EMBRYONIC STEM CELLS/INDUCED PLURIPOTENT STEM CELLS
A Boost of BMP4 Accelerates the Commitment of Human Embryonic
Stem Cells to the Endothelial Lineage
ORIT GOLDMAN,a,b,c,dOLIVIER FERAUD,b,eJULIE BOYER-DI PONIO,a,b,dCATHERINE DRIANCOURT,a,b,d
DENIS CLAY,d,fMARIE-CAROLINE LE BOUSSE-KERDILES,a,b,dANNELISE BENNACEUR-GRISCELLI,b,eGEORGES UZANa,b,d
aINSERM U972, Villejuif, France;bINSERM U602, Villejuif, France;cABCell-Bio SAS, Paris, France;
dUniversite ´ Paris-Sud, Institut Andre ´ Lwoff, Villejuif, France;eINSERM U935, Villejuif, France;
fIFR89, Universite ´ Paris-Sud, Institut Andre ´ Lwoff, Villejuif, France
Key Words. Human embryonic stem cells•Endothelial differentiation•BMP4•Endothelial cells
Embryoid bodies (EBs) generated during differentiation of
human embryonic stem cells (hESCs) contain vascular-like
structures, suggesting that commitment of mesoderm pro-
genitors into endothelial cells occurs spontaneously. We
showed that bone morphogenetic protein 4 (BMP4), an in-
ducer of mesoderm, accelerates the peak expression of
CD133/kinase insert domain-containing receptor (KDR)
and CD144/KDR. Because the CD1331KDR1population
could represent endothelial progenitors, we sorted them at
day 7 and cultured them in endothelial medium. These
cells were, however, unable to differentiate into endothe-
lial cells. Under standard conditions, the CD1441KDR1
population represents up to 10% of the total cells at day
12. In culture, these cells, if sorted, give rise to a homoge-
neous population with a morphology typical of endothelial
cells and express endothelial markers. These endothelial
cells derived from the day 12 sorted population were func-
tional, as assessed by different in vitro assays. When EBs
were stimulated by BMP4, the CD1441KDR1peak was
shifted to day 7. Most of these cells, however, were
CD312, becoming CD311
pressed von Willebrand factor and were functional.
This suggests that, initially, the BMP4-boosted day 7,
CD1441KDR1CD312population represents immature en-
dothelial cells that differentiate into mature endothelial
cells in culture. The expression of OCT3/4, a marker of
immaturity for hESCs decreases during EB differentia-
tion, decreasing faster following BMP4 induction. We also
show that BMP4 inhibits the global expression of GATA2
and RUNX1, two transcription factors involved in heman-
gioblast formation, at day 7 and day 12. STEM CELLS
in culture. They then ex-
Disclosure of potential conflicts of interest is found at the end of this article.
In the embryo, the formation of the vascular network is gov-
erned by two distinct processes, vasculogenesis and angiogen-
esis. Vasculogenesis occurs during early morphogenesis and
is defined as the de novo development of a vascular tree from
intrinsic endothelial cells . Angiogenesis is defined as the
development of the pre-existing vascular network by a remod-
eling process [2, 3]. Comprehension of the mechanisms
involved in vessel formation during development has numer-
ous applications, including the treatment of cancer and ische-
mic disease [4, 5]. However, the human embryo is too com-
plex to permit the study of the fine-tuning mechanisms
controlling these processes. In this context, human embryonic
stem cells (hESCs) represent a unique model for studying suc-
cessive events occurring during human development .
ESCs have the capacity to self-renew and to differentiate into
the three germ layers, mesoderm, endoderm, and ectoderm
[7–9]. Different studies have described the differentiation of
hESCs into endothelial cells using various culture conditions.
The first hESC-derived endothelial cells were isolated from
embryoid bodies (EBs) by CD31 cell sorting. These cells
expressed endothelial markers in culture and were functional
in vitro . Further studies established that this differentia-
tion occurs in early mesoderm, from a common stem cell, the
hemangioblast, which gives rise to both hematopoietic and en-
dothelial cells [11, 12]. In a recent study, Wang et al. have
shown that endothelial cells produced by hESCs are fully
functional, as assessed in vivo where they have been observed
to be incorporated into mouse vessels in a model of ischemia
. Several studies have pointed to the critical role of bone
morphogenetic protein 4 (BMP4) in the induction of both he-
matopoietic and endothelial differentiation from hESCs. The
BMPs are transforming growth factor b related. BMP4 indu-
ces ventral mesoderm, suppresses induction of dorsal meso-
derm by activin, and inhibits dorsoanterior development of
embryos, suggesting that it is a ventralizing factor .
Author contributions: O.G.: performed most experiments shown in this study, and wrote the manuscript; O.F.: codesigned experiments,
was involved in the manuscript revision; J.B.-D.P.: was associated with the experiments; C.D.: performed qPCR assays; D.C.: performed
cell sorting experiments; M.-C.L.B.-K.: was involved in the revision of the manuscript; A.B.-G.: was involved in the revision of the
manuscript; G.U.: designed the projects, obtained grants, codesigned experiments, wrote the manuscript.
Correspondence: Georges Uzan, Ph.D., INSERM U972, Ho ˆpital Paul Brousse, 12-14 Avenue Paul Vaillant Couturier, 94807 Villejuif,
France. Telephone: 33-145595267; Fax: 33-145595329; e-mail: firstname.lastname@example.org
lication April 11, 2009; first published online in STEM CELLS EXPRESS April 23, 2009. V
STEM CELLS 2009;27:1750–1759 www.StemCells.com
Received January 20, 2009; accepted for pub-
C AlphaMed Press 1066-5099/2009/$30.00/0 doi:
Treatment of hESCs with BMP4 yields different effects
according to the duration of the treatment. Long-term expo-
sure of hESCs to BMP4 results in trophoblast and extraem-
bryonic endoderm differentiation, whereas short-term treat-
ment promotes early mesoderm induction. It has been shown
that members of the BMP family regulate the proliferation
and differentiation of primitive human hematopoietic stem
cells. High levels of BMP4 increase hematopoietic stem cell
survival from hESCs, whereas low levels of BMP4 induce
differentiation and proliferation of these stem cells [15–18].
In this study, we analyzed the influence of BMP4 on en-
dothelial cell appearance. We show that a short induction
with a high-dose of BMP4 accelerates the rate of appearance
and the number of cells committed to the endothelial lineage
during the course of EB differentiation. Endothelial cells
emerging after a BMP4 ‘‘boost’’ display an immature pheno-
type, and mature into fully functional endothelial cells in
MATERIALS AND METHODS
hESCs (H9; WiCell Research Institute, Madison, WI, http://
www.wicell.org) were cultured according to the supplier’s
instructions. Cells were maintained on a mitomycin C inactivated
mouse embryonic fibroblast feeder layer. The medium—Dulbec-
co’s modified Eagle’s medium/F12 supplemented with 20%
Knock Out serum replacer, 1 mM L-glutamine, 1% penicillin/
streptomycin, 0.1 mM b-mercaptoethanol, 1% nonessential amino
acids, and 10 ng/ml recombinant human basic fibroblast growth
factor (all from Invitrogen, Carlsbad, CA, http://www.invitrogen.
com)—was changed daily.
Differentiated EB Formation
For EB formation, H9 colonies were harvested and cultured under
different conditions. We first used the human cytokine cocktail
described by Chadwick et al. . Briefly, every 4 days, 100 ng/
ml human stem cell factor (hSCF) (AbCys s.a., Paris, France,
http://www.abcysonline.com), 100 ng/ml human Flt3 ligand
(hFlt3L) (AbCys s.a.), 10 ng/ml human interleukin (hIL)-3
(AbCys s.a.), 10 ng/ml hIL-6 (AbCys s.a.), 50 ng/ml hG-CSF
(AbCys s.a.), 10 ng/ml human vascular endothelial growth factor
(hVEGF) (PromoCell Biosciences, Heidelberg, Germany, http://
www.promocell.com), and 10 ng/ml hBMP4 (AbCys s.a.) were
added to the culture medium. These conditions are hereafter
referred as ‘‘standard conditions.’’ To analyze the effect of cyto-
kines, EBs were cultured under standard conditions, with one
cytokine removed at a time (except for VEGF and BMP4). We
then analyzed the effect of a 1- or 3-day BMP4 induction using
different doses of BMP4 (10, 50, and 100 ng/ml) in the presence
of 100 ng/ml hSCF, 100 ng/ml hFlt3L, and 50 ng/ml hVEGF. Af-
ter 1 or 3 days, new medium supplemented with hSCF, hFlt3L,
and hVEGF was used. Conditions in which BMP4 at 50 ng/ml
was added for 24 hours are hereafter referred to as ‘‘boost
To block the effect of BMP4, EBs were cultivated for 3 days
with 150 ng/ml Noggin (R&D Systems Inc., Minneapolis, http://
www.rndsystems.com). At day 3, new medium supplemented
with 100 ng/ml hSCF, 100 ng/ml hFlt3L, and 50 ng/ml hVEGF
RNA Extraction and Quantitative Polymerase
At day 0, day 7, and day 12, EBs were dissociated with 1 mg/ml
collagenase IV for 2 hours at 37?C. Total RNA was extracted
using the RNeasy mini-kit (Qiagen, Hilden, Germany, http://
www1.qiagen.com) and cDNA was prepared with the cDNA
Archive kit (Applied Biosystems, Foster City, CA, http://www.
appliedbiosystems.com). Quantitative polymerase chain reaction
(PCR) was performed with TaqMan Gene Expression Assays on
an ABI7000 system (Applied Biosystems) in the presence of
10 ng of RNA. GATA2, OCT3/4, RUNX1, and TATA box bind-
ing protein (endogenous control) were investigated (references,
respectively, Hs00231119_m1, Hs00742896_s1, Hs01021970_m1,
Hs99999910_m1). The 2-[/delta][/delta]Ct method was used and
results were expressed as fold increase in expression of each
gene relative to day 0.
EBs at day 7 and day 12 were dissociated by collagenase IV
treatment. Cells were then labeled with control isotypes—mouse
IgG1-phycoerythrin (PE), IgG1-fluorescein isothiocyanate (FITC),
and IgG1-allophytocyanin (APC) (Beckman Coulter, Villepinte,
France, http://www.beckmancoulter.com) or CD133/1-PE (Milte-
nyi Biotec, Paris, France, Germany, http://www.miltenyibiotec.
com), kinase insert domain-containing receptor (KDR)-APC
(R&D Systems Inc., Minneapolis, http://www.rndsystems.com),
or CD45-PE (Beckman Coulter) antibodies. Cells were analyzed
using a fluorescence-activated cell sorting (FACS)Calibur (BD
Pharmingen, San Jose,CA,
bdbiosciences.com/index_us.shtml) with Weasel Software (WEHI,
Melbourne, Australia, http://www.wehi.edu.au).
For cell sorting, dissociated EB cells were labeled with con-
trol mouse isotypes or with anti-CD144-PE (Beckman Coulter),
anti-KDR-APC, or anti-CD31-FITC (R&D Systems). Cells were
sorted using a FACSDiva cell sorter (Becton, Dickinson and
Company, Franklin Lakes, NJ, http://www.bd.com) and plated on
48 fibronectin-coated plates (Sigma-Aldrich, St. Louis, http://
www.sigmaaldrich.com) in 80% Medium-199 (Invitrogen), 20%
fetal bovine serum, 50 lg/ml bovine pituitary extract, 10 IU/ml
heparin, 5 ng/ml VEGF, and antibiotics. After the first passage,
cells were cultured in EGM-2 (Lonza, Paris, France, http://
Cultured cells were analyzed with monoclonal antibodies
(CD144-PE, VEGFR-2-APC, CD133/1-PE, CD34-FITC, CD34-
PE, CD31-FITC, CD43-FITC, CD14-PE, or CD45-FITC) or with
control mouse isotypes.
CD144þKDRþsorted cells were fixed in 4% paraformalde-
hyde/phosphate-buffered saline (PBS). When required, cells were
permeabilized with 0.1% Triton X100 PBS for 10 minutes at
room temperature. Cells were incubated with the primary antibod-
ies stage-specific embryonic antigen (SSEA)-3, tumor rejection
antigen (TRA)1-60, and OCT3/4 (Invitrogen) and labeled with
secondary antibodies (Alexa Fluor 488 goat anti-mouse IgM,
Alexa Fluor 488 goat anti-mouse IgG, Alexa Fluor 488 goat anti-
rabbit IgG; Invitrogen).
Cells were fixed in 4% paraformaldehyde/PBS for 15 minutes at
room temperature and rinsed with PBS/0.025% Tween. For intra-
cellular staining, cells were permeabilized with 0.1% Triton
X100/PBS for 10 minutes at room temperature.
CD144þKDRþcells were incubated for 60 minutes with con-
trol isotype (mouse IgG1-Alexa Fluor 488) or with monoclonal
antibodies—anti-CD144 (Beckman Coulter), anti-CD31 (Dako,
Trappes, Francehttp://www.dako.com), anti-von Willebrand factor
(vWF) (Dako)—and labeled with secondary antibodies—Alexa
Fluor 488 goat IgG (Invitrogen), Alexa Fluor 488 goat anti-rabbit
IgG (Invitrogen). Cells were stained with 40,6-diamidino-2-phe-
nylindole (DAPI) and examined with a fluorescence microscope
(Leica DMS, Wetzlar, Germany, http://www.leica.com).
Endothelial Function Assays
Tumor Necrosis Factor a Treatment.
for 18 hours with 10 ng/ml TNF-a (R&D Systems) and incubated
Cells were incubated
Goldman, Feraud, Boyer-Di Ponio et al.
with monoclonal antibodies (intercellular adhesion molecule
[ICAM]-1; R&D Systems) or with control isotypes (mouse IgG;
Beckman Coulter) and labeled with secondary antibody: Alexa
Fluor 488 goat IgG. Cells were analyzed using a FACSCalibur
Vascular Tube Formation.
Matrigel (BD Biosciences, San Diego, http://www.bdbiosciences.
com). Cells (200,000) were plated onto Matrigel in EGM-2 with
50 ng/ml VEGF and incubated for 18 hours at 37?C. Photographs
were takenevery 2hours.
Endothelial Cell Migration.
pended in 100 ll EGM-2 without VEGF and seeded onto Costar
Transwell inserts (BD Biosciences) precoated with type I rat tail
collagen (BD Biosciences). Inserts were placed in a 12-well plate
containing 600 ll EGM-2 without VEGF. After 24 hours, 50 ng/ml
VEGF was added to the lower chamber. Cells that had migrated
to the lower side of the insert after 12 hours were fixed in 4%
paraformaldehyde/PBS and stained with DAPI. The insert, cut and
mounted in Glycergel, was examined with a fluorescence micro-
scope. Migrated cell numbers were counted in three different rep-
resentative high-power fields per insert.
Cells were trypsinized and sus-
EGM-2. A scratch was made using a pipette cone and photo-
graphs were taken before, just after, and 24 hours after the
Cells were plated onto fibronectin in
Population Doubling Assay.
sage 3, 20 days after the initial seeding of CD144/KDR sorted
cells. At confluence, cells were detached and counted. The cumu-
lative population doubling was estimated using the following
equation: (Ln(number of cells counted/number of cells at the be-
ginning of the assay)/Ln2). Population doubling of boosted day 7
and standard culture day 12 endothelial cells was compared by
analyzing the number of population doublings at day 66 (n ¼ 6).
The statistical analysis was performed using Student’s t-test.
These population doubling curves were compared with those
obtained with human umbilical vein endothelial cells (HUVECs)
seeded at passage 1.
The experiment started at pas-
Effect of Hematopoietic Cytokines and BMP4 on
The standard culture conditions described in Materials and
Methods were designed for obtaining both hematopoietic and
endothelial cells from a common progenitor, the hemangio-
blast. For this reason, a panel of hematopoietic cytokines,
including Il-3, IL-6, G-CSF, Flt3L, and SCF, was used in
addition to VEGF and BMP4 at a low concentration (10 ng/
ml). In the present study, we first analyzed if any of these he-
matopoietic cytokines was also critical for endothelial differ-
entiation and proliferation. EBs were cultured either in the
standard medium containing all the cytokines or under condi-
tions in which one cytokine was removed at a time. The per-
centage of the CD144þKDRþpopulation, which corresponds
to our definition of endothelial cells, was analyzed at day 12,
a stage at which endothelial cells were shown to be present
under these culture conditions . When EBs were cultured
without SCF or Flt3L under standard conditions, the percent-
age of CD144þKDRþcells at day 12 was more than twofold
lower (p < .005) than with the standard conditions. The other
cytokines (IL-3, IL-6, G-CSF) did not induce any change in
the percentage of CD144þKDRþcells (Fig. 1A, 1B).
To determine whether BMP4 had an effect on the appear-
ance of endothelial cells during the course of EB differentia-
tion, we added this factor on day 1 at a concentration of 10,
50, or 100 ng/ml and cultured for 24 hours or 3 days. We ana-
lyzed the effect of this BMP4 boost on the kinetics of appear-
ance of the CD144þKDRþpopulation. These conditions were
compared with the standard conditions. The CD144þKDRþ
population was analyzed by flow cytometry at different time
points during EB differentiation. When EBs were cultured for
24 hours with BMP4 at 50 or 100 ng/ml, the percentage of
CD144þKDRþcells at day 7 was 2.2-fold higher than with
the standard conditions (n ¼ 3; p < .005) and lower by half at
day 12 (n ¼ 3; p < .005). A longer duration of BMP4 induc-
tion (3 days) at the same concentration (50 or 100 ng/ml) did
CD144þKDRþcells. (A): CD144/KDR expression at day 12 in the
presence or absence of IL-3, IL-6, G-CSF, SCF, and Flt-3. (B):
CD144/KDR expression during differentiation at various times and
concentrations of BMP4. (C): CD144/KDR expression under BMP4
boost conditions in the presence or absence of Flt-3 and SCF. Abbre-
viations: BMP4, bone morphogenetic protein 4; IL, interleukin; KDR,
kinase insert domain-containing receptor; SCF, stem cell factor;
VEGF, vascular endothelial growth factor.
Effect of cytokines and BMP4 on the number of
Effects of BMP4 on hESC-Derived Endothelial Cells
not yield any variation in the percentage of the CD144þKDRþ
population at either day 7 or day 12 as compared with the
standard conditions (Fig. 1B). Moreover, treatment of EBs
with BMP4 (10 ng/ml) for 24 hours led to a decrease in the
CD144þKDRþpopulation at day 12 but not at day 7 (p <
.005). A longer duration of BMP4 treatment (10 ng/ml for 3
days) did not modify the kinetics of CD144þKDRþcells.
Finally, we show that SCF and Flt3L, the two cytokines that
have been shown to be critical for maximal CD144/KDR
expression, were also required in the BMP4 boost culture con-
ditions (BMP4, 50 ng/ml for 24 hours) (Fig. 1C).
Effect of BMP4 on the Kinetics of Expression
of Endothelial Markers
To determine if a 24-hour BMP4 boost modified the kinetics
of appearance of endothelial/progenitor cell markers, we fol-
lowed the expression of these markers, either individually
(KDR, CD31, CD144, CD34, and CD133) (Fig. 2A) or in
combination (CD133/KDR and CD144/KDR) (Fig. 2B) (n ¼
5). Expression of KDR and CD133 was detected in undiffer-
entiated hESCs. The BMP4 boost induced a significant
increase in KDRþcells at day 7. The BMP4 boost resulted in
a significantly lower expression of CD133 at day 9 and day
12, compared with noninduced cells. CD144 expression
peaked at day 7 upon BMP4 stimulation, whereas the peak in
CD144 expression was detected at day 12 under standard con-
ditions. Differences between the BMP4-induced and standard
conditions were significant at day 7 and day 12 (Fig. 2A).
When CD133 and KDR expression was measured in combina-
tion, three peaks in expression were observed: at day 2 under
standard conditions, at day 7 after BMP4 stimulation, and at
day 9 under standard conditions. At day 2, CD133/KDR
expression was significantly higher under standard condition
(p < .005), at day 7 this population was significantly higher
in BMP4-boosted cells (p < .005 ), and at day 9 expression
was significantly higher under standard conditions (p < .005 )
(Fig. 2B). CD144/KDR combined expression yielded two
peaks: the first at day 7 after BMP4 boost (p < .005) and the
second at day 12 under standard conditions (p < .005) (Fig.
2B). To make sure that the differences observed between
standard conditions and BMP4 stimulation were a result of a
direct effect of BMP4, BMP4 was blocked by adding Noggin
(an antagonist of BMP4) to the culture during the first 3 days
of EB differentiation. Noggin treatment abolished the effect
of BMP4 on CD144/KDR expression during EB differentia-
tion (Fig. 2C) (n ¼ 3). Similar results were observed with the
CD133þKDRþpopulation (data not shown).
The population of cells present in the different peaks was
sorted and analyzed in culture. Because CD133 and KDR are
expressed in undifferentiated hESCs, CD133þKDRþcells at
day 2 were not further analyzed. At day 9, CD133þKDRþ
sorted cells did not proliferate and were not analyzed further.
Finally, CD133þKDRþcells sorted at day 7 proliferated in
culture, but these cells did not express any endothelial
markers (data not shown).
ual endothelial markers (A) or a combination of endothelial markers (B). (C): Inhibition of CD144/KDR BMP4-boosted expression by Noggin
during the course of EB differentiation. Abbreviations: BMP4, bone morphogenetic protein 4; EB, embryoid body; KDR, kinase insert domain-
Effect of BMP4 on the kinetics of endothelial marker expression. Show in the effect of BMP4 induction on the expression of individ-
Goldman, Feraud, Boyer-Di Ponio et al.
Sorting and Characterization of Endothelial Cells in
Differentiated EBs at Day 12
When EBs were cultured under standard conditions, the
expression of CD144/KDR peaked at day 12 (Fig. 1 C). We
then analyzedthis population
CD144þKDRþcells represent an average of 10% of the total
cells (?2, n > 5) (Fig. 3A), and they express CD31 (97%)
and CD34 (97%), confirming their endothelial phenotype
(Fig. 3B). In contrast, these CD144þKDRþsorted cells
expressed neither CD133 nor the hematopoietic marker CD43
(Fig. 3B). Also, it was observed that CD144þKDRþcells
from EBs cultured under BMP4 boost conditions and sorted
at day 12 expressed endothelial markers, including CD31, im-
mediately after sorting (data not shown). CD144þKDRþday
12 sorted cells from EBs cultured under standard conditions
gave rise to a homogeneous population of cells with the mor-
phology of endothelial cells. At day 30 of culture, the cells
were still positive for the endothelial markers KDR and
CD144. The culture did not contain any hematopoietic cells,
as attested by the absence of expression of CD14, CD45, and
CD133 (Fig. 4A). Immunofluorescent microscopy indicated
that these cells expressed vWF in the cytoplasm and CD31
and CD144 on their membranes (Fig. 4B). These data confirm
that cells derived from CD144þKDRþday 12 EBs have the
phenotypic features of endothelial cells.
These cells were activated in response to the proinflam-
matory factor TNF-a, as shown by the upregulation of ICAM-
1 (Fig. 4C).
We compared the cumulative population doubling curve
of endothelial cells derived from CD144þKDRþday 12 EBs
to that of HUVECs, taken as a reference of mature vascular
wall endothelial cells. As shown in Figure 4D, hESC-derived
endothelial reached a plateau at day 47 (21 days after passage
3) and HUVECs reached a plateau at day 60 (n ¼ 6).
When CD144þKDRþcells were placed onto a Matrigel
film, they formed vascular-like network structures (Fig. 4E).
These endothelial cells take up diacetylated low-density lipo-
protein (data not shown).
sorted cells did not display
features of progenitor cells, because we did not observe any
colonies when they were seeded under limiting dilution condi-
tions, suggesting that these cells are mature.
Phenotypic and Functional Properties of
BMP4-Induced Endothelial Cells at Day 7
Under BMP4 boost conditions, the expression of CD144/KDR
peaked at day 7 (Fig. 1 C). The CD144þKDRþpopulation was
sorted by flow cytometry at day 7 (n > 5). Sorted cells did not
express the hematopoietic markers CD45 (Fig. 5A) and CD43
(data not shown). The majority of the cells (73%) expressed
CD34 (Fig. 5A). The CD144þKDRþpopulation contained a
major population (65%) that did not express CD31 and a minor
population (35%) that did express this endothelial marker.
To characterize the CD144þKDRþCD31?
these sorted cells were analyzed at day 21 of culture. They
had the typical morphology of endothelial cells and acquired
CD31 expression (98% of the cells) (Fig. 5A, right panel).
This suggests that the initial CD144þKDRþCD31?sorted
cells displayed an immature endothelial phenotype and that
they differentiated into mature endothelial cells in culture
(Fig. 5A). The cumulative population doubling curve of the
day 7 CD144þKDRþcells was linear until day 66 (42 days
after passage 3) (Fig. 5B). At day 66, the cumulative popula-
tion doubling obtained with these cells was significantly
CD144þKDRþpopulation before sorting. (B): Phenotype of sorted CD144þKDRþcells. Expression of CD31, CD34, CD43, and CD133 was ana-
lyzed by flow cytometry just after sorting. Abbreviations: EB, embryoid body; FSC, forward scatter; KDR, kinase insert domain-containing recep-
tor; SSC, side scatter.
Characterization of CD144þKDRþcells after sorting at day 12 of EB differentiation. (A): Flow cytometry analysis of the
Effects of BMP4 on hESC-Derived Endothelial Cells
higher (n ¼ 6; p < .005) than that obtained with the day 12
CD144þKDRþcells cultured under standard conditions (Fig.
4D). This curve was comparable with that obtained with
HUVECs(day 60).As shown
CD144þKDRþCD31?-derived endothelial cells are activated
by TNF-a, as assessed by the induction of ICAM-1 by flow
cytometry. Migration of CD144þKDRþCD31?-derived endo-
thelial cells in response to a VEGF gradient was compared
with that of HUVECs and with CD144þKDRþ-derived endo-
thelial cells cultured under standard conditions sorted at day
12 (n > 5). We found that CD144þKDRþCD31?-derived en-
dothelial cells had a migration capacity comparable with that
of HUVECs. Moreover, this migration capacity was signifi-
CD144þKDRþ-derived endothelial cells (Fig. 5D) (p < .005).
Wethen measured the
CD144þKDRþcells in an in vitro wound-healing assay. We
showed that the cell-free wound gaps healed almost com-
pletelyafter24 hours (Fig.
CD144þKDRþCD31?sorted cells were seeded under limiting
dilution conditions. These cells, like CD144þKDRþcells
sorted at day 12, were unable to generate any colonies, sug-
gesting that even if they are immature, they are not fully
competent endothelial progenitors.
Expression of Transcriptional Factors in hESCs and
During EB Differentiation
GATA2 and RUNX1/AML1 are transcription factors involved
in the emergence of hematopoietic/endothelial stem cells
(hemangioblasts) from mesodermal progenitors. We analyzed
their transcriptional level at day 7 and day 12 of EB differen-
tiation, either under standard or BMP4-boost conditions, and
compared these levels of expression with those measured at
day 0 (expressed as fold increase in expression). At day 12,
the increase in GATA2 expression was higher under standard
conditions than under BMP4-boost conditions (n ¼ 3; p <
.005). GATA2 expression was greater at day 7 than at day 0
(ninefold), but no significant difference in expression level
was observed between standard and BMP4-boost conditions
(Fig. 6A, upper left). A relative decrease in GATA2 expres-
sion was observed between day 7 and day 12 under BMP4-
boost conditions, whereas greater expression of this factor
was observed under standard conditions (Fig. 6A, upper
RUNX1 expression was higher than at day 0 (n ¼ 3).
However, this increase was less important after BMP4 boost
than under standard conditions (p < .005). The increase in
RUNX1 expression between day 7 and day 12 was significant
under standard conditions but nonsignificant under BMP4-
boost conditions (Fig. 6A, lower left).
We also measured OCT3/4, a marker of undifferentiated
hESCs (Fig. 6A lower right) (n ¼ 3). We show that expres-
sion of this factor decreased after day 7 and this decrease was
significantly more prominent after BMP4 boost than under
standard conditions. After BMP4 boost, expression of OCT3/4
was dramatically lower at day 12 relative to day 0 (144-fold).
This decrease was less important under standard conditions
(27-fold). We then analyzed the expression of OCT3/4 and
two other hESC markers, SSEA-3 and TRA1-60, in the
surface marker expression in CD144þKDRþsorted cells. (B): Immunodetection of CD31, vWF, and CD144 on hES-derived endothelial cells.
(C): TNF-a induced upregulation of ICAM-1. (D) Population doubling of hES EC and HUVECs. (E): Vascular tube formation on Matrigel.
Abbreviations: FSC, forward scatter; hES EC, human embryonic stem endothelial cell; HUVEC, human umbilical vein endothelial cell; ICAM-1,
intercellular adhesion molecule 1; KDR, kinase insert domain-containing receptor; TNF-a, tumor necrosis factor a; vWF, von Willebrand factor.
Characterization of day 12 CD144þKDRþsorted cells after 30 days of expansion. (A): Flow cytometry analysis of endothelial cell
Goldman, Feraud, Boyer-Di Ponio et al.
CD144þKDRþboosted population sorted at day 7 (Fig. 6B).
We show that, in this population, none of these factors was
detected either in the cytoplasm (OCT3/4) or at the surface of
the cells (SSEA-3 and TRA1-60).
hESCs recapitulate the sequential events giving rise to endo-
derm, mesoderm, and ectoderm  in a manner similar to
what occurs during human development. Because these cells
can be expanded almost indefinitely, they represent a useful
model of human development, allowing molecular studies that
are difficult to perform in human embryos. In this study, we
have further defined the steps of commitment and differentia-
tion of ESCs into endothelial cells. We first used culture con-
ditions that were previously defined as inducing the formation
of hemangioblastic cells by the Bathia group [15, 19].
Because these conditions were designed for obtaining both he-
matopoietic and endothelial cells, the culture medium con-
tained a panel of cytokines acting on the hematopoietic sys-
tem (IL-3, IL-6, G-CSF, Flt-3, SCF). We checked if all these
cytokines were also required for endothelial differentiation.
We show that IL-3, IL-6, and G-CSF did not affect the num-
ber of endothelial cells, but removal of SCF and Flt-3L
yielded a significant lower number of these cells. We then
investigated if BMP4 had an effect on the endothelial differ-
entiation process. BMP4 has been shown to be required for
mesoderm formation in mouse ESCs . BMP4 also plays a
crucial role in the commitment of mesoderm to the hemato-
poietic and endothelial lineages [15, 19, 21]. It has been
shown that exposure of hESCs to a short BMP4 treatment
induces endothelial differentiation . We have confirmed
this observation and extended the analysis of its effect on the
sequential emergence of endothelial cells during the differen-
tiation of H9 hESC-derived EBs. We first defined the optimal
BMP4 concentration and showed that 50 ng/ml was the mini-
mal dose yielding the maximal effect. Time of exposure to
BMP4 was also critical to reach the maximal effect. We show
that exposure of EBs to a 24-hour BMP4 (50 ng/ml) induction
We show that, similar to standard conditions, SCF and
Flt3L are also required in BMP4 boost conditions. This obser-
vation is consistent with the role of these factors in the forma-
tion of hemangioblastic cells and further differentiation of
these cells into endothelial cells [22–24]. CD133 was first
considered as a marker of hematopoietic progenitors , but
it was further described as being expressed by stem/progenitor
cells in different tissues . A number of studies have
shown that endothelial progenitor cells (EPCs) could be
defined by a combination of stem cell (such as CD133) and
CD45) in CD144þKDRþcells just after sorting. Expression of CD31 after 21 days of culture. (B): Population doubling. (C): Expression of
ICAM-1 after TNF-a treatment. (D): Migration assays. (E): Wound-healing assay. Abbreviations: BMP4, bone morphogenetic protein 4; FSC,
forward scatter; hES EC, human embryonic stem endothelial cell; HUVEC, human umbilical vein endothelial cell; ICAM-1, intercellular adhesion
molecule 1; TNF-a, tumor necrosis factor a.
Phenotypic characterization of BMP4-boosted endothelial cells. (A): Flow cytometric analysis of a set of markers (CD31, CD34,
Effects of BMP4 on hESC-Derived Endothelial Cells
endothelial (such as KDR, CD144) markers [26, 27]. In this
study, we thus considered two populations during EB differ-
entiation: the CD144þKDRþpopulation, for the identification
of all cellswithan endothelial
CD133þKDRþpopulation, for the identification of stem/pro-
We measured the number of CD144þKDRþcells at day 7
obtained under BMP4-boost conditions, comparing different
times of exposure with the standard, a continuous low level
of BMP4 (10 ng/ml). We found that BMP4 delivered on day
1 of EB differentiation at 50 ng/ml (for only 1 day) yielded a
peak in the population of CD144/KDR-expressing cells. We
thus used these induction conditions in further experiments.
Complete inhibition of induction of the CD144þKDRþpopu-
lation upon BMP4 boost in the presence of Noggin confirms
the direct involvement of BMP4 in this induction (Fig. 2C).
We then analyzed the kinetics of expression of KDR,
CD144, and CD133, both individually and in combination,
under standard and BMP4 boost conditions during the EB dif-
ferentiation process. CD144 expression peaked at day 7 under
BMP4-boost conditions and at day 12 under standard condi-
tions, suggesting that the BMP4 boost shortens the onset of
mesodermal cell commitment to the endothelial lineage. KDR
was expressed from day 0, in both the BMP4-boost and stand-
ard conditions, confirming that KDR is expressed on cells
other than endothelial cells during development. It has been
suggested that early KDRþcells could be a common meso-
dermal precursor that later gives rise to hemangioblastic cells
. At day 7, KDR expression was, however, significantly
higher after the BMP4 boost. Similar to KDR, CD133 is
expressed on native hESCs, but surprisingly, in contrast to the
KDRþpopulation, BMP4 boost reduced the number of
CD133þcells at every time point, and this reduction was sig-
nificant at day 9 and day 12.
When CD133 and KDR expression were analyzed in com-
bination, a dramatically lower expression level of this popula-
tion was observed after BMP4 boost, which was not observed
when these two markers were analyzed individually. This
indicates that this early CD133þKDRþpopulation is rapidly
induced to differentiate after BMP4 boost. Because day 2 EBs
did not express endothelial markers other than KDR, this pop-
ulation was not further analyzed.
The CD133þKDRþpopulation peaked at day 7 under
BMP4 boost conditions and at day 9 under standard condi-
tions. Aiming at identifying a population of endothelial pro-
genitors, we sorted the CD133þKDRþcells corresponding to
these two peaks. Cells corresponding to the day 9 peak were
unable to proliferate in culture and were not further analyzed.
In contrast, the BMP4-boosted CD133þKDRþpopulation
in day 7 and day 12 BMP4-stimulated and unstimulated EBs. (B): Flow cytometry analysis of OCT3/4, TRA1-60, and SSEA-3 cell surface
markers on CD144þKDRþsorted cells. Abbreviations: BMP4, bone morphogenetic protein 4; EB, embryoid body; FSC, forward scatter; KDR,
kinase insert domain-containing receptor; RT-PCR, reverse transcription-polymerase chain reaction; SSEA, stage-specific embryonic antigen;
TRA-1-60, tumor rejection antigen 1-60.
Expression of transcriptional factors during EB differentiation. (A): Quantitative RT-PCR analysis of GATA2, RUNX1, and OCT3/4
Goldman, Feraud, Boyer-Di Ponio et al.
sorted at day 7 represents 7% of the total cells and gives rise
to a highly proliferative population. These cells did not
express any endothelial markers, and therefore they were not
considered within the present study.
We then analyzed the CD144þKDRþpopulation that
peaks at day 12 under standard culture conditions. The pheno-
type of these cells (CD144þKDRþCD31þCD34þ) corresponds
to that of endothelial cells, and they represent 10% of the
total cells in these cultures. No hematopoietic cells were
detected, because neither CD45þ
detected. The absence of CD133 expression suggests that no
progenitor cells were included in the sorted CD144þKDRþpop-
ulation. In culture, sorted cells expressed KDR, CD144, and
CD31 on the surface of the cells and vWF in the cytoplasm.
The CD144þKDRþpopulation sorted at day 7 under
BMP4-boost conditions represents approximately 5% of the
total cells. Interestingly, the majority of sorted cells did not
express CD31, but after a few days, expression of this marker
increased, and by day 21, all cells were CD31þ. This indicates
that the initial CD144þKDRþpopulation contained immature
endothelial cells (early) that differentiated in vitro. The immature
phenotype of the initial CD144þKDRþCD31?population was
confirmed by the analysis of their cumulative proliferation curve.
These cells remained in the linear phase of their proliferation
longer than day 12 sorted CD144þKDRþcells that were cul-
tured under standard conditions (day 66 versus day 47).
Expression of ICAM-1 in the CD144þKDRþCD31?cells
was also induced by TNF-a, but the basal level of ICAM-1
expression was much lower in these early cells than in late
endothelial cells derived from day 12 EBs. Early, day 7 EB-
derived endothelial cells also displayed a more prominent
migration capacity than the late, day 12 EB-derived ones.
Finally, early, day 7 EB-derived endothelial cells have the
capacity to fill the gap or wounded area in an in vitro model of
wound healing. Taken together, these data suggest that endothe-
lial cells derived from day 7 BMP4-boosted EBs or from day 12
EBs cultured under standard conditions have all the phenotypic
and functional features of endothelial cells. However, the former
displayed a globally more immature phenotype than the latter.
EPCs have the capacity to generate secondary and tertiary
colonies in culture . EPC-like cells have been identified
during avian development . In contrast, even early endo-
thelial cells isolated at day 7 under BMP4 conditions failed to
form colonies in limiting dilution assays. This indicates that,
under our culture conditions, hESC-derived endothelial cells
do not correspond to endothelial stem/progenitor cells compa-
rable with EPCs. One of the major differences between EPCs
and hESC-derived endothelial cells is the microenvironment
of these cells. In the bone marrow, these cells are probably in
close contact with the hematopoietic niche, and, when blood
outgrowth EPCs are seeded in culture to generate colonies,
the initial adhesion step is performed in the presence of blood
cells, which represent >95% (CD45þcells) of the total cells.
In contrast, hematopoietic cells appear later during hESC dif-
ferentiation, and at day 7 under BMP4 induction conditions
no CD45þcells were detected inside EBs. hESC-derived en-
dothelial cells may thus be unable to form colonies because
they have not been instructed by an inducible niche, of which
hematopoietic cells may represent an essential component.
Finally, the mRNA level of three transcription factors,
GATA2, RUNX1, and OCT3/4, was analyzed by quantitative
reverse transcription-PCR at different stages of EB differen-
tiation, either under standard or BMP4-boost conditions.
GATA2 is a transcription factor expressed in both hematopoi-
etic and endothelial cells. It is also highly expressed in
hemangioblasts , and interaction between BMP4 and
GATA2 has been shown in different studies [17, 32]. We
saw, however, a lower level of GATA2 expression after
BMP4 boost (relative to that observed under standard condi-
tions). Such a decrease was also observed with RUNX1,
another hematopoietic/endothelial transcription factor involved
in the development of hemangioblasts. The lower expression
level of GATA2 and RUNX1 was more prominent at day 12
than at day 7. This indicates a restriction in GATA2/RUNX1-
expressing cells following BMP4 boost, probably linked to
the commitment of hemangioblasts to the endothelial lineage,
and a disappearance of cells with hematopoietic potential,
which cannot survive in a culture medium lacking hematopoi-
etic growth factors. This indicates that, not only are these fac-
tors required for hemangioblastic cell appearance, they are
also essential to hematopoietic commitment, voiding the de-
velopment of endothelial cells. BMP4 also accelerates the
Sorted CD144þKDRþcells at day 7 after
BMP4 boost show an immature endothelial
from day 7 to day 12 in standard culture
conditions exhibited a mature endothelial
phenotype. Abbreviations: AP, phosphatase
alcalin; BMP4, bone morphogenetic protein
4; EB, embryoid body; hES, human embry-
onic stem; KDR, kinase insert domain-con-
embryonic antigen; vWF, von Willebrand
Model of BMP4-induced endo-
differentiationfrom hES cells.
Effects of BMP4 on hESC-Derived Endothelial Cells
decrease in OCT3/4 expression, consistent with the effect of Download full-text
BMP4 on EB differentiation.
In conclusion, our paper not only confirms that functional
endothelial cells can be obtained from hESCs but also defines
superior conditions for their development. Under our standard
conditions, these endothelial cells represented up to 12% of
the total EB cells at day 12, a yield significantly better than
in other studies [10, 13, 15]. After BMP4 boost, functional
endothelial cells were detected earlier than under standard
conditions, suggesting that this factor accelerates the rate of
endothelial differentiation from mesoderm. This factor also
increases the number of endothelial cells, suggesting that it
forces commitment of mesoderm into the endothelial lineage.
Our results thus provide optimized conditions for obtaining
functional endothelial cells from hESCs. Endothelial cells at
different maturation stages could also be obtained, which
would represent a useful model for testing molecules involved
in endothelial development and differentiation. The cell popu-
lations obtained in our different culture and sorting conditions
are illustrated in Figure 7.
We thus propose a model of endothelial differentiation
that could be followed in culture. This model could be useful
for analyzing the mechanisms controlling the endothelial mat-
This work was supported by a research grant (IngeCell) from the
Ile de France region. We acknowledge Vanina Jodon de Vil-
leroche ´, Je ´ro ˆme Avouac, and Vale ´rie Vanneaux for their helpful
advice in experimental design, and Christophe Desterke for his
helpwith thestatistical analyses.
DISCLOSURE OF POTENTIAL CONFLICTS
The authors indicate no potential conflicts of interest.
1Yancopoulos GD, Davis S, Gale NW et al. Vascular-specific growth
factors and blood vessel formation. Nature 2000;407:242–248.
Flamme I, Fro ¨lich T, Risau W. Molecular mechanisms of vasculogen-
esis and embryonic angiogenesis. J Cell Physiol 1997;173:206–210.
Folkman J. Tumor angiogenesis: Therapeutic implications. N Engl J
Oyamada N, Itoh H, Sone M et al. Transplantation of vascular cells
derived from human embryonic stem cells contributes to vascular
regeneration after stroke in mice. J Transl Med 2008;6:54.
Chamberlain MC. Antiangiogenesis: Biology and utility in the treat-
ment of gliomas. Expert Rev Neurother 2008;8:1419–1423.
Thomson JA, Itskovitz-Eldor J, Shapiro SS et al. Embryonic stem cell
lines derived from human blastocysts. Science 1998;282:1145–1147.
Reubinoff BE, Pera MF, Fong CY et al. Embryonic stem cell lines
from human blastocysts: Somatic differentiation in vitro. Nat Biotech-
Zaret KS, Grompe M. Generation and regeneration of cells of the liver
and pancreas. Science 2008;322:1490–1494.
Aberdam E, Barak E, Rouleau M et al. A pure population of ectoder-
mal cells derived from human embryonic stem cells. Stem Cells 2008;
10 Levenberg S, Golub JS, Amit M et al. Endothelial cells derived from
human embryonic stem cells. Proc Natl Acad Sci U S A 2002;99:
11 Kennedy M, D’souza SL, Lynch-Kattman M et al. Development of
the hemangioblast defines the onset of hematopoiesis in human ES
cell differentiation cultures. Blood 2007;109:2679–2687.
12 Lu SJ, Feng Q, Caballero S et al. Generation of functional hemangio-
blasts from human embryonic stem cells. Nat Methods 2007;4:501–509.
13 Wang ZZ, Au P, Chen T et al. Endothelial cells derived from human
embryonic stem cells form durable blood vessels in vivo. Nat Biotech-
14 Candia AF, Watabe T, Hawley SHB et al. Cellular interpretation of
multiple TGF-b signals: Intracellular antagonism between activin/
BVg1 and BMP-2/4 signaling mediated by Smads. Development
15 Chadwick K, Wang L, Li L et al. Cytokines and BMP-4 promote he-
matopoietic differentiation of human embryonic stem cells. Blood
16 Bhatia M, Bonnet D, Wu D et al. Bone morphogenetic proteins regu-
late the developmental program of human hematopoietic stem cells.
J Exp Med 1999;189:1139–1148.
17 Pearson S, Sroczynska P, Lacaud G et al. The stepwise specification
of embryonic stem cells to hematopoietic fate is driven by sequential
exposure to BMP4, activin A, bFGF and VEGF. Development 2008;
18 Zhang P, Li J Tan Z et al. Short-term BMP-4 treatment initiates mesoderm
induction in human embryonic stem cells. Blood 2008;111:1933–1941.
19 Wang L, Li L, Shojaei F et al. Endothelial and hematopoietic cell fate
of human embryonic stem cells originates from primitive endothelium
with hemangioblastic properties. Immunity 2004;21:31–41.
20 Winnier G, Blessing M, Labosky PA et al. Bone morphogenetic pro-
tein-4 is required for mesoderm formation and patterning in the
mouse. Genes Dev 1995;9:2105–2116.
21 Boyd NL, Dhara SK, Rekaya R et al. BMP4 promotes formation of
primitive vascular networks in human embryonic stem cell-derived
embryoid bodies. Exp Biol Med 2007;232:833–843.
22 Loges S, Fehse B, Brockmann MA et al. Identification of the adult
human hemangioblast. Stem Cells Dev 2004;13:229–242.
23 Chung YS, Zhang WJ, Arentson E et al. Lineage analysis of the
hemangioblast as defined by FLK1 and SCL expression. Development
24 Park C, Afrikanova I, Chung YS et al. A hierarchical order of factors
in the generation of FLK1- and SCL-expressing hematopoietic and en-
dothelial progenitors from embryonic stem cells. Development 2004;
25 Boxall SA, Cook GP, Pearce D et al. Haematopoietic repopulating ac-
tivity in human cord blood CD133þquiescent cells. Bone Marrow
26 Peichev M, Naiyer AJ, Pereira D et al. Expression of VEGFR-2 and
AC133 by circulating human CD34(þ) cells identifies a population of
functional endothelial precursors. Blood 2000;95:952–958.
27 Asahara T, Murohara T, Sullivan A et al. Isolation of putative progen-
itor endothelial cells for angiogenesis. Science 1997;275:964–967.
28 Ema M, Rossant J. Cell fate decisions in early blood vessel formation.
Trends Cardiovasc Med 2003;13:254–259.
29 Ingram DA, Mead LE, Tanaka H et al. Identification of a novel hier-
archy of endothelial progenitor cells using human peripheral and um-
bilical cord blood. Blood 2004;104:2752–2760.
30 Pardanaud L, Eichmann A. Identification, emergence and mobilization
of circulating endothelial cells or progenitors in the embryo. Develop-
31 Lugus JJ, Chung YS, Mills JC et al. GATA2 functions at multiple
steps in hemangioblast development and differentiation. Development
32 Xu RH, Ault KT, Kim J et al. Opposite effects of FGF and BMP-4 on
embryonic blood formation: Roles of PV.1 and GATA-2. Dev Biol
Goldman, Feraud, Boyer-Di Ponio et al.