www.elsevier.com/locate/ybbrc

Biochemical and Biophysical Research Communications 330 (2005) 934–942

BBRC

Noggin and bFGF cooperate to maintain the pluripotency of humanembryonic stem cells in the absence of feeder layers

Guangwen Wang a,1, Hong Zhang a,1, Yang Zhao a, Jian Li a, Jun Cai a,Peigang Wang a, Sha Meng a, Jingbo Feng b, Chenglin Miao a,

Mingxiao Ding a, Dongsheng Li b,*, Hongkui Deng a,*

a Department of Cell Biology and Genetics, College of Life Sciences, Peking University, Beijing, Chinab Regenerative Medicine Research Institute, Yunyang Medical College Tai-he Hospital, 29 S. Renmin Rd, Shiyan, Hubei, China

Received 19 February 2005Available online 18 March 2005

Abstract

Human embryonic stem (hES) cells are typically maintained on mouse embryonic fibroblast (MEF) feeders or with MEF-con-ditioned medium. However, these xenosupport systems greatly limit the therapeutic applications of hES cells because of the risk ofcross-transfer of animal pathogens. Here we showed that the bone morphogenetic protein antagonist noggin is critical in preventingdifferentiation of hES cells in culture. Furthermore, we found that the combination of noggin and basic fibroblast growth factor(bFGF) was sufficient to maintain the prolonged growth of hES cells while retaining all hES cell features. Since both noggin andbFGF are expressed in MEF, our findings suggest that they may be important factors secreted by MEF for maintaining undiffer-entiated pluripotent hES cells. Our data provide new insight into the mechanism how hES cell self-renewal is regulated. The newlydeveloped feeder-free culture system will provide a more reliable alternative for future therapeutic applications of hES cells.� 2005 Elsevier Inc. All rights reserved.

Keywords: Human embryonic stem cell; Self-renewal; Pluripotency; BMP antagonist; Noggin; bFGF; Feeder-free culture

Human embryonic stem (hES) cells derived from theinner cell mass of human blastocysts are capable ofboth self-renewal and maintenance of pluripotency todifferentiate [1–3]. The hES cells may thus provide anunlimited supply of any cell type for transplantationtherapy, drug screening, toxicology studies of variousagents, as well as functional genomic applications[2,4]. Similar to mouse ES (mES) cells, hES cells aretypically maintained in culture in an undifferentiatedstate on inactivated mouse embryonic fibroblast(MEF) feeder layers [1] or on matrigel with MEF-con-ditioned medium (CM) [4]. However, one caveat of

0006-291X/$ - see front matter � 2005 Elsevier Inc. All rights reserved.

doi:10.1016/j.bbrc.2005.03.058

* Corresponding authors. Fax: +86 10 6275 6474 (H. Deng).E-mail addresses: dsli@syth.sina.net (D. Li), hongkui_deng@pku.

edu.cn (H. Deng).1 These authors contributed equally to this work.

using MEF feeder or MEF-CM for culture is that theseprocedures are unsafe for the clinical use of hES cellsand their derivatives because of the risk of contamina-tion by infectious agents from animals. Recently, sev-eral groups have demonstrated that hES cells can bemaintained over a prolonged period without any obvi-ous differentiation on human feeders [5–7]. However,better defined feeder-free culture conditions withoutconditioned medium are required for controlled deriva-tion, maintenance, and scaled-up production of hEScells. Thus, the identification of the MEF-derived fac-tors essential for hES cell self-renewal and the develop-ment of an animal-free hES cell culture system areextremely important issues for eventual therapeuticapplication of hES cells.

Unlike mES cells, hES cells cannot be maintained inthe undifferentiated state by replacing the feeder layer

G. Wang et al. / Biochemical and Biophysical Research Communications 330 (2005) 934–942 935

with leukemia inhibitory factor [1,2,8]. And it has beenreported that the STAT3 activation is not sufficient toblock hES cell differentiation [9,10], indicating thatmES and hES cells require different signals to maintainpluripotency. Although much progress has recently beenmade in hES cell research, the factors generated byMEF that are crucial for the self-renewal of hES cells re-main unclear. Lim et al. [11] have carried out a prote-ome analysis of conditioned medium from mouseembryonic fibroblast feeder layers and identified 136 un-ique protein species which are required for further func-tional characterization with hES cell culture. Studiesabout signaling pathways controlling the self-renewalof hES cells can provide important clues for identifyingexternal factors essential to hES cells maintenance.Through transcriptional profiling of undifferentiatedhES cells, evidence has emerged that three pathwaysFGF, Wnt, and TGFb may have prevalent roles in theself-renewal of hES cells [12–15]. bFGF is a ligand inFGF signaling pathway, and previous studies have sug-gested that exogenous bFGF is required for derivationand continued self-renewal of hES cells [16]. The maincomponents of the canonical Wnt pathway are ex-pressed in undifferentiated hES cells [12–15], and it hasbeen shown that the GSK-3-specific inhibitor, BIO,can maintain the self-renewal of hES cells [17]. However,most Wnt ligands cannot be detected in undifferentiatedhES cells [12,17], raising the possibility that Wnt pro-teins secreted from MEF may activate Wnt signaling.TGFb/activin/nodal is a branch of the TGFbsuper-family. Nodal as well as the inhibitors of Nodalsignaling, lefties, were found to be enriched in the undif-ferentiated hES cells [12–14,18–20]. Vallier et al. [19]have shown that nodal could inhibit progression alongthe neuroectoderm pathway of hES cells while promot-ing extraembryonic visceral endoderm differentiationand maintaining the expression of pluripotency markers.Amit et al. [21] have shown TGFb1, combined with LIFand bFGF, could support prolonged undifferentiatedculture of hES cells. Recently, James et al. [22] demon-strated that the TGFb/activin/nodal branch was acti-vated in undifferentiated hES cells and was necessaryfor the maintenance of pluripotency of hES cells. BMPsignaling is another branch of TGFb super-family thathas been studied in hES cell research [18,23,24]. Previousstudies showed that active BMP signaling promotes dif-ferentiation of hES cells into trophoblast [23] or extra-embryonic endoderm cells [24], while intriguingly,BMP receptors and ligands could be detected in somehES cell lines [18,24,25]. These observations suggest thatthe inhibition of BMP signaling may be required formaintaining hES cells in the undifferentiated state.

In this study, we demonstrated that the BMP antag-onist noggin played an important role in suppressingthe differentiation of hES cells. Stable expression ofexogenous noggin in NIH/3T3 cells allows the condi-

tioned medium from this cell line to support the undif-ferentiated growth of hES cells. Furthermore, we havedeveloped a feeder-free hES cell culture medium anddemonstrated that the combination of noggin andbFGF was sufficient to maintain the prolonged undiffer-entiated growth of hES cells with normal karyotype andpluripotency to differentiate into derivatives of all threeprimitive germ layers. Our findings and this novel cul-ture system should be valuable for future study of themechanism involved in hES cell self-renewal, and shouldbe useful for ultimately developing an animal-free cul-ture medium to generate large numbers of hES cellsfor therapeutic application.

Materials and methods

Human ES cell cultures. Human ES cell line H1 was obtained fromWiCell Research Institute (Madison, WI), and the passage number ofH1 cells used in this paper was between 41 and 56. Regularly, hES cellswere maintained on irradiated MEFs, which were derived from ICRmouse embryos (E13.5), in the hES medium (also termed as non-conditioned medium): DMEM/F12 medium supplemented with 20%knockout serum replacement, 1 mM L-glutamine, 1% non-essentialamino acids, 0.1 mM b-mercaptoethanol, and 4 ng/ml basic fibroblastgrowth factor (bFGF) (all from Invitrogen/Gibco). The feeder-freeculture on matrigel (BD Biosciences) with conditioned medium wascarried out as described [4]. The MEF and NIH/3T3 cells were inac-tivated with mitomycin C treatment or irradiation and seeded at�5 · 104/cm2 to generate the conditioned medium for human ES cellcultures.

The establishment of noggin-expressing NIH/3T3 cell line. Thenoggin cDNA was cloned from MEF cells and ligated into a pMX-Neo retroviral vector (kindly provided by Dr. Toshio Kitamura, TheUniversity of Tokyo). The retrovirus was generated by transfection ofpMX-noggin-Neo and pCDNA-VSVg plasmids into 293gp cells usinga standard calcium-phosphate method. The viruses were then used toinfect the NIH/3T3 cells in the presence of polybrene (Sigma). After 2weeks of G418 selection, the NIH/3T3 cells constitutively expressingnoggin (NIH/3T3-noggin) were obtained.

Immunofluorescence. Immunostaining of tumor rejection antigens(TRA)-1-60, TRA-1-81, and stage-specific embryonic antigen (SSEA)-4 was carried out similarly as described [1] and detected with fluores-cein isothiocyanate (FITC)-labeled secondary antibodies. For noggin,neuro-filament heavy chain (NFH), glial fibrillary acidic protein(GFAP), cytokeratin (CK)19 or glucagon immunostaining, cells werefixed in 4% paraformaldehyde at room temperature for 15 min. Fora-fetoprotein (AFP) or muscle actin, cells were fixed in 100% methanolat room temperature for 5 min. The fixed cells were blocked for 1 h atroom temperature with 10% goat serum/0.2% Triton X-100/PBS.Antibodies against human Tra-1-60, Tra-1-81 or SSEA-4 (Chemicon),mouse noggin (R&D system), human AFP (Roche), human NFH,GFAP, CK19, glucagon or actin (Santa Cruz) were used and localizedwith FITC or tetramethylrhodamine isothiocyanate (TRITC) conju-gated antibodies (Santa Cruz).

Alkaline phosphatase detection. To detect alkaline phosphatase(AP) activity, the cells were washed with PBS three times, fixed in 4%paraformaldehyde, and then stained with BCIP/NBT (Promega) for15–20 min at room temperature.

Flow cytometric analysis of human ES cells. The human ES cellswere dissociated with trypsin/EDTA and resuspended in PBS with 10%goat serum for 15 min at 4 �C. Then the cells were incubated for 30 minat 4 �C with TRA-1-60 antibody (Chemicon) and then with a

936 G. Wang et al. / Biochemical and Biophysical Research Communications 330 (2005) 934–942

FITC-conjugated secondary antibody (Santa Cruz). Finally, the cellswere analyzed by use of a flow cytometer (MoFlo High-PerformanceCell Sorter; Dako Cytomation). Flow cytometry data were analyzedwith Summit software (Dako Cytomation).

RT-PCR analysis. Total RNA was isolated from cells using TRIzol(Invitrogen) and reverse-transcribed using MMLV reverse transcrip-tase (Promega) according to the manufacturer�s protocol. PCRamplification of different genes was performed using EXTaq poly-merase (Takara). The PCRs were denatured at 94 �C for 5 min andcycled at 94 �C for 30 s, 54–62 �C for 30 s, and 72 �C for 30 s followedby final extension at 72 �C for 10 min after 30 cycles. The primers usedin the detection are shown in supplemental information.

In vitro differentiation of hES cells. The undifferentiated hES cellcolonies were removed from the matrigel with dispase (Gibco) and re-plated onto fibronectin-coated wells. To permit differentiation, thehuman ES cells were cultured in differentiation medium for about 12–14 days. Differentiation medium contains 80% DMEM/F12, 20% fetalbovine serum (Gibco), 1 mM L-glutamine, 1% non-essential aminoacids, and 0.1 mM b-mercaptoethanol.

Karyotype analysis. For karyotype analysis, hES cells were treatedwith 0.1 lg/ml Colcemid (KaryoMax Colcemid solution, Gibco) for 2–3 h. Cells were trypsinized, resuspended in 0.075 M KCl at 37 �C for20 min, and then fixed in methanol and acetic acid (3:1). The standardG-banding technique was performed for karyotype analysis.

Results

Conditioned medium from noggin-expressing NIH/3T3

cells could support the undifferentiated growth of hEScells

To identify the potential factors secreted by MEFwhich contribute to the self-renewal of hES cells, ourstrategy was to compare the expression patterns of can-didate factors in MEF and another cell line which couldnot maintain the undifferentiated growth of hES cells.The NIH/3T3 cell is an immortalized cell line, and simi-lar to MEF, it is also derived from mouse embryonicfibroblasts. We initially determined whether NIH/3T3cells could support hES cell culture. The hES cells werecultured on matrigel containing the CM generated bymitosis-inactivated NIH/3T3. We found that after cul-turing for 4 days, the hES cells within most colonies be-gan to differentiate into a flattened, epithelium-likemorphology. They lost the expression of Tra-1-60 andSSEA-4, two surface markers characterizing undifferen-tiated hES cells, and also showed low alkaline phospha-tase (AP) activity, a property of undifferentiated hEScells (Fig. 1A). These observations indicate that NIH/3T3 cells lack the ability to maintain the undifferentiatedstate of hES cells. Thus, we chose NIH/3T3 as the cellline to compare with MEF for the differential expressionof candidate genes through RT-PCR analysis, using thehousekeeping gene b-actin as the internal control. Thegenes we have analyzed include: Wnt1, 2, 3, 3a, 4, 5a,7a, and 10b, which belong to Wnt signaling pathway;bFGF in FGF pathway; TGFb1, -2, -3 in TGFb family;and five BMP antagonists noggin, chordin, Dan, cer-berus, and gremlin. In this experiment, we used three dif-

ferent lots of MEF cells to assure the reliability of ouranalysis. According to the results, while Wnt5a, bFGF,TGFb1, -2, -3, and cerberus were expressed in bothMEF and NIH/3T3 (Fig. 1B), there was no expressionof Wnt1, Wnt3, Wnt3a, Wnt7a, chordin, and Dan ineither MEF or NIH/3T3 (data not shown). In addition,the expression of Wnt10b was restricted to NIH/3T3cells; conversely, Wnt2, Wnt4, noggin, and gremlin wereexpressed in MEF but were hardly detectable in NIH/3T3 (Fig. 1B). The differences in the expression ofBMP antagonists noggin or gremlin between MEF andNIH/3T3 were quite significant (Fig. 1B), and we foundtranscripts of the BMP ligands BMP-2, -4 as well as BMPreceptors BMPR1A, 1B and BMPR2 could be detectedin both undifferentiated and spontaneously differentiatedH1 hES cells (Fig. 1C). We therefore assumed that inhi-bition of BMP signaling by BMP antagonists secretedfrom MEF may be required for maintaining hES cellsin undifferentiated state and focused on studying theroles of BMP antagonist.

The ability of gremlin to bind and block BMP-2 and-4 has been demonstrated [26–28]. However, biochemi-cal characterization of gremlin showed that, althoughit could be secreted to the culture medium, a significantfraction of this protein tended to remain non-covalentlybound to the outer cell surface [28], thus reducing thepossibility that abundant gremlin exists in MEF-CMto inhibit BMPs. Noggin is a relatively specific inhibitorof BMP activity and has not been reported as having ac-tions independent of BMP binding, and it has been usedas a tool to block BMP functions, such as BMP-2, -4,and -7 [26,27]. These promoted us to speculate that nog-gin is the most likely BMP antagonist secreted by MEF,which plays the key role in suppressing hES differentia-tion. To evaluate this, we first performed immunostain-ing with noggin antibody to confirm the expression ofnoggin in MEF cells (Fig. 1D), then we cloned noggincDNA from MEF cells and delivered it into NIH/3T3cells through the retroviral vector pMX-Neo. After drugselection for several weeks and validating noggin expres-sion by RT-PCR analysis (data not shown), we obtaineda NIH/3T3 cell line (NIH/3T3-Nog) with a stable nog-gin expression. We subsequently cultured hES cells onmatrigel with the CM from NIH/3T3-Nog, usingMEF-CM and the CM from empty vector-transfectedNIH/3T3 as controls. As shown in Fig. 1B, hES cellscultured with the CM from NIH/3T3-Nog are very sim-ilar to those cultured with MEF-CM, sustaining highlevels of AP activity as well as high expression of Tra-1-60 and SSEA-4 (Fig. 1A), while the hES cells underthe CM from empty vector-transfected NIH/3T3showed differentiated characteristics, same as those cul-tured with the CM from normal NIH/3T3 (data notshown). These results suggest that noggin may play animportant role in suppressing the differentiation ofhES cells.

Fig. 1. Identification of noggin as a potential factor to inhibit hES cell differentiation. (A) Expression of pluripotent stem cell markers—alkalinephosphatase (AP), TRA-1-60, and SSEA-4—of ES colonies cultured with the CM from NIH/3T3 (3T3), noggin-expressing 3T3 (3T3-NOG), andMEF (MEF) cells. Most of the cells in 3T3-CM cultured colonies lost stem cell marker expression, while the cells in MEF-CM and 3T3-NOG-CMwere positive to these markers. Scale bar: 150 lm. (B) Expression analysis of Wnt, TGFb, FGF pathway components, and BMP antagonists withRT-PCR. b-Actin expression was used as internal control. (C) RT-PCR analysis for the expression of BMP-2, BMP-4, BMPR1A, BMPR1B, andBMPR2 in undifferentiated hES cells (ES) cultured with MEF-conditioned medium and spontaneously differentiated ES cells (dES) cultured withnon-conditioned medium. GAPDH expression was used as internal control. (D) Immunofluorescence staining of MEF cells with noggin antibody.Scale bar: 50 lm.

G. Wang et al. / Biochemical and Biophysical Research Communications 330 (2005) 934–942 937

Combination of noggin and bFGF could maintain

prolonged undifferentiated growth of hES cells in theabsence of feeders or conditioned medium

To further determine whether non-conditioned med-ium (termed as non-CM, same to the hES medium de-scribed under Materials and methods) supplementedwith noggin can maintain hES cells in undifferentiatedstate, we added different concentrations of recombinantnoggin protein (R&D system) into the non-CM and cul-

tured hES cells on matrigel with these noggin-containedmedia. The differentiated states of hES cells were de-tected by morphological observations, AP activity anal-ysis, and Tra-1-60 antibody staining. While most hEScells cultured with 0, 20 or 100 ng/ml noggin underwentdifferentiation after 5–6 days; a large percentage of hEScells cultured with 250, 500 or 1000 ng/ml noggin werestill maintained in an undifferentiated state, having highAP activity and Tra-1-60 expression; and amongst allthese conditions, the non-CM containing 500 ng/ml

938 G. Wang et al. / Biochemical and Biophysical Research Communications 330 (2005) 934–942

noggin showed the best hES cell-supporting ability (datanot shown). However, when hES cells were cultured forup to 10–12 days (two passages), many colonies ap-peared with a differentiated morphology even in mediumcontaining 500 ng/ml noggin. To verify their status, westained these hES cells with the antibody of Tra-1-60at the 10th day of culture and distinguished the differen-tiated cells from undifferentiated ones by flow cytometryanalysis. As shown in Fig. 2A, the percentage of undif-ferentiated cells (Tra-1-60-positive) decreased to 70%from 92% at the 5th day of culture. These results sug-gested that while noggin can suppress differentiation ina dosage dependent manner (the best functional concen-tration in our culture system is 500 ng/ml), there must beother factors required for the maintenance of undifferen-tiated hES cells in the presence of noggin.

Fig. 2. Characterization of human ES cells cultured on matrigel with recomexpression of day 5 (d5) and day 10 (d10) hES cells cultured in non-conditnoggin plus bFGF (NFM, 500 and 40 ng/ml, respectively), and MEF-cundifferentiated TRA-1-60 positive population. The proportion decreased infigures represent data from triplicate experiments with similar results. (B–E) MNFM (B) and MCM (C) after five passages (about 30 days). Colonies at highsystems, the hES cells grow similarly as compact colonies with a high nuclearhES cells cultured in NFM after five passages also retained high expression ofin MCM and NFM cultured hES cells detected with RT-PCR. GAPDH expre(D,E), 20 lm.

Currently, hES cells are commonly cultured on MEFfeeders or on matrigel in MEF-CM and 4 ng/ml exoge-nous bFGF is added to both culture conditions. Previousstudies have found that hES cells became differentiatedwithin 2 weeks after plating if the medium lacked exoge-nous bFGF, suggesting that bFGF was required for con-tinued undifferentiated proliferation for hES cells [16].This was also proved in our unpublished study aboutSU5402, a small molecular inhibitor of FGFR1. WhenSU5402 was added into undifferentiated hES cells, mostcells in the middle of colonies underwent a remarkabledifferentiation fate with enlarged flattened phenotypeand significantly reduced AP activity within 4 days (datanot shown). In the experiment described earlier in thispaper, we have shown that bFGF was expressed at sub-stantial levels in both MEF and NIH/3T3 cells (Fig. 1B),

binant noggin and bFGF. (A) Flow cytometric analysis of TRA-1-60ioned medium with noggin (NM, 500 ng/ml), bFGF (FM, 40 ng/ml),onditioned medium (MCM). Values denote the percentage of thethe culture of noggin or bFGF alone, but not in MCM and NFM. Theorphology of hES cell colonies grown on matrigel supplemented with

magnification were shown in (D, NFM) and (E, MCM). In both culturecytoplasmic ratio. (F–H) Immunofluorescence staining showed that theTRA-1-60, TRA-1-81, and SSEA-4. (I) Expression of Oct4 and Nanogssion was used as internal control. Scale bar: (B, C, and F–H), 150 lm;

G. Wang et al. / Biochemical and Biophysical Research Communications 330 (2005) 934–942 939

suggesting that, in addition to the exogenous bFGF,these fibroblasts may also produce bFGF needed byhES cells. Thus, we increased bFGF from 4 to 8, 16,40, and 80 ng/ml, respectively, in the non-CM with orwithout noggin and used them to culture hES cells onmatrigel. We found that when there was no noggin, 40or 80 ng/ml bFGF alone could maintain most hES cellcolonies in undifferentiated morphology, and couldmaintain the percentage of Tra-1-60-positive cells near90% within 5–6 days� culture (Fig. 2A); while 4, 8 or16 ng/ml bFGF showed poor ability to support hES cellculture (data not shown). However, as the culture timeextended, the hES cells cultured with 40 or 80 ng/mlbFGF alone began to differentiate: the percentage ofTra-1-60-positive cells decreased to 72% at the 10thday under 40 ng/ml bFGF (Fig. 2A) and decreased to73% at the 15th day under 80 ng/ml bFGF (data notshown). In contrast, the hES cells cultured with the com-bination of 500 ng/ml noggin and 40 ng/ml bFGF couldbe maintained in undifferentiated states very well andwere able to keep Tra-1-60-positive cells at 92% at the10th day of culture, similar to the results obtained when

Fig. 3. Karyotype analysis and in vitro differentiation of hES cells cultured wthat hES cells remained normal 46, XY karyotype after prolonged culture in nPCR analysis of the differentiated derivatives from hES cells after five-passaexpression was used as internal control. (C) Immunofluorescence stainingindicating that the prolonged hES cells can differentiate into cell types of ectoAFP, and glucagons). Scale bar: 50 lm.

cultured with MEF-CM instead (Fig. 2A). Even afterbeing cultured with this noggin and bFGF-containednon-CM for 5 passages (about 30 days), the cells didnot show obvious differentiated appearance, having fewdifferences to those cultured with MEF-CM in morphol-ogy—tightly compacted colonies composed of small cellswith a high nuclear cytoplasmic ratio and prominentnucleoli (Figs. 2B–E). More detailed immunostainingand RT-PCR analysis showed that the hES cell surfacemarkers Tra-1-60, Tra-1-81, and SSEA-4, and two ofthe most ES cell-enriched transcription factors, Nanogand Oct-4, were all highly expressed by these hES cells(Figs. 2F–I). The hES cells cultured with noggin plusbFGF or MEF-CM were passaged routinely every 5–6days, at the same ratio of 1:3, indicating they had similarpopulation doubling time. We analyzed the expression ofTra-1-60 with flow cytometry after culturing hES cellswith the above medium for 7 passage (about 40 days),and found that the percentage of Tra-1-60-positive cellskept over 90% (data not shown). Our results suggest thatmaintenance of undifferentiated hES cells could beachieved by the combination of 500 ng/ml noggin and

ith noggin and bFGF for five passages. (A) G-banding analysis showedon-conditioned medium containing noggin and bFGF (NFM). (B) RT-ge culture in MEF-conditioned medium (MCM) and NFM. GAPDHof the cells derived from hES cells after prolonged culture in NFM,derm (NFH, GFAP), mesoderm (muscle actin), and endoderm (CK19,

940 G. Wang et al. / Biochemical and Biophysical Research Communications 330 (2005) 934–942

40 ng/ml bFGF in the absence of feeder layers or MEF-CM.

The hES cells cultured with Noggin and bFGF retained

normal karyotype and pluripotency for differentiation

Unfavorable culture condition often leads to chromo-some instability in ES cells, resulting in ES cells withunbalanced chromosomes. To determine whether thenon-CM containing noggin and bFGF alter the normal-ity of hES cells, we analyzed the karyotype of our H1hES cells which had been cultured with noggin andbFGF for five passages by standard G-banding tech-niques. Our result showed that H1 cells kept a normalXY karyotype in this medium (Fig. 3A). To further eval-uate the effect of noggin and bFGF in the non-CM onhES cells, we measured the pluripotency of hES cells cul-tured under this condition. We induced the in vitro dif-ferentiation of hES cells by removing the hES cellcolonies from the matrigel and replating them ontofibronectin-coated wells. These cells were then culturedwith the differentiation medium with or without retinoicacid (RA, Sigma) for 10–14 days. The differentiatedderivatives of hES cells were analyzed by RT-PCR todetermine their final fates. We found that the markersof all three germ layers, as well as the extra embryonictrophoblast, could be detected. The markers are neu-ro-filament heavy chain (NFH) and keratin15 for ecto-derm; T gene and b-globin for mesoderm; a-fetoprotein (AFP) and amylase for endoderm; and hu-man chorionic gonadotrophin-b (hCG-b) and glial cellsmissing-1 (Gcm-1) for extraembryonic trophoblast (Fig.3B). In addition, we also performed immunostaining tomonitor the outgrowth of differentiated hES clustersand found that NFH or glial fibrillary acidic protein(GFAP)-positive neural ectoderm cells, AFP, glucagonor cytokeratin (CK) 19-positive endoderm cells andmuscle actin-positive mesoderm cells could all be ob-tained after differentiation (Fig. 3C). These results dem-onstrate that the combination of noggin and bFGF issufficient to maintain the pluripotency of hES cells with-out the MEF feeders or MEF-CM.

Discussion

In this study, we showed that the BMP antagonistnoggin contributed to preventing differentiation of hEScells (Fig. 1). When noggin cDNA was transfected intoNIH/3T3 cells, the CM from this noggin-expressingNIH/3T3 cell line could support hES cell growth in anundifferentiated state (Fig. 1A). More importantly, wedemonstrated that the hES cells grown in the non-CMsupplemented with the combination of 500 ng/ml nogginand 40 ng/ml bFGF could maintain the pluripotency todifferentiate into three germ layer derivatives and keep

normal karyotype. Previous studies have shown thatBMP4 could suppress differentiation and support self-renewal of mES cells [29,30], while in hES cell culture,BMP4 promotes differentiation of hES cells [23]. Inthe culture of mES cells, noggin, as the BMP antagonist,could promote neuronal differentiation of mES cells[31]. However, according to our results in hES cell cul-ture, noggin could suppress the differentiation of hEScells (Fig. 1), suggesting that inhibition of BMP signal-ing may be required for the pluripotency of hES cells.This is consistent with the observations that the phos-phorylated activation of BMP signaling downstreammolecules Smad1/5/8 is barely evident in undifferenti-ated hES cells, while upon differentiation, their phos-phorylation is globally increased [18,22]. These resultsunderscore the striking difference between hES andmES cells [2,8–10,22]. Interestingly, BMP inhibitionhas also been suggested as being important for theself-renewal of intestinal stem cells (ISCs). He et al.[32] have shown that BMP signaling inhibits ISCs� self-renewal by suppressing Wnt-b-catenin activity, andtreatment of noggin leads to the relocation of b-cateninand promotes the proliferation of ISC. Previous studieshave shown that Wnt signaling is active during hES cellself-renewal [12–14,17]. Therefore, it is possible that sim-ilar interaction between BMP inhibition and Wnt activa-tion may exist in the self-renewal of hES cells.

In our newly developed feeder-free culture system,bFGF is another indispensable factor, which is also ex-pressed in MEF, and only the combination of bFGFand noggin can maintain the undifferentiation of hEScells for extended culture. At the time of writing, Kimet al. [33] have suggested that bFGF maintains hES cellself-renewal by supporting stable expression of ECMmolecules through activation of the PI3K/Akt/PKApathway. Previous research on the development ofXenopus embryos had shown that there are some inter-actions between FGF and BMP signals [34]. Under cer-tain conditions, FGF and BMP signals oppose eachother, and FGF signaling can cause inhibition of BMPsignaling via phosphorylation of Smad1 [34]. This effectmeans that FGF and BMP antagonists may have similareffects on regulation of BMP pathways. However, themechanism of how bFGF and noggin play their roles,and how FGF interacts with BMP signaling pathwaysin the self-renewal process of hES cells are unclear andrequire further investigation.

To our knowledge, we provide here the first evidencesuggesting that noggin and bFGF are sufficient to main-tain the pluripotency of hES cells in the absence of fee-der layers or conditioned medium. Our findings aboutthe roles of noggin and bFGF give new insight intothe mechanism how hES cell self-renewal is regulated.Although our novel hES cell culture medium cannot ex-clude the animal products from matrigel and ‘‘serumreplacement,’’ in which Neu5Gc is included and would

G. Wang et al. / Biochemical and Biophysical Research Communications 330 (2005) 934–942 941

induce an immune reaction in vivo [35], this mediumgreatly reduces the application risk of hES cells fromanimal pathogen contamination, and is valuable for ulti-mately developing a completely defined culture systemthat is compatible with future therapeutic applicationsof hES cells.

Acknowledgments

This research was supported by Ministry of Scienceand Technology Grant (2001CB510106), Science andTechnology Plan of Beijing Municipal Government(H020220050290) and National Nature Science Founda-tion of China for Outstanding Young Scientist Award(30125022) to Hongkui Deng, and Ministry of Scienceand Technology Grant (1999053900) to Mingxiao Ding.We thank Dr. Toshio Kitamura for kindly providingthe PMX-Neo vector, Dr. Hui Zhang, Zhiwei Chen forcritical reading of the manuscript, Yan Shi, TingtingQing, Jiefang You, Yuezhou Chen, Chengyan Wang,Chunbo Teng, and other colleagues in our laboratoryfor technical assistance and advice during experiments,Liying Du for providing the technical support of flowcytometric analysis, and Guoping Zhou for karyotypeanalysis.

Appendix A. Supplementary data

Supplementary data associated with this article canbe found, in the online version, at doi:10.1016/j.bbrc.2005.03.058.

References

[1] J.A. Thomson, J. Itskovitz-Eldor, S.S. Shapiro, M.A. Waknitz,J.J. Swiergiel, V.S. Marshall, J.M. Jones, Embryonic stem celllines derived from human blastocysts, Science 282 (1998) 1145–1147.

[2] B.E. Reubinoff, M.F. Pera, C.Y. Fong, A. Trounson, A. Bongso,Embryonic stem cell lines from human blastocysts: somaticdifferentiation in vitro, Nat. Biotechnol. 18 (2000) 399–404.

[3] C.A. Cowan, I. Klimanskaya, J. McMahon, J. Atienza, J.Witmyer, J.P. Zucker, S. Wang, C.C. Morton, A.P. McMahon,D. Powers, D.A. Melton, Derivation of embryonic stem-cell linesfrom human blastocysts, N. Engl. J. Med. 350 (2004) 1353–1356.

[4] C. Xu, M.S. Inokuma, J. Denham, K. Golds, P. Kundu, J.D.Gold, M.K. Carpenter, Feeder-free growth of undifferentiatedhuman embryonic stem cells, Nat. Biotechnol. 19 (2001) 971–974.

[5] M. Richards, S. Tan, C.Y. Fong, A. Biswas, W.K. Chan, A.Bongso, Comparative evaluation of various human feeders forprolonged undifferentiated growth of human embryonic stemcells, Stem Cells 21 (2003) 546–556.

[6] M. Richards, C.Y. Fong, W.K. Chan, P.C. Wong, A. Bongso,Human feeders support prolonged undifferentiated growth ofhuman inner cell masses and embryonic stem cells, Nat. Biotech-nol. 20 (2002) 933–936.

[7] K. Miyamoto, K. Hayashi, T. Suzuki, S. Ichihara, T. Yamada,Y. Kano, T. Yamabe, Y. Ito, Human placenta feeder layerssupport undifferentiated growth of primate embryonic stem cells,Stem Cells 22 (2004) 433–440.

[8] M.K. Carpenter, E.S. Rosler, G.J. Fisk, R. Brandenberger, X.Ares, T. Miura, M. Lucero, M.S. Rao, Properties of four humanembryonic stem cell lines maintained in a feeder-free culturesystem, Dev. Dyn. 229 (2004) 243–258.

[9] R.K. Humphrey, G.M. Beattie, A.D. Lopez, N. Bucay, C.C.King, M.T. Firpo, S. Rose-John, A. Hayek, Maintenance ofpluripotency in human embryonic stem cells is STAT3 indepen-dent, Stem Cells 22 (2004) 522–530.

[10] L. Daheron, S.L. Opitz, H. Zaehres, W.M. Lensch, P.W.Andrews, J. Itskovitz-Eldor, G.Q. Daley, LIF/STAT3 signalingfails to maintain self-renewal of human embryonic stem cells,Stem Cells 22 (2004) 770–778.

[11] J.W. Lim, A. Bodnar, Proteome analysis of conditioned mediumfrom mouse embryonic fibroblast feeder layers which support thegrowth of human embryonic stem cells, Proteomics 2 (2002) 1187–1203.

[12] N. Sato, I.M. Sanjuan, M. Heke, M. Uchida, F. Naef, A.H.Brivanlou, Molecular signature of human embryonic stem cellsand its comparison with the mouse, Dev. Biol. 260 (2003) 404–413.

[13] R. Brandenberger, I. Khrebtukova, R.S. Thies, T. Miura, C.Jingli, R. Puri, T. Vasicek, J. Lebkowski, M. Rao, MPSS profilingof human embryonic stem cells, BMC Dev. Biol. 4 (2004) 10.

[14] R. Brandenberger, H. Wei, S. Zhang, S. Lei, J. Murage, G.J. Fisk,Y. Li, C. Xu, R. Fang, K. Guegler, M.S. Rao, R. Mandalam, J.Lebkowski, L.W. Stanton, Transcriptome characterization eluci-dates signaling networks that control human ES cell growth anddifferentiation, Nat. Biotechnol. 22 (2004) 707–716.

[15] C.L. Wei, T. Miura, P. Robson, S.K. Lim, X.Q. Xu, M.Y. Lee,S. Gupta, L. Stanton, Y. Luo, J. Schmitt, S. Thies, W. Wang, I.Khrebtukova, D. Zhou, E.T. Liu, Y.J. Ruan, M. Rao, B. Lim,Transcriptome profiling of human and murine ESCs identifiesdivergent paths required to maintain the stem cell state, Stem Cells23 (2005) 166–185.

[16] M. Amit, M.K. Carpenter, M.S. Inokuma, C.P. Chiu, C.P.Harris, M.A. Waknitz, J. Itskovitz-Eldor, J.A. Thomson, Clonallyderived human embryonic stem cell lines maintain pluripotencyand proliferative potential for prolonged periods of culture, Dev.Biol. 227 (2000) 271–278.

[17] N. Sato, L. Meijer, L. Skaltsounis, P. Greengard, A.H.Brivanlou, Maintenance of pluripotency in human and mouseembryonic stem cells through activation of Wnt signaling by apharmacological GSK-3-specific inhibitor, Nat. Med. 10 (2004)55–63.

[18] D. Besser, Expression of nodal, lefty-a, and lefty-B in undiffer-entiated human embryonic stem cells requires activation ofSmad2/3, J. Biol. Chem. 279 (2004) 45076–45084.

[19] L. Vallier, D. Reynolds, R.A. Pedersen, Nodal inhibits differen-tiation of human embryonic stem cells along the neuroectodermaldefault pathway, Dev. Biol. 275 (2004) 403–421.

[20] B. Bhattacharya, T. Miura, R. Brandenberger, J. Mejido, Y.Luo, A.X. Yang, B.H. Joshi, I. Ginis, R.S. Thies, M. Amit, I.Lyons, B.G. Condie, J. Itskovitz-Eldor, M.S. Rao, R.K. Puri,Gene expression in human embryonic stem cell lines: uniquemolecular signature, Blood 103 (2004) 2956–2964.

[21] M. Amit, C. Shariki, V. Margulets, J. Itskovitz-Eldor, Feederlayer- and serum-free culture of human embryonic stem cells, Biol.Reprod. 70 (2004) 837–845.

[22] D. James, A.J. Levine, D. Besser, A. Hemmati-Brivanlou,TGF{beta}/activin/nodalsignaling is necessary for the mainte-nance of pluripotency in human embryonic stem cells, Develop-ment 132 (2005) 1273–1282.

[23] R.H. Xu, X. Chen, D.S. Li, R. Li, G.C. Addicks, C. Glennon,T.P. Zwaka, J.A. Thomson, BMP4 initiates human embryonic

942 G. Wang et al. / Biochemical and Biophysical Research Communications 330 (2005) 934–942

stem cell differentiation to trophoblast, Nat. Biotechnol. 20 (2002)1261–1264.

[24] M.F. Pera, J. Andrade, S. Houssami, B. Reubinoff, A. Trounson,E.G. Stanley, D. Ward-van Oostwaard, C. Mummery, Regulationof human embryonic stem cell differentiation by BMP-2 and itsantagonist noggin, J. Cell Sci. 117 (2004) 1269–1280.

[25] X. Zeng, T. Miura, Y. Luo, B. Bhattacharya, B. Condie, J. Chen,I. Ginis, I. Lyons, J. Mejido, R.K. Puri, M.S. Rao, W.J. Freed,Properties of pluripotent human embryonic stem cells BG01 andBG02, Stem Cells 22 (2004) 292–312.

[26] W. Balemans, W. Van Hul, Extracellular regulation of BMPsignaling in vertebrates: a cocktail of modulators, Dev. Biol. 250(2002) 231–250.

[27] E. Canalis, A.N. Economides, E. Gazzerro, Bone morphogeneticproteins, their antagonists, and the skeleton, Endocr. Rev. 24(2003) 218–235.

[28] L.Z. Topol, B. Bardot, Q. Zhang, J. Resau, E. Huillard, M.Marx, G. Calothy, D.G. Blair, Biosynthesis, post-translationmodification, and functional characterization of Drm/Gremlin, J.Biol. Chem. 275 (2000) 8785–8793.

[29] Q.L. Ying, J. Nichols, I. Chambers, A. Smith, BMP induction ofId proteins suppresses differentiation and sustains embryonic stemcell self-renewal in collaboration with STAT3, Cell 115 (2003)281–292.

[30] X. Qi, T.G. Li, J. Hao, J. Hu, J. Wang, H. Simmons, S.Miura, Y. Mishina, G.Q. Zhao, BMP4 supports self-renewal ofembryonic stem cells by inhibiting mitogen-activated proteinkinase pathways, Proc. Natl. Acad. Sci. USA 101 (2004) 6027–6032.

[31] T.E. Gratsch, K.S. O�Shea, Noggin and chordin have distinctactivities in promoting lineage commitment of mouse embryonicstem (ES) cells, Dev. Biol. 245 (2002) 83–94.

[32] X.C. He, J. Zhang, W.G. Tong, O. Tawfik, J. Ross, D.H. Scoville,Q. Tian, X. Zeng, X. He, L.M. Wiedemann, Y. Mishina, L. Li,BMP signaling inhibits intestinal stem cell self-renewal throughsuppression of Wnt-beta-catenin signaling, Nat. Genet. 36 (2004)1117–1121.

[33] S.J. Kim, S.H. Cheon, S.J. Yoo, J. Kwon, J.H. Park, C.G. Kim,K. Rhee, S. You, J.Y. Lee, S.I. Roh, H.S. Yoon, Contributionof the PI3K/Akt/PKB signal pathway to maintenance of self-renewal in human embryonic stem cells, FEBS Lett. 579 (2005)534–540.

[34] E.M. De Robertis, H. Kuroda, Dorsal–ventral patterning andneural induction in Xenopus embryos, Annu. Rev. Cell Dev. Biol.20 (2004) 285–308.

[35] M.J. Martin, A. Muotri, F. Gage, A. Varki, Human embryonicstem cells express an immunogenic nonhuman sialic acid, Nat.Med. 11 (2005) 228–232.