specific glycosaminoglycans modulate neural specification of mouse embryonic stem cells

EMBRYONIC STEM CELLS/INDUCED PLURIPOTENT STEM CELLS

Specific Glycosaminoglycans Modulate Neural Specification of Mouse

Embryonic Stem Cells

CLAIRE E. PICKFORD,aREBECCA J. HOLLEY,

aGRAHAM RUSHTON,

bMARIOS P. STAVRIDIS,

cCHRISTOPHER M. WARD,

d

CATHERINE L.R. MERRYa

aStem Cell Glycobiology Group, School of Materials Science, University of Manchester, Manchester,

United Kingdom; bTheraputic Angiogenesis Group, CRUK Paterson Institute for Cancer Research, University of

Manchester, Manchester, United Kingdom; cCentre for Oncology and Molecular Medicine, Division of Medical

Sciences, University of Dundee, Ninewells Hospital, Dundee, United Kingdom; dCore Technology Facility,

Faculty of Medical and Human Sciences, University of Manchester, Manchester, United Kingdom

Key Words. Heparan sulfate • Embryonic stem cell • Sox1 • Fibroblast growth factors • Pluripotency • Differentiation

ABSTRACT

Mouse embryonic stem (mES) cells express a low sulfatedform of heparan sulfate (HS). HS chains displayed by EScells and their progeny become more complex and more

sulfated during progression from pluripotency to neuroec-todermal precursors. Sulfated epitopes are important for

recognition and binding of a variety of ligands includingmembers of the fibroblast growth factor (FGF) family. Wedemonstrated previously that mES cells lacking HS cannot

undergo neural specification but this activity can be recov-ered by adding soluble heparin, a highly sulfated glycos-

aminoglycan (GAG). Therefore, we hypothesized thatsoluble GAGs might be used to support neural differentia-tion of HS competent cells and that the mechanisms

underlying this activity might provide useful informationabout the signaling pathways critical for loss of pluripo-

tency and early lineage commitment. In this study, we

demonstrate that specific HS/heparin polysaccharides sup-port formation of Sox11 neural progenitor cells fromwild-type ES cells. This effect is dependent on sulfation

pattern, concentration, and length of saccharide. Using aselective inhibitor of FGF signal transduction, we show

that heparin modulates signaling events regulating exitfrom pluripotency and commitment to primitive ectodermand subsequently neuroectoderm. Interestingly, we were

also able to demonstrate that multiple receptor tyrosinekinases were influenced by HS in this system. This sug-

gests roles for additional factors, possibly in cell prolifera-tion or protection from apoptosis, during the process ofneural specification. Therefore, we conclude that soluble

GAGs or synthetic mimics could be considered as suitablelow-cost factors for addition to ES cell differentiation

regimes. STEM CELLS 2011;29:629–640

Disclosure of potential conflicts of interest is found at the end of this article.

INTRODUCTION

Embryonic stem (ES) cells can self-renew indefinitely whileretaining the pluripotent capacity to differentiate into anyadult lineage [1, 2]. They, along with adult stem cells andinduced pluripotent stem cells, hold great therapeutic poten-tial, particularly for cell or tissue replacement strategies fordisorders where no current long-term alternative exists (e.g.,type I diabetes and Parkinson’s disease). Successful therapiesrequire regimens for differentiating stem cells into thedesired population with high efficiency and yield. The devel-opmental steps that control differentiation to specified prog-eny are being elucidated, with identification of discrete steps

possible via surface antigen expression and transcription fac-tor profiling. Typically, cells are driven to differentiate withspecific media preparations, often containing cocktails ofcostly protein factors or chemical inhibitors. This approach,combined with a relatively poor yield of desired end-pointcells, make many current regimens impractical on the thera-peutic scale.

Heparan sulfate proteoglycans (HSPGs) are ubiquitouscomponents of the pericellular environment, where they playcrucial roles in signaling regulation and the creation ofmorphogen gradients during development. HS has a complexsugar structure, consisting of a backbone of repeating disac-charides of glucuronic acid (GlcUA) and N-acetylglucosamine(GlcNAc), polymerized by a heteromeric complex of EXT1/

Author contribution: C.E.P.: conception and design, collection and assembly of data, data analysis and interpretation, manuscriptwriting; R.J.H.: conception and design, data analysis and interpretation; G.R.: collection of data; M.P.S.: collection of data, data analysisand interpretation, manuscript writing, approval of manuscript; C.M.W.: conception and design, approval of manuscript; C.L.R.M.:conception and design, manuscript writing, final approval of manuscript.

Correspondence: Catherine L.R. Merry, B.Sc. (Hons), Ph.D., School of Materials, University of Manchester, Materials Science Building,Manchester M13 9PL, United Kingdom; Telephone: 44-0-161-3068871; Fax: 44-0-161-3063586.; e-mail: [email protected] Received October 14, 2010; accepted for publication January 16, 2011; first published online in STEM CELLS EXPRESS February 4,2011. VC AlphaMed Press 1066-5099/2009/$30.00/0 doi: 10.1002/stem.610

STEM CELLS 2011;29:629–640 www.StemCells.com

EXT2 enzymes. This backbone is subject to modifications,including de-N-acetylation and N-sulfation of glucosamine,epimerisation of GlcUA resides to iduronic acid (IdoUA),2-O-sulfation of IdoUA residues, and 6-O-sulfaction and 3-O-sulfation of glucosamine residues [3]. Postbiosynthetic proc-essing is carried out by endo-6-sulfatases present at theplasma membrane [4]. The enzymes are both interdependentand incomplete in action, resulting in the typical HS structurewhere domains of dense sulfation, flanked by regions of inter-mediate sulfation, are spaced along a nonsulfated backbone[5]. The bank of enzymes responsible for HS biosynthesis areoften members of multi-isoform families with specific sub-strate activities and discrete patterns of expression in embry-onic and adult tissues [6, 7]. Regulation of these enzymes islikely to enable the tight tissue specificity of hypervariableHS patterning. This is critical because sulfation patternswithin HS chains govern their ability to bind and influencethe activity of soluble and matrix proteins including key fac-tors, for example, members of the fibroblast growth factor(FGF) family [8, 9]. Mice mutant for many components ofthe biosynthetic pathway have demonstrated that the majorityof modifications are critical for successful development [10–13]. HS-deficient mice were created by deleting either Ext1 orExt2 [14, 15], and the pregastrulation lethality of the pheno-types indicates a critical need for HS in tissue organizationand migration.

HS is known to be essential for FGF receptor (FGFR)interactions [16, 17] and is important in the regulation ofgrowth factor signaling. Early studies of primary neural cellsand differentiated teratocarcinoma cells demonstrated rapidalteration in HS sulfation patterns and core protein expressionconcomitant with temporal expression of FGFs [18–20]. NowHSPGs are well recognized as important modulators ofgrowth factor signaling and morphogen gradients across arange of developmental processes [21–24].

We previously used monolayer neural differentiationcoupled with a Sox1-GFP reporter line (46C) to demonstratethat the fine patterning of HS alters as ES cells differentiateto Sox1þ neural progenitor cells (NPCs), which reflectsaltered ability of HS in this system to bind protein ligands,and therefore potentially influence cell fate. We then used theExt1�/� ES cell line to show that, although in the absence ofcell surface HS ES cells fail to achieve successful neuroecto-dermal differentiation, this can be compensated by solubleheparin [25]. In this study, we have interrogated the specificrole of HS in early signaling events during exit from selfrenewal into primitive ectoderm and subsequently neuroecto-derm [26–28]. We have used the 46C ES cell line to investi-gate the effect of glycosaminoglycan (GAG) addition toHS-competent cells, demonstrating that GAG can be a posi-tive or negative regulator of differentiation, dependent on sizeand sulfation pattern of the chains. 46C ES cells with Ext1knocked down by RNA interference (46Cext1RNAi cells),with 80% reduced HS and altered sulfation patterning did notcommit to Sox1þ NPCs, confirming an absolute requirementfor correctly patterned GAG species to enable the transductionof key FGF-mediated signals across the cell membrane. Anal-ysis of phosphorylated receptor tyrosine kinases (pRTKs)reveals that, although as suggested by others the Erk pathwayis involved, a multitude of signaling events are associatedwith neural specification, many of which are HS-dependent.Therefore, these data suggest that the addition of specificGAG species, for example, those generated by chemical meth-ods with specific sequences or mimetics [29], may provide acost-effective alternative to the growth factors currently usedto drive differentiation for the production of therapeuticallyrelevant cell types.

MATERIALS AND METHODS

Materials

Heparin, HS (Celsus), chondroitin and dermatan sulfate (CSand DS), K5 polysaccharide, selectively desulfated heparins,and all GAG lyases were from Iduron (Manchester, U.K.,www.iduron.co.uk). For size-defined oligosaccharides, degreeof polymerization (dp) is dp2 for disaccharide, and so on. The46C ES cell line was a gift from Austin Smith [30, 31], andthe Ext1�/� line was generously provided by Prof. Dan Wells.PD173074 was used at 250 nM (Sigma Aldrich, St. Louis,www.sigmaaldrich.com).

Maintenance of ES Cell Lines

ES cells were maintained or differentiated in N2B27 asdescribed [25] with ES cells plated overnight in ES cell mediabefore switching to N2B27. For monitoring Sox1-GFP, cellswere trypsinized, resuspended in 0.1% formaldehyde/phos-phate-buffered saline (PBS), and analyzed on a Beckton Dick-inson FACScan.

Semiquantitative Reverse Transcriptase PolymeraseChain Reaction (RT-PCR)

A total of 1 lg RNA was used for cDNA synthesis with avianmyeoblastosis virus reverse transcriptase using oligodT primeraccording to manufacturer’s instructions (Promega, WI,www.promega.com). cDNA of 1 ll was used with BioMixRed (Bioline, London, U.K., www.bioline.com) and specificprimers as detailed in Supporting Information under the fol-lowing conditions: 95�C for 10 minutes; 30 cycles of 95�Cfor 1 minute, 55�C for 1 minute; and 72�C for 1 minute witha final 72�C for 10 minutes extension.

Quantitative RT-PCR

Quantitative polymerase chain reaction (qPCR) was per-formed using TaqMan Custom Gene Expression Assays, runand analyzed on a StepOne Plus Real-Time PCR system(Applied Biosystems, Carlsbad, CA, www.appliedbiosystems.com) using the 2�DDCT quantification method. Further detailsare available on request.

Flow Cytometry

Cells were prepared for flow cytometry as described [25].10E4 (3702551), 3G10 (3702601) (Seikagaku Corp., Tokyo,Japan, www.seikagaku.com), and SSEA1 antibodies (SantaCruz Biotechnology, Santa Cruz, CA, www.scbt.com) wereused at 1:200 dilution. Phycoerythrin-conjugated detectionantibodies (Santa Cruz Biotechnology) were used at 1:500.Fluorescence was measured in a Beckton-Dickinson (FranklinLakes, NJ, www.bd.com) FACSCalibur with CellQuest Anal-ysis software.

Isolation and Characterization of Radiolabeled HS

HS preparation and analyses were conducted and presented asdescribed previously [25] with two additions: an aliquot(50 K cpm of 3H) of HS from each pool was treated with2 M NaBH4/100 mM NaOH at 45�C for 48 hours to liberateHS chains. These were then resolved on a Sepharose CL-6Bcolumn to estimate their molecular size distribution and com-pared against the Wasteson calibration [32] for an estimate ofmolecular mass.

For analysis of cells spiked with 3H heparin, we fractio-nated the culture as follows: media was removed, cell layerwas PBS washed, treated with trypsin, and centrifuged. Thesupernatant was retained (pericellular coat fraction) and the

630 Specific GAGs Modulate Neural Specification

cell pellet lysed with 1% Triton in PBS. Radioactivity wasmeasured in a scintillation counter.

Generation of Ext1shRNAi Lines

Three short-hairpin RNAi sequences (sh2, sh3, and sh4) tar-geting Ext1 (50GATCCCGTACATGCTGGTATTCAAGGGGAAGCGGTGAAGCTTGaccgcttccccttgaataccagcatgtacTTTTTT 30, 50GATCCC GCACAAGGATTCTCGCTGTGACAGAGACAGAAGCTTGtgtctctgtcacagcgagaatccttgtgc TTTTTT30, and 50GATCCCGTGGACGAATGACTACTCCATGGTGTTGAGAAGCTTGtcaacaccatggagtagtcattcgtccacTTTTTT 30)were cloned separately into the pGE1 expression vector andamplified. 46C ES cells were transfected with equal amountsof three plasmids using Transfast Reagent (Promega). Cellswere cultured for 2 days post-transfection and G418 selectionapplied. Resistant colonies were picked, expanded, and ana-lyzed for HS knockdown by flow cytometry (10E4) and RT-PCR (Ext1). Clones were stably maintained in standard EScell culture media plus G418 for more than 20 passages.Alternatively, 46C cells were transfected with pGE1 contain-ing irrelevant shRNA targeted to Zebrafish myosin (AgilentTechnologies, Santa Clara, CA, www.genomics.agilent.com)and selected and expanded as described above.

Quantitative Western Analysis of dPErk1/2

A total of 6 lg of lysates was resolved on 4%–12% Nupagegels (Invitrogen, Paisley, U.K., www.invitrogen.com) andblotted on polyvinylidene fluoride membranes (Millipore,Billerica, MA, www.millipore.com) using standard protocols.Membranes were blocked in 5% nonfat milk in Tris-bufferedsaline (TBS) and rinsed twice in TBS. Antibodies: antiphos-pho-Erk1/2 and antiErk1/2 (Cat. Nos. 4,370 and 4,696 fromCell Signaling Technologies, Danvers, MA, www.cellsignal.com) were used at 1:2,000 in 5% bovine serum albumin/TBSwith 0.1% Tween 20 (Sigma Aldrich) (TBST) overnight.Secondary antibodies (anti-mouse IRDye800 and anti-rabbitA680 [Rockland 610-732-002 and Life Technologies A21209,respectively]) were incubated in 5% milk/TBST for 1 hour atroom temperature. The membrane was then rinsed in TBSTand scanned on a LiCor Odyssey infrared scanner. Densitome-try was performed with Odyssey 2.1 software. Relative dP-Erk levels were measured by dividing the intensity of thephosphorylated Erk bands with their corresponding totalErk band.

RTK Array Analysis

A total of 250 lg of each protein lysate was applied to indi-vidual RTK arrays (ARY014, R&D Systems, Minneapolis,MN, www.rndsystems.com) that contain duplicate spots ofcontrol and capture antibodies for different RTKs. Arraymembranes were blocked, incubated with lysates overnight at4�C, washed and incubated with antiphosphotyrosine–horseradish peroxidase antibody for 2 hours. Membranes weredeveloped with enhanced chemiluminescence reagents (GEHealthcare) and visualized with HyperFilm (GE Healthcare,Chalfont, U.K., www.gehealthcare.com). Pixel density wasanalyzed using GeneTools (Syngene, Cambridge, U.K.,www.syngene.com).

RESULTS

Soluble Heparin-Like Saccharides Can InfluenceNeural Specification in HS-Competent ES Cells

We have shown that Ext1�/� ES cells, which lack HS chains,cannot undergo neural specification, and this phenotype can

be rescued by addition of soluble heparin [25]. Therefore, wehypothesized that GAGs could be used as soluble agents tomodulate ES cell differentiation. To investigate the effect ofGAGs on HS-competent systems, we used the 46C Sox1-GFPreporter line and the N2B27 monolayer differentiation proto-col [30] to examine the effects of heparin and other GAGs onthe rate of acquisition of a Sox1þ phenotype.

We differentiated 46C cells in N2B27 for 10 days in thepresence of a range of GAG species. GAGs were added frominitiation of N2B27 culture, and included in media changesexcept where stated. Initially, polymeric heparin (>dp24)(typical structure: [IdoUA2S-GlcNS6Sþ/-3S]n) was comparedwith K5 polysaccharide (structure [GlcUA-GlcNAc]n) at 1 lg/ml. Acquisition of Sox1 expression was quantified by flowcytometry for GFP (Fig. 1A). The control condition displayedtypical Sox1 expression kinetics [30]. The addition of GAGsmodified this pattern, with heparin promoting Sox1þ pheno-type acquisition at a faster rate than control, and K5 polysac-charide resulting in a slight delay.

To probe the relationship between sulfation patterning andpotentiation of neural differentiation, we tested a panel of fulllength GAG polysaccharides (for structures see Supporting In-formation Fig. 1) and selectively desulfated heparin species(Fig. 1B). All conditions generated a similar pattern of Sox1-GFP activation to that shown in Figure 1A and no GAG spe-cies completely repressed neural specification; however, cleardifferences in activity were observed between the saccharides.Data are presented as the percentage difference in Sox1-GFPpositive cell number when compared with control at day 4,when Sox1 switch-on is exponential (for representative FACSplots, see Supporting Information Fig. 2). For heparin andHS, the proportion of Sox1þ cells within the population isincreased �60% over control. Cultures containing CS displayonly a subtle increase over control; whereas, DS results in a20% increase in the proportion of cells expressing Sox1.Addition of K5 polysaccharide results in a decrease of 20%when compared with control; interestingly N-desulfated hepa-rin is comparable in activity with K5 polysaccharide despiteretaining 2-, 6-, and 3-O-sulfate groups. Heparin lacking 2-Oor 6-O sulfate groups had negligible effect, similar to com-pletely desulfated heparin. We conclude that not only aresoluble GAGs capable of modulating neural specification in aHS-competent background but also that sulfation patterning/charge is critically important in defining this effect. In almostall conditions, the Sox1þ patterns are largely comparable atlater stages (7þ days, data not shown) suggesting that GAGsare influencing growth factor signaling during loss of self-renewal and transition to the ‘‘primed state’’ for differentiation[26, 33]. To test this, we compared 46Cs differentiated withheparin present for 10 days with cultures where heparin wasremoved at day 4 (Fig. 1C). The heparin-induced promotionof Sox1þ phenotype was conserved in both conditions sup-porting the hypothesis that GAGs are influencing the earliestevents in neural specification.

The Effect of Heparin Is Concentration- andSize-Dependent

We differentiated 46C cells in N2B27 adding increasing con-centrations of heparin from 1–10 lg/ml. The optimum con-centration for promotion of Sox1 expression was 6 lg/mlwith 65% cells Sox1þ at day 4 when compared with 30%Sox1þ in controls (Fig. 1D). Finally, we differentiated 46Ccells in the presence of heparin oligosaccharides of definedsizes (Fig. 1E). The data indicate that dp12 heparin is slightlymore efficient at promoting Sox1 acquisition than dp24,whereas dp6 heparin seems to suppress Sox1 acquisition

Pickford, Holley, Rushton et al. 631

www.StemCells.com

between days 3 and 7, such that the culture does not reach

the peak of Sox1 expression until day 8 when compared with

day 6 in dp24 and dp12-supplemented cultures.

Soluble Heparin Is Not Internalized by ES Cells

To assess whether exogenous GAGs were internalized/metab-olized by ES cells; we cultured 46C cells in N2B27 for 48

Figure 1. Soluble glycosaminoglycan species modulate Sox1-acquisition. Sox1-GFP expression was monitored by flow cytometry with triplicateinternal controls and presented as average 6 SD. (A): 46C embryonic stem cells were differentiated in N2B27 (^) containing 1 lg/ml heparin(n) or 1 lg/ml K5 polysaccharide (~). (B): Day 4 of differentiation, which is supplemented with 1 lg/ml of either heparin, heparan sulfate, CS,DS, K5 polysaccharide, or various modified heparins. (C): 46C þ heparin (day 0–10) compared with 46C þ heparin (day 0–4). (D): Day 4 of46C differentiated with degree of polymerization (dp) >24 heparin from 1 to 10 lg/ml. Statistical testing against control was performed by two-tailed t test with equal variance; *, p < .05. (E): 46C were differentiated with >dp24 heparin 6 lg/ml (n), dp12 heparin 3 lg/ml (~), dp6 hepa-rin 1.25 lg/ml (n), hence equivalent numbers of heparin chains. (F): 46C differentiated with 1 lg/ml >dp24 heparin plus 25,000 3H cpm permilliliter tritiated heparin. At 24 and 48 hours, the following fractions were measured in a scintillation counter: culture media, supernatant, andcell extract. The percentage 3H counts in each fraction is shown. Abbreviations: CS, chondroitin sulfate; DS, dermatan sulfate; GFP, green fluo-rescent protein; HS, heparan sulfate.


hours in the presence of 1 lg/ml heparin spiked with tritiatedheparin (25,000 3H cpm per milliliter). After 24 and 48 hours,media, pericellular/supernatant, and cell pellet fractions werethen assayed for retention of 3H heparin. More than 98% ofthe 3H heparin remained in the culture media after 48 hours(Fig. 1F) with a minor increased proportion detected in thepericellular coat, suggesting that most of the heparin has notbeen internalized by the cells.

Characterization of 46Cext1RNAi ES Cell Lines

The Ext1�/� mouse ES cells (mES) used in this and otherstudies [25, 34] contain a disruption of the Ext1 gene and

were obtained by gene targeting and homologous recombina-

tion [14]. To confirm that the inability of Ext1�/� ES cells to

undergo neural specification was solely due to lack of HS, weused a stable plasmid-based shRNAi approach to knockdownExt1 expression in the 46C line. Clones were propagated foranalysis, and we found differing levels of HS-knockdown ineach clone (Supporting Information Fig. 2). The clone withgreatest knockdown of HS production by immunocytochemis-

try for 10E4 was designated 46Cext1RNAi:5 along with irrele-vant control transfectants 46CirrRNAi:9 and 46CirrRNAi:10.In contrast to previous reports [35], clones could be culturedfor extended passage and displayed typical ES cell colonymorphology (Fig. 2A–2C). Flow cytometry analysis of 46Cwhen compared with 46Cext1RNAi:5 and 46CirrRNAi:9 con-firmed similar expression of pluripotency-associated glycanSSEA1 (Fig. 2D), knockdown of cell surface 10E4 (anti-HSchain) reactivity in 46Cext1RNAi:5 only (Fig. 2E) while 3G10(anti-HS ‘‘stub’’) reactivity was unaffected (Fig. 2F), suggest-ing that chain initiation was conserved whereas chain elonga-tion was abrogated. Additionally, qPCR analysis for Ext1demonstrated knock-down of transcript expression in46Cext1RNAi:5 only (Fig. 2G), whereas pluripotency andearly lineage marker expression patterns were conserved(Fig. 2H).

To quantify the effect of Ext1 knockdown on HS produc-tion, 46C parental and 46Cext1RNAi:5 ES cell lines weremetabolically radiolabeled with D-[6-3H] glucosamine hydro-chloride, which will incorporate into newly synthesized HSchains. For cell extract fractions, total protein was determined

Figure 2. Phenotype of 46Cext1RNAi mouse embyronic stem (ES) cell line. 46C were transfected with plasmids stably expressing shRNAi tar-geted to Ext1 (clone: 46Cext1RNAi:5) or irrelevant control (clone 46CirrRNAi:9 or 46CirrRNAi:10). Clones expressing the plasmids wereselected by G418 resistance. (A–C): Knockdown and control clones formed stable ES cell lines that were maintained in standard conditions for>20 passages. Pluripotency was confirmed by flow cytometry (46C, black; 46Cext1RNAi:5, blue; and 46CirrRNAi:9, pink) for SSEA1 (D), andknockdown of heparan sulfate production was assessed by flow cytometry for 10E4 (E) and 3G10 (F). Ext1 transcript levels were assessed byquantitative PCR (G) along with RT-PCR analysis of relevant ES cell markers (H).


www.StemCells.com

using standard methods. 46C cells express 11,000 HS cpmper milligram protein, whereas 46Cext1RNAi:5 cells express2,000 HS cpm per milligram protein (Table 1). This repre-sents a loss of 80% of HS production in 46Cext1RNAi:5 EScells.

HS disaccharide composition analysis of secreted and cellextracted fractions revealed that HS from the 46Cext1RNAi:5cell line had almost twice the level of 6-O-sulfation of both46C parent line and E14, an independent ES cell line, withsmaller increases in 2-O and N-sulfation (Fig. 3A, 3B). Thisis an interesting comparison with the Ext1þ/� cell line [14],which also has shorter HS chains than wild-type, increased Nand 2-O sulfation but not the increase in 6-O-sulfation seen in46Cext1RNAi:5 (our unpublished data). Alkali-release of HSchains from media fractions of 46C and 46C ext1RNAi:5revealed Kav values of 0.45 for 46C and 0.6 for 46Cext1R-NAi:5 (Fig. 3C), corresponding to chains with approximatemasses of 26 kDa and 12 kDa, respectively [32]. Assumingan approximate 450 Da per disaccharide, this corresponds to areduction in HS chain length from approximately 58 disaccha-rides in 46C cells to 27 disaccharides in 46Cext1RNAi:5.

Finally, we tested the ability of 46Cext1RNAi:5 ES cellsto commit to Sox1-GFP þ NPCs, with HS elongation sup-pressed (selection with G418) or active (G418 removed fromculture). Flow cytometry analysis for both GFP and 10E4reactivity at day 8 shows that 46Cext1RNAi:5 cells þ G418display minimal 10E4 reactivity and exhibit decreased Sox1-

Table 1. Determination of heparan sulfate (HS) produced relativeto total protein for cell extract fractions using standard methods

Cell extracted

preparations

Total protein

(mg)

Total HS 3H

cpm

HS cpm per

mg protein

46C 153 1.7 � 106 11,06246Cext1RNAi:5 141 2.9 � 105 2,055

46C and 46Cext1RNAi:5 embryonic stem cells were radiolabeledwith 3H-GlcN. Cell extract fractions were protease-treated andresolved via anion-exchange chromatography. HS-containingpools were treated with chondroitinase ABC and resolved on CL-6B sepharose to remove CS/DS disaccharides. HS-3H cpm wasthen calculated for each cell line.

Figure 3. Analysis of 46Cext1RNAi cell lines. (A, B): Disaccharide composition analysis of secreted and cell extracted heparan sulfate fromE14, 46C, and 46Cext1RNAi:5 ES cell lines (*, p < .05, **, p, .005, Student’s t test comparing 46C vs. 46Cext1RNAi:5). (C): Alkali treatmentof HS revealed Kav values of 0.45 and 0.6, respectively. Day 8 of N2B27: under selection 46Cext1RNAi:5 do not express appreciable levels ofSox1-GFP or HS (D), without selection both are upregulated (E). (F, G): 46CirrRNAi:10 clone expresses 10E4þ and Sox1-GFPþ populations inboth conditions. Abbreviations: GFP, green fluorescent protein; HS, heparan sulfate.


GFP expression (<10%). In contrast, if G418 selection isremoved, a significant proportion of the cells become 10E4reactive (35%) and 15% of the culture express 10E4 andSox1-GFP (Fig. 3D, 3E). However, the control clone 46CirrR-NAi:10 expresses both 10E4 and Sox1 6 G418 (Fig. 3F, 3G),demonstrating no impairment in neural specification. There-fore, endogenous HS-production is dispensable for self-renewal and loss of cell surface HS, via genetic deletion ortranscription inhibition, which results in a defect in commit-ment to Sox1þ NPCs.

GAGs Modulate the FGF-Dependent Step ThatInitiates Sox1 Expression

ES cells respond to autocrine FGF signaling to convert toSox1þ NPCs [28, 30]. We hypothesize that it is this step atwhich exogenous GAGs have their effect [25]. To confirm this,we used an inhibitor of FGFR-signaling, PD173074 [28]. Ashas been reported previously, 46C cells cultured in N2B27 inthe presence of PD173074 do not activate Sox1 when comparedwith dimethyl sulfoxide–only controls (Fig. 4A). Addition ofheparin dp12 to cells in the absence of inhibitor results in pro-motion of Sox1þ phenotype; however, if heparin is added in thepresence of PD173074, this effect is abrogated. This demon-strates that when FGF-signaling is inhibited, heparin cannotinfluence the rate of Sox1-acquisition and conversion to NPCs.

It has been shown recently that ES cells in vitro fluctuatebetween self-renewal and a ‘‘threshold of differentiation,’’ andthat FGF/Erk signaling is necessary to progress from this state[27]. To identify where Ext1�/� cells arrest, we performedquantitative expression analyses, comparing 46C and Ext1�/�

ES cells alongside 46C, Ext1�/� and Ext1�/� and the rescuecondition, Ext1�/� þ heparin, at day 4 of differentiation. Thenaı̈ve pluripotency marker Klf4 behaves indistinguishablybetween the three conditions (Fig. 4B). Interestingly Ext1�/�

cells have higher Nanog expression than 46C in ES cells, butthis difference disappears following differentiation (Fig. 4C).Nevertheless, a third naı̈ve pluripotency marker Rex1 is notsignificantly reduced in Ext1�/� cells at day 4, whereas inboth control and rescue conditions it is downregulated (Fig.4D). Oct4 mRNA is reduced by about half in control cells atday 4, but has not significantly downregulated from ES cell-levels in Ext1�/� cells, even with the inclusion of heparin(Fig. 4E). To discover whether Ext1�/� ES cells are capableof progressing to primed primitive ectoderm state, we ana-lyzed Fgf-5 expression (Fig. 4F). Fgf-5 is absent in ES cellsbut upregulated in all conditions by day 4. Taken together,and considering the heterogeneity of differentiating ES cellcultures, these data suggest that Ext1�/� cells have left naivepluripotency/ground state as defined by NanogþRex1þKlf4þ

and a proportion have entered the primitive ectoderm-likestage (Fgf-5þRex1�oct4þ) [27] although the high Rex1 at day4 indicates that not all the population are in this state.

Multiple Pathways Are Altered During NeuralDifferentiation of Ext12/2

To investigate the involvement of non-Erk1/2 signaling path-ways in the acquisition of neural specification, we used aRTK antibody array to compare 46C versus Ext1�/� ES cellswith 46C, Ext1�/� and Ext1�/� þ heparin at day 4 of N2B27.Membranes are prespotted with 39 specific RTK antibodies,allowing simultaneous detection of receptor tyrosine phospho-rylation (Supporting Information Fig. 4). Spot density analysissoftware assigns positive control spots (A1, 2; F1, 2; A23, 24)a value of 100 and PBS spots (E7, 8) a value of 0. 46C EScells have considerably higher levels of phospho-ErbB2 thanExt1�/� (Fig. 5A) and at day 4 ErbB2 is still highly phospho-

rylated in 46C but is much lower in Ext1�/�; the addition ofheparin results in increased pErbB2 (Fig. 5B). Phosphorylatedplatelet-derived growth factor receptor (PDGFR) at equivalentlevels in ES cells (Fig. 5A) is considerably increased in 46Cat day 4, but is not detectable in Ext1�/� day 4, unless hepa-rin is included (Fig. 5B). Phosphorylated macrophage-stimu-lating protein receptor (pMSPR), a hepatocyte growth factorreceptor family member, is not detectable in Ext1�/� ES cellsor day 4 cells, whereas wild-type ES cells have pMSPR andincrease its phosphorylation with differentiation (Fig. 5A vs.Fig. 5B). In this case, the inclusion of heparin has only asmall effect on increasing pMSPR (Fig. 5B).

Epidermal growth factor receptor (EGFR) is not phospho-rylated in either ES cell line (Fig. 5A), but its phosphorylationis also increased significantly in 46C by day 4, Ext1�/� havealso increased pEGFR but not to the same level, and in thiscase the inclusion of heparin has no effect (Fig. 5C). Simi-larly phosphorylated Ephrin B4, which is not detected in EScells (nor are other members of the Ephrin family [data notshown]), is highly abundant in wild-type cells at day 4, butExt1�/� cells do not phosphorylate EphB4 even with theinclusion of heparin (Fig. 5C).

Finally, pTrkC, a neural-associated RTK, is absent fromES cells (Fig. 5D) and strongly detected in 46C by day 4 ofdifferentiation, not in Ext1�/� or, interestingly, Ext1�/� þheparin. Similarly, two vascular endothelial growth factor(VEGF) receptor isoforms are phosphorylated in wild-typecells as a function of differentiation, but not in Ext1�/� andthe inclusion of heparin cannot recover this (compare Fig. 5Dand Fig. 5E). The differential effect of heparin on two mem-bers of the same family, ErbB2 and EGFR, suggests that hep-arin may interact with multiple ligands at the cell surfacewith differential effects on downstream signaling pathways.

Ext12/2 mES Cells Have Reduced pERK1/2 Levelsand This Can Be Partially Rescued by HeparinAddition

To determine if the block in Ext1�/� neural differentiationwas indeed due to lack of FGF-stimulated Erk1/2 signaling,we examined 46C (HS competent) and Ext1�/� ES cells (6heparin) to assay their steady-state Erk1/2 signaling duringN2B27-mediated neural specification. Progression was moni-tored by Sox1-GFP upregulation (Fig. 6A). We performedquantitative assessment of dP-Erk1/2 using fluorescent West-ern analysis with LiCor Odyssey software. The relativedPErk1/2 level was calculated as the sum of the ratios ofdPErk1/Erk1 and dPErk2/Erk2 (Fig. 6B). In ES cell mediacontaining 10% serum, 46C but not Ext1�/� has high dPErk1/2 levels, indicating that one consequence of loss of HS is thedepleted Erk signaling in multiple contexts. 46C cells displaythe expected pattern of dPErk1/2, initial reduction as the cellsare plated into serum-free N2B27 and increasing dPErk1/2between days 3 and 4, concomitant with rapid increase in theproportion of Sox1-GFPþ cells (Fig. 6A, 6B, red boxes).Ext1�/� cells have extremely low levels of dPErk1/2 up to day3, and although there is an increase between days 3 and 4, it isless than 46C and, as we have demonstrated before, insuffi-cient to allow neural specification [25]. Interestingly, in theExt1�/� culture treated with heparin, the peak of dPErk1/2 atday 4 is higher than both Ext1�/� and 46C control. These find-ings are in good agreement with Stavridis et al. [28] statingthat a defined window of Erk activation is necessary to induceSox1-expression (increasing dPErk1/2 is correlating withSox1-switch on), but it also implies that this is not the onlyimportant factor because the dPErk1/2 peak seen in the Ext1�/�

cells is not sufficient to allow progression to Sox1þ NPCs.


www.StemCells.com

Figure 4. Heparan sulfate/heparin regulates a fibroblast growth factor (FGF)-dependent step in neural specification that causes Ext1�/� cells toarrest at the primed state. (A): 46C were differentiated in N2B27 either supplemented with DMSO (n), FGF receptor inhibitor PD173074 (~),degree of polymerization (dp) 12 heparin 3 lg/ml (n), and PD173074 þ dp12 heparin 3 lg/ml (n). Sox1-GFP was monitored by flow cytometryand triplicate samples were analyzed. (B): Quantitative expression analysis of pluripotency and early lineage markers during N2B27 differentia-tion. Abbreviations: DMSO, dimethyl sulfoxide; ES cells, embryonic stem cells; GFP, green fluorescent protein.


DISCUSSION

As an essential part of this study, we generated the 46Cext1R-NAi cell line to confirm the observed block in neural specifi-cation in Ext1�/� cells, which was indeed the result of loss ofendogenous HS production. Using RNAi to inhibit glycosyla-tion (the result of a multienzyme biosynthetic pathway) is not

facile and is only useful when combined with careful analysesof the products generated from the RNAi-kd cells. Purificationof HS chains from 46C and 46Cext1RNAi:5 revealed an 80%reduction in the amount of HS produced by the RNAi cellline. In addition, HS chains that were produced were signifi-cantly reduced in size compared with 46C, and the sulfationprofile was markedly altered. We compared 46Cext1RNAi:5against control preparations from independent wild-type ES

Figure 5. Receptor tyrosine kinase (RTK) array analysis. (A): A total protein lysate of 250 lg from each sample was used, and arrays wereexposed identically to detection reagents and x-ray film. Pixel density analysis phosphorylated RTKs in 46C versus Ext1�/� ES cells (A, D) and46C versus Ext1�/� 6 heparin at day 4 (B, C, E). Abbreviations: EGFR, epidermal growth factor receptor; ES cells, embryonic stem cells.


www.StemCells.com

cell lines 46C and E14 to confirm that the increased 6-O-sul-fation of material produced by 46Cext1RNAi:5 is outside thevariation in HS profiles typically obtained from wild-type EScells. More minor increases in N-sulfation and 2-O-sulfationwere also in evidence. The increase in 6-O-sulfation suggestspossible altered activities of 6-O-sulfotransferases or the cell-membrane located sulfatase enzymes in the 46Cext1RNAi:5cell line. In related analyses, we have also characterized HSfrom Ext1þ/� ES cells (generated alongside Ext1�/� cellsused in this study), which also display shorter chain lengthand alterations in disaccharide composition (our unpublisheddata), distinct from the sulfation changes described in thisstudy. 46Cext1RNAi:5 and Ext1þ/� ES cells are likely to havedifferent levels of Ext1 enzyme activity, and the above find-ings correlate well with those demonstrating that altered lev-els of Ext1 enzyme influenced levels of N-deacetylase N-sul-fotransferase 1 activity and subsequent HS structure,including increased O-sulfation [36]. We hope that this datawill help refine the GAGosome theory, in which the GAGbiosynthetic enzymes exist in a physical complex in the Golgiwith inter-related activities [36, 37].

46Cext1RNAi:5 ES cells have greatly reduced ability toswitch on Sox1 during neural differentiation. This indicatesthat although the HS produced by 46Cext1RNAi:5 hasincreased 6-O-sulfation, the reduced length and overall reduc-tion of 80% of HS have resulted in a loss of commitment toneural specification. This raises the interesting question of a

possible threshold level of endogenous HS production neces-sary for successful lineage progression; in fact, although mice(and indeed, humans) lacking one copy of either EXT1 orEXT2 have phenotypic abnormalities as adults, they survivedevelopment [14, 15].

The Ext1�/� cells in this study were generated by conven-tional knockout [14]. Other researchers have used alternativesources of HS deficient ES cells, for example, from Ext1flox/flox

blastocysts [38] and by RNAi targeting of EXT1 in wild-typemES cells [35]. The phenotypes of these cells, primarily theirability to maintain self-renewal, differ significantly [35, 38].Both the knockout Ext1�/� ES cell lines (described previously[25, 34] and here), the stably transfected RNAi line (this study)and the Ext1flox/flox ES cell line [38], can self-renew and main-tain pluripotency in contrast to previous reports of Ext1kd cellsgenerated by transient RNA interference [35]. The loss ofendogenous HS is not deleterious to self-renewal and pluripo-tency. However, HS is critical when cells are challenged to dif-ferentiate by at least three distinct protocols, in vitro neural dif-ferentiation, embryoid-body mediated hematopoiesis, andleukemia inhibitory factor (LIF) withdrawal [25, 34, 38]. Oneof the hallmarks of early differentiation is the response to auto-crine signals, many of which are dependent on HS, for exam-ple, the FGF-mediated system described in this study. Indeed,our previous study demonstrated that wild-type ES cells makea low sulfated species of HS [25] in comparison with manyother embryonic tissues [39]. This is in good agreement withthe ‘‘ground state’’ hypothesis in which naı̈ve pluripotency ismaintained by protection from Erk signaling [33]. We foundthat the 46Cext1RNAi:5 ES cell line (and other clones) grewstably for many passages without obvious spontaneousdifferentiation in traditional LIF-containing ES cell media,appearing morphologically very similar to Ext1�/� mES cells.Anecdotally, we can report a much lower incidence of sponta-neously differentiated colonies in Ext1�/� ES cell culturescompared with many other wild-type lines. We suggest thatcontrasting results in Ext1kd ES cell lines perhaps reflects thedifference between a stable and transient transfection approachto generating the knockdown, potentially leaving sufficient re-sidual HS on ES cells to disrupt the balance between prodiffer-entiation and antidifferentiation signals.

ES cells are of great interest as a source of cell-basedtherapies; however, the generation of sufficient numbers ofdifferentiated progeny using affordable, nonanimal derivedproducts is a significant block to their application. To date,little consideration has been given to GAG oligosaccharidesas potential in vitro modulators of cell fate decisions. Wedemonstrate that heparin addition to ES cells enhances theyield of Sox1þ NPCs, even in a HS-competent background.We show minimum size, sulfation, and concentration require-ments for this effect and that suboptimal GAG species (K5polysaccharide, N-desulfated heparin, or heparin dp6) can infact delay the onset of Sox1 expression. In this study, theoptimal GAG for support of neural commitment was heparindp12; conversely, in a parallel study, in vitro differentiationto hemangioblast and subsequent blood lineages has a distinctrequirement for longer heparin oligosaccharides (>dp24) andat concentrations that were found to be inhibitory for neuraldifferentiation [34]. This likely reflects the different regula-tory factors involved in the formation of separate lineages.Sox1 expression initially requires upregulation of FGF-medi-ated signals [27, 28, 30, 40]; whereas early hematopoiesisrequires signaling via BMP4 and VEGF [41, 42]. This demon-strates both how versatile oligosaccharides could be as cultureadditives, but also highlights the need for careful selectionwith a thorough knowledge of the specific pathways involvedand the GAG structures needed to regulate them.

Figure 6. Comparison of wild-type and Ext1�/� embroyonic stemcell degree of polymerization (dP)-Erk1/2 activation. (A): Sox1-GFPexpression was monitored in 46Cs by flow cytometry. (B): Quantita-tion of (dP) Erk1/2 levels on cells sampled daily using fluorescentdetection and LiCor Odyssey software. Abbreviations: GFP, green flu-orescent protein; ES cells, embryonic stem cells.


We have shown that Ext1�/� mES cells cannot differenti-ate to Sox1þ NPCs [25]. We demonstrate here that Ext1�/�

cells are intrinsically able to upregulate Fgf-5 and transit toprimitive ectoderm, thus they resemble wild-type ES cellsthat are subjected to Erk1/2 signaling inhibition [28]. More-over, it has been demonstrated that ES cells can fluctuatebetween self-renewal and a primitive ectoderm-like state,characterized by Rex-1�Fgf-5þ, yet still expresses Oct-4/Sox2master regulators and will (under the influence of Nanog)revert to pluripotency [27]. The transcription profile of Ext1�/�

cells in N2B27 is consistent with a cell population in this stateof flux, yet unable to progress beyond it since Rex-1 remainshigh. Ext1�/� cells upregulate Fgf-5; however, they cannotgenerate signals in response to this without the addition of HS/heparin, which will allow the ‘‘push’’ out of pluripotency andinto neural specification.

Our analysis of dP-Erk1/2 in this system demonstrates thatin ES cell culture, 46C have significantly higher dP-Erk1/2 thanExt1�/�. This correlates well with reports of Ext1-deficientfibroblasts generated by the genetrap method, which contain re-sidual amounts of HS [43], where dPErk1/2 in response toFGF-2 stimulation was markedly reduced. During differentia-tion, Ext1�/� have a lower level of dP-Erk1/2 compared with46C, which is not sufficient to allow neural specification in theExt1�/� line [25]. Addition of heparin results in a higher dP-Erk1/2 level at day 4 than in 46C, indicative of heparin alsomodulating non-FGF mediated signals directed through Erk1/2.

It is important to note that, in a previous study, when wedifferentiated Ext1�/� þ heparin and analyzed b3-tubulinþ

neuronal cells, b3-tubulinþ cells were present but were notorganized into the extended networks seen in wild-type cul-tures [25]. Therefore, we hypothesize that heparin addition(dp24 at 1 lg/ml or dp12 at 3 lg/ml) restores the competenceto differentiate, but both timing and efficiency of differentia-tion are decreased when compared with control cells.

We applied a broad assay to simultaneously report on thephosphorylation of multiple RTKs in both wild-type and HS-deficient ES cells when compared with day 4 of N2B27 cul-ture. First, the pRTK profile of the ES cell lines is broadlysimilar with the exception of loss of pMSPR in Ext1�/� andlower levels of pErbB2. The 46C profile at day 4 displays anincrease in phosphorylation of many of the pRTKs detectedin pluripotency, along with the additional appearance of dif-ferentiation-specific pRTKs, for example, TrkC. The Ext1�/�

at day 4 conversely have failed to increase phosphorlyation ofmany of the RTKs much beyond the levels in ES cells, withpPDGFRa and pMSPR no longer detectable and no differen-tiation-associated pRTKs being detected. pPDGFRa andpMSPR can be, to a greater and lesser extent, recovered bythe addition of heparin. It is possible that serum factors in EScell media are responsible for the low levels of pPDGFRaand pMSPR seen in the Ext1�/� ES cell profile; in contrast,N2B27 media is serum-free. The subtle changes in the Ext1�/�

day 4 profile are consistent with a population of cells whichcannot progress beyond the early stages of exiting the groundstate, and this correlates with the transcription factor expression

profile of the cells (Fig. 4). The inclusion of heparin to Ext1�/�

cultures highlights that some pathways are HS-dependent,for example, ErbB2 and PDGFRa, and some are not, forexample, EGFR and EphB4. This supports the hypothesisthat differentiation of ES cells to any lineage requires a bal-ance of multiple signaling pathways to promote certain cellfates (e.g., neural) and critically, defines HS as a key ele-ment regulating these balancing acts. PDGFRa is involved incell growth, motility, and embryonic development [44, 45].EGFR and ErbB2 are involved in many normal developmen-tal processes but have critical roles in tumorigenesis, whereupregulation correlates with poor prognosis in many cancers[46, 47]. Levels of pEGFR and pErbB2 were significantlyreduced in Ext1�/� cells supporting the targeting of HS inantitumor therapy.

The Ext1�/� cell line provides a useful model system todissect the kinetics of the transition from naı̈ve ground stateto ‘‘primed’’ state. The transcription profile and phospho-RTKanalysis suggest a slower exit to primed state than wild-typecells (Rex1 still high at day 4), which implies many morepathways are defective in this cell line than FGF/Erk signal-ing, and not all of them are recoverable by the addition ofheparin, for example, EGFR and EphB4.

CONCLUSION

Therefore, in conclusion, we have used a combination ofstrategies to demonstrate that HS is required for efficient pro-gression of ES cells from self-renewal to primitive ectodermand to subsequently enable further progression to neural com-mitment. This regulation by GAGs can be exploited by addi-tion of selected oligosaccharides to differentiating cultures todrive the formation of therapeutically relevant neural celltypes. The underlying mechanisms include the known role ofHS as a positive regulator of signaling via FGF family mem-bers but importantly also include positive and negative regula-tion of multiple signaling pathways.

ACKNOWLEDGMENTS

We thank John Gallagher for critical reading of the manuscript.This work was supported by the Medical Research Council (toC.E.P. and C.L.R.M.), the Human Frontier Science Program (toR.J.H. and C.L.R.M.), and the Biotechnology and BiologicalSciences Research Council (to C.M.W. and C.L.R.M.).

DISCLOSURE OF POTENTIAL CONFLICTS OF

INTEREST

The authors indicate no potential conflicts of interest.

REFERENCES

1 Evans MJ, Kaufman MH. Establishment in culture of pluripotentialcells from mouse embryos. Nature 1981;292:154–156.

2 Smith AG. Embryo-derived stem cells: Of mice and men. Annu RevCell Dev Biol 2001;17:435–462.

3 Esko JD, Lindahl U. Molecular diversity of heparan sulfate. J ClinInvest 2001;108:169–173.

4 Ai X, Do AT, Kusche-Gullberg M et al. Substrate specificity and do-main functions of extracellular heparan sulfate 6-O-endosulfatases,QSulf1 and QSulf2. J Biol Chem 2006;281:4969–4676.

5 Murphy KJ, Merry CL, Lyon M et al. A new model for the domainstructure of heparan sulfate based on the novel specificity of K5 lyase.J Biol Chem 2004;279:27239–27245.

6 Ford-Perriss M, Guimond SE, Greferath U et al. Variant heparan sul-fates synthesized in developing mouse brain differentially regulateFGF signaling. Glycobiology 2002;12:721–727.

7 Habuchi H, Tanaka M, Habuchi O et al. The occurrence of three iso-forms of heparan sulfate 6-O-sulfotransferase having different


www.StemCells.com

specificities for hexuronic acid adjacent to the targeted N-sulfoglucos-amine. J Biol Chem 2000;275:2859–2868.

8 Lanner F, Lee KL, Sohl M et al. Heparan sulfation dependent FGFsignalling maintains embryonic stem cells primed for differentiation ina heterogeneous state. Stem Cells 2009;28:191–200.

9 Sugaya N, Habuchi H, Nagai N et al. 6-O-sulfation of heparan sulfatedifferentially regulates various fibroblast growth factor-dependent sig-nalings in culture. J Biol Chem 2008;283:10366–10376.

10 Forsberg E, Pejler G, Ringvall M et al. Abnormal mast cells in micedeficient in a heparin-synthesizing enzyme. Nature 1999;400:773–776.

11 Ringvall M, Ledin J, Holmborn K et al. Defective heparan sulfate bio-synthesis and neonatal lethality in mice lacking N-deacetylase/N-sulfo-transferase-1. J Biol Chem 2000;275:25926–25930.

12 Bullock SL, Fletcher JM, Beddington RS et al. Renal agenesis inmice homozygous for a gene trap mutation in the gene encoding hepa-ran sulfate 2-sulfotransferase. Genes Dev 1998;12:1894–1906.

13 Habuchi H, Nagai N, Sugaya N et al. Mice deficient in heparan sulfate6-O-sulfotransferase-1 exhibit defective heparan sulfate biosynthesis,abnormal placentation, and late embryonic lethality. J Biol Chem2007;282:15578–15588.

14 Lin X, Wei G, Shi Z et al. Disruption of gastrulation and heparan sulfatebiosynthesis in EXT1-deficient mice. Dev Biol 2000;224:299–311.

15 Stickens D, Zak BM, Rougier N et al. Mice deficient in Ext2 lackheparan sulfate and develop exostoses. Development 2005;132:5055–5068.

16 Rapraeger AC, Krufka A, Olwin BB. Requirement of heparan sulfatefor bFGF-mediated fibroblast growth and myoblast differentiation. Sci-ence 1991;252:1705–1708.

17 Yayon A, Klagsbrun M, Esko JD et al. Cell surface, heparin-like mol-ecules are required for binding of basic fibroblast growth factor to itshigh affinity receptor. Cell 1991;64:841–848.

18 Nurcombe V, Ford MD, Wildschut JA et al. Developmental regulationof neural response to FGF-1 and FGF-2 by heparan sulfate proteogly-can. Science 1993;260:103–106.

19 Brickman YG, Ford MD, Gallagher JT et al. Structural modificationof fibroblast growth factor-binding heparan sulfate at a determinativestage of neural development. J Biol Chem 1998;273:4350–4359.

20 Jiang R, Kato M, Bernfield M et al. Expression of syndecan-1 changesduring the differentiation of visceral and parietal endoderm from mu-rine F9 teratocarcinoma cells. Differentiation 1995;59:225–233.

21 Gallagher JT. The interaction and regulation of basic and acidic fibro-blast growth factors by heparan sulphate. Trends Glycosci Glycotech-nol 1998;10:137–144.

22 Allen BL, Rapraeger AC. Spatial and temporal expression of heparansulfate in mouse development regulates FGF and FGF receptor assem-bly. J Cell Biol 2003;163:637–648.

23 Koziel L, Kunath M, Kelly OG et al. Ext1-dependent heparan sulfateregulates the range of Ihh signaling during endochondral ossification.Dev Cell 2004;6:801–813.

24 Baldwin RJ, ten Dam GB, van Kuppevelt TH et al. A developmen-tally regulated heparan sulfate epitope defines a subpopulation withincreased blood potential during mesodermal differentiation. StemCells 2008;26:3108–3118.

25 Johnson CE, Crawford BE, Stavridis M et al. Essential alterations ofheparan sulfate during the differentiation of embryonic stem cells toSox1-enhanced green fluorescent protein-expressing neural progenitorcells. Stem Cells 2007;25:1913–1923.

26 Kunath T, Saba-El-Leil MK, Almousailleakh M et al. FGF stimulationof the Erk1/2 signalling cascade triggers transition of pluripotent em-bryonic stem cells from self-renewal to lineage commitment. Develop-ment 2007;134:2895–2902.

27 Stavridis MP, Collins BJ, Storey KG. Retinoic acid orchestrates fibro-blast growth factor signalling to drive embryonic stem cell differentia-tion. Development 2010;137:881–890.

28 Stavridis MP, Lunn JS, Collins BJ et al. A discrete period of FGF-induced Erk1/2 signalling is required for vertebrate neural specifica-tion. Development 2007;134:2889–2894.

29 Liu R, Xu Y, Chen M et al. Chemoenzymatic design of heparan sul-fate oligosaccharides. J Biol Chem 2010;285:34240–34249.

30 Ying QL, Stavridis M, Griffiths D et al. Conversion of embryonicstem cells into neuroectodermal precursors in adherent monoculture.Nat Biotechnol 2003;21:183–186.

31 Aubert J, Stavridis MP, Tweedie S et al. Screening for mammalianneural genes via fluorescence-activated cell sorter purification of neu-ral precursors from Sox1-gfp knock-in mice. Proc Natl Acad Sci2003;100(suppl 1):11836–11841.

32 Wasteson A. A method for the determination of the molecular weightand molecular-weight distribution of chondroitin sulphate. J Chroma-togr 1971;59:87–97.

33 Ying QL, Wray J, Nichols J et al. The ground state of embryonicstem cell self-renewal. Nature 2008;453:519–523.

34 Holley RJ, Pickford CE, Rushton G et al. Influencing hematopoieticdifferentiation of mouse embryonic stem cells using soluble heparinand heparan sulfate saccharides. J Biol Chem 2010 (in press).

35 Sasaki N, Okishio K, Ui-Tei K et al. Heparan sulfate regulates self-renewal and pluripotency of embryonic stem cells. J Biol Chem 2008;283:3594–3606.

36 Presto J, Thuveson M, Carlsson P et al. Heparan sulfate biosynthesisenzymes EXT1 and EXT2 affect NDST1 expression and heparan sul-fate sulfation. Proc Natl Acad Sci USA 2008;105:4751–4756.

37 Pinhal MA, Smith B, Olson S et al. Enzyme interactions in heparansulfate biosynthesis: Uronosyl 5-epimerase and 2-O-sulfotransferaseinteract in vivo. Proc Natl Acad Sci USA 2001;98:12984–12989.

38 Kraushaar DC, Yamaguchi Y, Wang L. Heparan sulfate is requiredfor embryonic stem cells to exit from self-renewal. J Biol Chem 2010;285:5907–5916.

39 Holmborn K, Ledin J, Smeds E et al. Heparan sulfate synthesized bymouse embryonic stem cells deficient in NDST1 and NDST2 is 6-O-sulfated but contains no N-sulfate groups. J Biol Chem 2004;279:42355–42358.

40 Sterneckert J, Stehling M, Bernemann C et al. Neural induction inter-mediates exhibit distinct roles of FGF signaling. Stem Cells 2010;28:1772–1781.

41 Conlon FL, Lyons KM, Takaesu N et al. A primary requirement fornodal in the formation and maintenance of the primitive streak in themouse. Development 1994;120:1919–1928.

42 Gadue P, Huber TL, Paddison PJ et al. Wnt and TGF-beta signalingare required for the induction of an in vitro model of primitive streakformation using embryonic stem cells. Proc Natl Acad Sci USA 2006;103:16806–16811.

43 Yamada S, Busse M, Ueno M et al. Embryonic fibroblasts with agene trap mutation in Ext1 produce short heparan sulfate chains. JBiol Chem 2004;279:32134–32141.

44 Ball SG, Shuttleworth CA, Kielty CM. Platelet-derived growth factorreceptors regulate mesenchymal stem cell fate: Implications for neo-vascularization. Expert Opin Biol Ther 2010;10:57–71.

45 Heldin CH, Westermark B. Mechanism of action and in vivo role ofplatelet-derived growth factor. Physiol Rev 1999;79:1283–1316.

46 Yarden Y. Biology of HER2 and its importance in breast cancer. On-cology 2001;61(suppl 2):1–13.

47 Barros FF, Powe DG, Ellis IO et al. Understanding the HER family inbreast cancer: interaction with ligands, dimerization and treatments.Histopathology. 2010;56:560–572.

See www.StemCells.com for supporting information available online.


specific glycosaminoglycans modulate neural specification of mouse embryonic stem cells

Documents