnt signaling (51). We tend to believe that cell surface HS functions in both modes at the same time to finely modulate Wnt signaling. However the apparent effect is determined by which mode plays a dominant role in the specific experimental system. The up-regulation of Wnt signaling in our EXT1 Ϫ/Ϫ mESCs indicates that HS functions to negatively modulate Wnt signaling and suggests that the "catch" mode plays a dominant role in the context of differentiating mESCs. Based on our belief, we also tentatively postulate that in the cases of EXT1-KD and PAPTS-KD mESCs, the short or undersulfated HS carried by the cells might selectively lose its co-receptor function, thereby enhancing the already dominant "catch" function mode to lead to down-regulation of Wnt signaling in EXT1-KD or PAPST-KD mESCs.
During embryogenesis, development is controlled by the concerted action of different cell signaling pathways that determine the fate of any given cell. Pluripotent ESCs respond to growth factors and morphogens to activate developmental signaling in much the same fashion as the pluripotent cells of the epiblast, therefore representing a suitable system to study the complexities of early mammalian development (2). EXT1 Ϫ/Ϫ mice have been observed to lack organized mesoderm, but the related developmental signaling defect has not been known (12). In this study, the differentiation of EXT1 Ϫ/Ϫ mESCs recapitulated the aberrant mesoderm phenotype of EXT Ϫ/Ϫ embryos, suggesting that HS may function through a similar molecular mechanism, that is the FGF and BMP signaling, to modulate mesoderm development during early embryogenesis. The BMP receptor and FGF receptor knock-out mice both exhibited failure in gastrulation and mesoderm formation (25,52), phenocopying the developmental defect of EXT1 Ϫ/Ϫ mice. Therefore, these observations are in line with our EXT1 Ϫ/Ϫ mESC study and support that HS deficiency may disrupt BMP and FGF signaling to lead to the mesoderm development defect of the EXT1 Ϫ/Ϫ mice.
In conclusion, our results demonstrate that HS directs lineage fate decisions of mESCs into mesodermal precursors through modulation of both FGF and BMP signaling. We show that HS is required for the transitioning of mESCs through a Brachyuryϩ pan-mesodermal precursor stage as well as for dorso-ventral patterning of ESC via facilitation of FGF and BMP signaling, respectively. In this respect we have also elucidated that HS modulates BMP signaling by enhancing ligand stability. Further studies that examine the fine structure requirements of HS for developmental signaling and their effects on fate decision during ESC differentiation will provide us with means to generate specific cell types for cell replacement therapies.