Identification and Characterization of Hemoangiogenic Progenitors During Cynomolgus Monkey Embryonic Stem Cell Differentiation

Identification and Characterization of HemoangiogenicProgenitors During Cynomolgus Monkey EmbryonicStem Cell Differentiation

KATSUTSUGU UMEDA,a TOSHIO HEIKE,a MOMOKO YOSHIMOTO,a GEN SHINODA,a MITSUTAKA SHIOTA,a

HIROFUMI SUEMORI,b HONG YUAN LUO,c DAVID H. K. CHUI,c RYUZO TORII,d MASABUMI SHIBUYA,e

NORIO NAKATSUJI,f TATSUTOSHI NAKAHATAa

aDepartment of Pediatrics, Graduate School of Medicine, bLaboratory of Embryonic Stem Cell Research, Stem Cell

Research Center, Institute for Frontier Medical Sciences, fDepartment of Development and Differentiation, Institute

for Frontier Medical Science, Kyoto University, Kyoto, Japan; cDepartment of Medicine, Boston University School

of Medicine, Boston, Massachusetts, USA; dResearch Center for Animal Life Science, Shiga University of Medical

Science, Ohtsu, Japan; eDivision of Genetics, Institute of Medical Science, University of Tokyo, Tokyo, Japan

Key Words. Embryonic stem cell • Primate • Hemangioblast • Vascular endothelial growth factor

ABSTRACT

We identified intermediate-stage progenitor cells that havethe potential to differentiate into hematopoietic and endo-thelial lineages from nonhuman primate embryonic stem(ES) cells. Sequential fluorescence-activated cell sorting andimmunostaining analyses showed that when ES cells werecultured in an OP9 coculture system, both lineages devel-oped after the emergence of two hemoangiogenic progeni-tor-bearing cell fractions, namely, vascular endothelialgrowth factor receptor (VEGFR)-2high CD34– and VEGFR-2high CD34� cells. Exogenous vascular endothelial growthfactor increased the proportion of VEGFR-2high cells, par-ticularly that of VEGFR-2high CD34� cells, in a dose-depen-

dent manner. Although either population of VEGFR-2high

cells could differentiate into primitive and definitive hema-topoietic cells (HCs), as well as endothelial cells (ECs), theVEGFR-2high CD34� cells had greater hemoangiogenic po-tential. Both lineages developed from VEGFR-2high CD34–

or VEGFR-2high CD34� precursor at the single-cell level,which strongly supports the existence of hemangioblasts inthese cell fractions. Thus, this culture system allows differenti-ation into the HC and EC lineages to be defined by surfacemarkers. These observations should facilitate further studiesboth on early developmental processes and on regenerationtherapies in human. STEM CELLS 2006;24:1348–1358

INTRODUCTIONIt has been reported previously that development of hematopoi-etic cells (HCs) and endothelial cells (ECs) is closely associated[1, 2]. These observations suggest that both lineage cells sharea common precursor, which has been called the hemangioblast.Further supporting the putative existence of the hemangioblastis an immunohistochemical study of murine embryos that re-vealed HC clusters adhering to the ventral floor of the dorsalaorta [3]. Additionally, both HCs and ECs share common sur-face markers such as vascular endothelial growth factor receptor(VEGFR)-2, CD34, CD31, tyrosine kinase with Ig and EGFhomology domain (Tie)-1, and Tie-2 [3, 4]. Of these antigens,VEGFR-2 (also known as Flk-1 in the mouse) is a candidatemarker for hemangioblasts, given that murine embryos or ES

cells that do not express VEGFR-2 completely fail to producecells from either lineage [5, 6]. Moreover, recent studies on thedifferentiation of murine ES cells have shown that both HCs andECs can be generated from VEGFR-2� cells under variousculture conditions [7, 8].

Some immunohistochemical studies of human embryoshave shown a spatially close association in the development ofboth HCs and ECs [9–11]. However, the developmental rela-tionship between HCs and ECs in human embryogenesis has notbeen further elucidated because of various obstacles, includingthe ethical restrictions on experiments using human embryos.

To understand the mechanisms that regulate the differenti-ation of HCs and ECs in humans, it is currently promising to useprimate (human and monkey) ES cells. We have recently

Correspondence: Tatsutoshi Nakahata, M.D., Ph.D., Department of Pediatrics, Graduate School of Medicine, Kyoto University, 54Kawahara-cho, Shogoin, Sakyo-ku, Kyoto 606-8507, Japan. Telephone: 81-75-751-3290; Fax: 81-75-752-2361; e-mail:[email protected] Received April 11, 2005; accepted for publication December 15, 2005; first published online in STEMCELLS EXPRESS January 12, 2006. ©AlphaMed Press 1066-5099/2006/$20.00/0 doi: 10.1634/stemcells.2005-0165

EMBRYONIC STEM CELLS

STEM CELLS 2006;24:1348–1358 www.StemCells.com

established a culture system, which induces hematopoietic celldifferentiation from cynomolgus monkey ES cells by coculturewith OP9 stromal cells [12]. We investigated the close associ-ation of HC and EC development during ES cell differentiationin this culture system using fluorescence-activated cell sorting(FACS) and subsequent culture of the sorted fractions. Here, weshow that both lineages developed after VEGFR-2high cellsemerged on day 6, when neither lineage was observed. BothHCs and ECs were generated from single-cell cultures ofVEGFR-2high cells, which strongly supports the existence ofhemangioblasts in primates.

MATERIALS AND METHODS

Cell LinesThe ES cell line CMK6, which was established from cynomol-gus monkey blastocysts, was maintained as described previously[13]. The enhanced green fluorescent protein (GFP)-transfectedES cell subline, which we established previously [14], was usedto distinguish ES cell-derived cells, except in the experimentsthat involved immunostaining with antibodies (Abs) to humanhemoglobin. The OP9 stromal cell line, a kind gift from Dr.Hiroaki Kodama, was maintained as previously reported [12].

Cytokines and Growth FactorsRecombinant human granulocyte colony-stimulating factor (G-CSF), erythropoietin (EPO), interleukin (IL)-3, and stem cellfactor (SCF) were kindly provided by Kirin Brewery (Tokyo,http://www.kirin.co.jp/english). Recombinant human vascularendothelial growth factor (VEGF) was purchased from R&DSystems (Minneapolis, http://www.rndsystems.com).

AntibodiesThe primary Abs used in this study were mouse anti-humanCD34 (clone 563) and CD41a (clone HIP8) (BD Pharmingen,San Diego, http://www.bdbiosciences.com/pharmingen); mouseanti-human c-kit and rabbit anti-human vWF (Nichirei, Tokyo,http://www.nichirei.co.jp), mouse anti-human CD45 (clone2B11�PD7/26), and CD41 (clone 5B12) (DAKO, Kyoto, Ja-pan, http://www.dako. com), mouse anti-human vascular endo-thelial cadherin (VE-cadherin, clone TEA1/31) (Immunotech,Luminy, France, http://www.immunotech.com), mouse anti-hu-man CD31 (clone WM59) (eBioscience, San Diego, CA, http://baybio.co.jp), mouse anti-human CD11b (clone Bear1) (Beck-man Coulter, Fullerton, CA, http://www.beckmancoulter.com),rabbit anti-human hemoglobin (Hb) (Cappel, Aurora, OH),mouse anti-human �-globin (Hb�) (Santa Cruz Biotechnology,Santa Cruz, CA, http://www.scbt.com), and mouse anti-TRA-1–60 (clone TRA-1–60) (Chemicon, Temecula, CA, http://www.chemicon.com). Mouse anti-stage-specific embryonic an-tigen (SSEA)-4 monoclonal antibody (mAb) developed byKannagi et al. [15] was obtained from the Developmental Stud-ies Hybridoma Bank developed under the auspices of the Na-tional Institute of Child Health and Human Development andmaintained by the Department of Biological Sciences, Univer-sity of Iowa (Iowa City, IA). The mouse anti-human �-globin(Hb�) [16] and VEGFR-2 mAbs [17] were used as previouslyreported. All primary antibodies against human antigens thatwere used in this study cross- react to cynomolgus monkeyproteins, as previously reported [12, 18, 19]. The secondary Abs

used in this study were Cy3-conjugated donkey anti-mouse IgG,horseradish peroxidase-conjugated donkey anti-mouse IgG, flu-orescein isothiocyanate (FITC)-conjugated donkey anti-rabbitIgG, and alkaline phosphatase (ALP)-conjugated donkey anti-mouse IgG (Jackson ImmunoResearch Laboratories Inc, WestGrove, PA, http://www.jacksonimmuno.com), phycoerythrin(PE)-conjugated goat anti-mouse IgG (Dako), PE-conjugatedgoat anti-mouse IgM (eBioscience), and allophycocyanin(APC)-conjugated goat anti-mouse IgG (BD Pharmingen).

Staining and the Dil-Ac-LDL Incorporation AssayMay-Giemsa staining and immunostaining were performed aspreviously reported [12]. For the 1,1�-dioctadecyl-1,3,3,3�,3�-tetramethylindocarbocyanine-labeled acetylated low-density li-poprotein (Dil-Ac-LDL) incorporation assay, adherent cellswere incubated with 10 �g/ml Dil-Ac-LDL (Molecular ProbesInc., Eugene, OR, http://probes.invitrogen.com) in culture me-dium for 4 hours at 37°C. These cells were then washed with�-minimum essential medium (�-MEM) (Gibco, Grand Island,NY, http://www.invitrogen.com) and observed by fluorescencemicroscopy (FLUOVIEW System; Olympus, Tokyo, http://www.olympus-global.com). After the DiI-Ac-LDL incorporation assay,the cells were then fixed and used for antibody staining.

FACS Analysis and Cell SortingStaining procedures, FACS analysis, and cell sorting were per-formed as reported previously [12]. Briefly, the cultured cellswere harvested with cell dissociation buffer (Invitrogen, Carls-bad, CA, http://www.invitrogen.com) and incubated with PE- orAPC-conjugated Abs or unconjugated Abs for 30 minutes. Sam-ples stained with unconjugated Abs were then incubated withPE- or APC-conjugated goat anti-mouse Abs. Nonviable cellswere excluded from the analysis by propidium iodide costain-ing. FACS analysis was performed with a FACScaliber instru-ment with the CellQuest program (Becton Dickinson Labware,Bedford, MA, http://www.bd.com). Cell sorting with PE-conju-gated CD34 and APC-conjugated VEGFR-2 mAbs was performedusing a FACSVantage flow cytometer (Becton Dickinson).

Reverse Transcription-Polymerase Chain ReactionRNA isolation and reverse transcription-polymerase chain reac-tion (RT-PCR) were performed as described previously [12].Samples were initially denatured at 94°C for 5 minutes, fol-lowed by 35–40 amplification rounds consisting of 94°C for 1minute (denaturing), 60°C for 1 minute (annealing), and 72°Cfor 1 minute (extension), followed by a final extension at 94°Cfor 7 minutes. The primers used for RT-PCR were as follows:GATA-1 (498 bp), forward, 5�-CAC ATC CCC AAG GCGGCC GAA C-3�, reverse, 5�-AGG TCT GGG CTC AGC CGCTCT-3�; MYB (307 bp), forward, 5�-CAC GCT GGG CCTGTC ATC AAC-3�, reverse, 5-GCA TGG CTC TTC GTG TTATAG C-3�; FLI-1 (412 bp), forward, 5�-ATG GAT CCA GGGAGT CTC CGG T-3�, reverse, 5�-TTG GTC GGT GTG GGAGGT TGT-3�; Tie-1 (308 bp), forward, 5�-TGG TCG GAGAGA ACC TAG CC-3�, reverse, 5�-GAC GCA TCA GCT CGTACA CTT C-3�; eNOS (557 bp), forward, 5�-GAC ATT TTCGGG CTC ACG CTG-3�, reverse, 5�-TGG GGT AGG CACTTT AGT AGT TC-3�; GATA-2 (303 bp), forward, 5�-TGGCGC ACA ACT ACA TGG AAC-3�, reverse, 5�-GAG GGGTGC AGT GGC GTC TT-3�; SCL (185 bp), forward, 5�-TCT

1349Umeda, Heike, Yoshimoto et al.

www.StemCells.com

CGG CAG CGG GTT CTT TG-3�, reverse, 5�-AAG GCC CCGTTC ACA TTC TGC-3�; FLT-1 (508 bp), forward, 5�-GCTCAC CAT GGT CAG CTA CTG-3�, reverse, 5�-CAG TGATGT TAG GTG ACG TGA ACC-3�; Rex-1 (489 bp), forward,5�-CGC GGT GTG GGC CTT ATG TG-3�, reverse, 5�-TCTCAG GGC AGC TCT ATT CCT C-3�; Oct-4 (246 bp), forward,5�-CGT GAA GCT GGA GAA GGA GAA GCT G-3�, reverse,5�-CAA GGG CCG CAG CTT ACA CAT GTT C-3�; GAPDH(360 bp), forward, 5�-CAC CAG GGC TGC TTT TAA CTCTG-3�, reverse, 5�-ATG GTT CAC ACC CAT GAC GAA C-3�.cDNA from cynomolgus monkey bone marrow, human umbil-ical vein endothelial cells (HUVECs), and human erythroleuke-mia K562 cells served as positive controls. For semiquantitativecomparisons, samples were normalized by dilution to giveequivalent signals for GAPDH.

In Vitro Differentiation of ES CellsFor initial differentiation induction, trypsin-treated undifferen-tiated ES cells were transferred onto fresh confluent OP9 cells insix-well plates at a concentration of 4 � 103 cells per well andcultured with various concentrations of VEGF (0, 10, 20, and 40ng/ml) in �-minimum essential medium supplemented with10% fetal calf serum (FCS) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) and 50 �M 2-mercaptoethanol (2ME).The cultured cells were harvested in cell dissociation buffer(Invitrogen) and analyzed by FACS, as described above, ondays 4, 6, 8, and 10.

To induce the differentiation of HCs and ECs, cells that hadbeen cultured for 6 days in the presence of 20 ng/ml VEGF wereharvested and sorted by FACS according to the expression ofVEGFR-2 and CD34, as detailed in Results. Each sorted cellfraction was transferred onto fresh confluent OP9 cells in six-wellplates at a concentration of 1 � 104 cells per well or in 12-wellplates at a concentration of 1 � 103 cells per well. To analyze thedevelopment of HCs, the sorted cells were cultured in �-MEMsupplemented with 10% FCS (Sigma-Aldrich), 50 �M 2ME, and amixture of 10 ng/ml G-CSF, 2 U/ml EPO, 20 ng/ml IL-3, and 100ng/ml SCF (hematopoietic cytokine mixture). Floating and adher-ent HCs were analyzed sequentially, as previously reported [12,20]. To analyze EC development, the sorted cells were cultured in�-MEM supplemented with 10% FCS, 50 �M 2ME, and 20 ng/mlVEGF. Six days after cell sorting, the cells were stained withanti-VE-cadherin and ALP-conjugated anti-mouse IgG, and the ECclusters were scored by microscopy. At least three independentexperiments were conducted.

Single-Cell Deposition Assay for Hematopoietic andEndothelial DifferentiationThe deposition of single sorted cells into individual wells of96-well plates was carried out by using the Clon-Cyt system ofthe FACSVantage flow cytometer (Becton Dickinson). Individ-ual sorted cells from each fraction were seeded onto OP9stromal cells in �-MEM supplemented with 10% FCS and ahematopoietic cytokine mixture. After 6 days in culture, HCdevelopment was evaluated by immunostaining with the anti-CD45, CD41, and HbF mAbs, whereas EC development wasanalyzed by immunostaining with the anti-VE-cadherin mAb orby the DiI-Ac-LDL incorporation assay. The concomitant de-velopment of both lineage of cells was confirmed by immuno-staining with the anti-CD34 mAb.

StatisticsDifferences in the number of HC or EC cluster between twogroups were assessed using Student’s t test. Differences in thefrequency of HC and/or EC cluster development in the single-cell deposition assay were assessed using the �2 test. Statisticalsignificance was defined as p values less than .05.

RESULTS

FACS Analysis of Hematopoietic and/or EndothelialSurface Markers During Early PrimateES DifferentiationThe ES cell line CMK6 and the GFP-transfected ES cell sublinethat were used in this study both expressed the undifferentiated-state marker SSEA-4, even after being maintained in culture formore than a year (data not shown). We confirmed that both EScell lines were equally capable of differentiating into HCs andECs (data not shown).

By RT-PCR analysis of cultures in the OP9 coculture sys-tem, we have demonstrated previously that sequential expres-sion of genes associated with both HC and EC lineage devel-opment was equivalent to that seen during primate ontogeny invivo [12]. FACS analysis was used to determine the expressionpatterns of various surface markers involved in HC and ECdevelopment when GFP-transfected ES cells were induced todifferentiate by coculture with OP9 stromal cells in the presenceor absence of exogenous VEGF (Fig. 1). The numbers of cellsexpressing particular markers were quantified as a percent of thetotal live GFP� cells in the culture (Fig. 1A). Although sub-stantial fraction of the GFP� cells were dead after being har-vested with cell dissociation buffer, harvesting by other means,such as by using trypsin or collagenase, did not significantlyalter the proportion of dead cells (data not shown). The markerscould be classified into three groups depending on their expres-sion kinetics. The first group includes CD34 and CD31, whichare expressed by both early HCs and ECs (Fig. 1B, 1C). Theirexpression was not detected in undifferentiated ES cells, butbecame upregulated on day 6. Thereafter, CD34� and CD31�

cells increased, especially in the presence of exogenous VEGF.The second group includes CD45, CD41a, and VE-cadherin(Fig. 1D–1F), whose expression is specific to either HCs (CD45and CD41a) or ECs (VE-cadherin). Expression of these proteinswas not detected in undifferentiated ES cells but became slightlyupregulated by day 10 in the presence of VEGF. The third groupincludes c-kit and VEGFR-2 (Fig. 1G, 1H). Although expres-sion of these proteins was detected in almost all undifferentiatedES cells, it became downregulated on day 4 and thereafter, withor without exogenous VEGF. These kinetics were similar to thatof SSEA-4 (Fig. 1I), which is expressed by undifferentiated EScells [13, 21]. These observations showed that during in vitroHC and EC differentiation, common surface markers such asCD34 and CD31 were expressed first, followed by the expres-sion of lineage-specific markers.

Generation of VEGFR-2highCD34– and VEGFR-2highCD34� Cells Before HC or EC DevelopmentDuring Primate ES Cell DifferentiationSeveral studies have demonstrated that VEGFR-2 is a keymarker of hemangioblasts during early murine development[5–8]. CD34 and CD31 are also expressed by early

1350 Hemoangiogenic Progenitors from Monkey ES Cell

hematopoietic and endothelial progenitors [9, 11]. Therefore, toidentify bipotential progenitor cells in primates, we analyzed thedifferentiating ES cell-derived cells by FACS using VEGFR-2,CD34, and CD31 mAbs. The undifferentiated ES cells did notexpress CD34 or CD31, whereas approximately 80% of themexpressed VEGFR-2 at low levels (Fig. 2A, 2B). We examinedthe expression levels of VEGFR-2 more closely and found thatthe proportion of VEGFR-2low cells gradually decreased duringcoculture and that VEGFR-2high cells could be detected on day6. More than half of the VEGFR-2high cells were CD34-negativeon day 6, but VEGFR-2high CD34� cells increased by day 8 andthereafter (Fig. 2A). The same temporal expression pattern ofCD31 among the VEGFR-2high cells was observed (Fig. 2B).FACS and immunostaining analysis showed that the VEGFR-2high cells that emerged on day 6 were negative for CD41a,CD45, VE-cadherin, or any hemoglobins (data not shown),indicating that these cells did not yet express HC or EC lineage-specific markers.

We further examined the differentiation state of theVGEFR-2high cells by double staining with mAbs for VEGFR-2and TRA-1–60, a surface marker indicative of undifferentiatedES cells. The expression of TRA-1–60 was detected in undif-ferentiated ES cells, as previously reported [13, 21], and wasupregulated on day 4. However, although approximately half ofthe GFP� ES cell-derived cells were positive for TRA-1–60 onday 6 and thereafter, the VEGFR-2high cells were always neg-ative for TRA-1–60 at all time points (Fig. 2C).

To verify the potential of day 6 VEGFR-2high CD34– andVEGFR-2high CD34� cells to differentiate into the HC and EClineages, their gene expression profiles were investigated byRT-PCR. The genes analyzed were those representing HC

(GATA-1, MYB, and FLI-1) [22], EC (Tie-1 and eNOS) [23,24], or HC-EC potentials (SCL, FLT-1, and GATA-2) [25–27].As shown in Figure 2D, GATA-1 and SCL were expressed byVEGFR-2high CD34� cells but not by VEGFR-2high CD34–

cells, whereas FLI-1 expression was up-regulated in bothVEGFR-2high cell populations. In contrast, the expression pro-files of the other HC and/or EC markers did not correlate withthe development of VEGFR-2high cells.

We also analyzed the expression of Rex-1 and Oct-4,which are marker genes for undifferentiated ES cells [21].Although their expression was still detected in the total GFP�

cell populations on day 6, they were not expressed by eitherVEGFR-2high cell population. These observations togetherindicate that the day 6 VEGFR-2high cell population differsfrom other ES cell-derived cells that emerge during thedifferentiation induction, as they express genes characteristicfor the HC and/or EC lineages.

We then analyzed the effect of VEGF on VEGFR-2high celldevelopment by adding various concentrations of VEGF to theculture (Table 1). FACS analysis demonstrated that the presenceof VEGF increased both the proportion of VEGFR-2high cells inthe culture and the percentage of VEGFR-2high CD34� cellsamong the VEGFR-2high cell fraction. This effect was dose-dependent and saturated at 20 ng/ml VEGF. The same trendswere observed by analysis with VEGFR-2 and CD31 mAbs.

HC Development from VEGFR-2high FractionsGiven the results described above, we hypothesized that theVEGFR-2high CD34– or the VEGFR-2high CD34� cells on day6 may contain hemoangiogenic progenitors. As shown in Figure1D–1F, rather few HCs and ECs develop from ES cells over 10

Figure 1. Fluorescence-activated cell sort-ing (FACS) analysis of surface markers ex-pressed by hematopoietic cells and/or endo-thelial cells during embryonic stem (ES) celldifferentiation. Undifferentiated green fluo-rescent protein (GFP)-transfected ES cellswere seeded onto confluent OP9 stromalcells in six-well plates in the presence orabsence of 20 ng/ml exogenous vascular en-dothelial growth factor (VEGF). The undif-ferentiated ES cells and the subsequent cul-tures on days 4, 6, 8, and 10 were stainedwith various monoclonal antibodies andquantified by FACS analysis. (A): Theamounts of GFP� ES-derived live cells(boxed region) are shown as a percentage ofthe total cells in cultures. (B–I): Cells thatstained positive for each antigen, CD34 (B),CD31 (C), CD45 (D), CD41a (E), VE-cad-herin (F), c-kit (G), VEGFR-2 (H), andSSEA-4 (I), are shown as a percentage of thetotal GFP� cells. The white and black circlesshow the results in the absence and presenceof VEGF, respectively. The data representthe mean � SD of three independent exper-iments. Abbreviations: SSEA, stage-specificembryonic antigen; VE-cadherin, vascularendothelial cadherin; VEGFR, vascular en-dothelial growth factor receptor.


www.StemCells.com

Table 1. Effects of VEGF on the generation of VEGFR-2high cells

�VEGF�VEGFR-2high cells(% of GFP� cells)

CD34� cells(% of VEGFR-2high cells)

CD31� cells(% of VEGFR-2high cells)

Nil 5.7 � 0.7 21.2 � 0.9 18.6 � 4.4V10 5.8 � 0.5 23.8 � 1.2 23.5 � 1.2V20 8.2 � 0.5 36.0 � 2.7 30.0 � 0.6V40 8.7 � 1.0 35.0 � 7.4 35.7 � 7.5

ES cells were cultured with various concentrations of VEGF (0, 10, 20, and 40 ng/ml; Nil, V10, V20, and V40, respectively) for 6 days,and the resultant cells were then analyzed with monoclonal antibodies specific for VEGFR-2 and CD34, or VEGFR-2 and CD31. Thedata are displayed as either the average percentage of VEGFR-2high cells among the total live GFP� ES cell-derived cells, or the % ofthe VEGFR-2high CD34� or CD31� cells among the total VEGFR-2high cells. All percentages are the mean � SD of three independentexperiments.Abbreviations: ES, embryonic stem; GFP, green fluorescent protein; VEGF, vascular endothelial growth factor; VEGFR, vascularendothelial growth factor receptor.

Figure 2. Fluorescence-acti-vated cell sorting and reversetranscription-polymerase chain re-action (RT-PCR) analysis of theVEGFR-2high cells that emergeduring ES cell differentiation. Thegreen fluorescent protein (GFP)-transfected ES cells and the subse-quent cultures on days 4, 6, 8, and10 in the absence of exogenousvascular endothelial growth factorwere analyzed by FACS with an-tibodies against VEGFR-2 andCD34 (A), VEGFR-2 and CD31(B), or VEGFR-2 and TRA-1–60(C). (A, B): The amounts ofVEGFR-2high CD34– or VEGFR-2high CD31– cells (upper left quad-rant), or VEGFR-2high CD34� orVEGFR-2high CD31� cells (upperright quadrant), are shown as a per-centage of the total GFP� ES cells.(C): The numbers in each quadrantrepresent percentages of cellsamong the total GFP� ES cells.(D): ES cells cocultured with OP9cells for 6 days were subjected toRT-PCR analysis of genes associ-ated with HC and/or EC develop-ment. The lanes contained mRNAfrom the following cells: adultcynomolgus monkey bone marrowcells (lane 1), the human erythro-blastic cell line K562 (lane 2), hu-man umbilical vein endothelialcells (lane 3), OP9 stromal cells(lane 4), undifferentiated ES cells(lane 5), the total GFP� ES cell-derived cells (lane 6), the VEGFR-2high CD34– cells (lane 7), and theVEGFR-2high CD34� cells (lane8), harvested on day 6. Represen-tative results from one of three in-dependent experiments are shown.Abbreviations: EC, endothelial cells;ES, embryonic stem; HC, hemato-poietic cells; M, size marker;VEGFR, vascular endothelialgrowth factor receptor.


days of culture. However, we have shown previously that thenumbers of HCs that develop from ES cells markedly increasedif the cells are replated onto a new confluent OP9 cell layer andthat abundant hematopoiesis, in particular definitive hematopoi-esis, cannot develop without additional VEGF [12]. Therefore,we sorted the cultures using anti-VEGFR-2 and CD34 mAbsafter initial 6-day VEGF treatment, and each fraction was re-plated onto a new confluent OP9 cell layer. Fractions ofVEGFR-2high CD34–, VEGFR-2high CD34�, VEGFR-2low

CD34–, VEGFR-2– CD34–, and VEGFR-2low or – CD34� cellswere collected, replated, and analyzed for their capacity togenerate HCs and/or ECs, as shown schematically in Figure 3A.FACS reanalysis of the sorted VEGFR-2high CD34� cellsshowed their purity ranged from 93.0%–97.0%, whereas thepurity of the other sorted fractions ranged from 99.0%–99.7%(Fig. 3B and data not shown).

Adherent HCs first emerged from the VEGFR-2high CD34–

and VEGFR-2high CD34� fractions 2 days after cell sorting andreplating (Fig. 4A, 4B). The adherent HCs, which formed clus-ters on or underneath the OP9 stromal layer, are known tocontain immature hematopoietic progenitors [12, 20]. Almost allof the adherent HC clusters were positive for CD34 (Fig. 4H)and also contained cells that were positive for VEGFR-2- (Fig.4I), CD45- (Fig. 4J), CD11b- (Fig. 4K), CD41- (Fig. 4L), andHb� (Fig. 4M). The number of adherent HC clusters was max-imal on day 10 but decreased thereafter. On day 10, the numberof clusters generated from the VEGFR-2high CD34� fraction

was approximately four times the number generated from theVEGFR-2high CD34– fraction. Furthermore, larger clusters weredetected from the VEGFR-2high CD34� fractions (Fig. 4M), andadherent clusters from the VEGFR-2high CD34� fraction cov-ered the stromal layer by day 20 (Fig. 4E), whereas adherentclusters from the VEGFR-2high CD34– fraction were rarelyobserved (Fig. 4D) and disappeared over time in culture.

In this coculture system, HC development in the floatingfractions first occurred on day 8 (2 days after seeding on day 6).The floating cells were found to consist exclusively of matureHCs, such as erythrocytes, myeloid lineage cells, andmegakaryocytes (data not shown) [12]. We examined whetherboth VEGFR-2high fractions were capable of both primitive anddefinitive hematopoiesis by sequential May-Giemsa stainingand immunostaining with hemoglobin Abs. Until day 15 (9 daysafter sorting), all of the floating HCs from both fractions werelarge nucleated erythrocytes positive for Hb�, Hb�, and Hb (Fig.4C, 4N–4Q). These cells correspond to primitive erythrocytes(EryP). On day 18 and thereafter, both VEGFR-2high CD34– andVEGFR-2high CD34� cultures contained some small erythro-cytes, including enucleated erythrocytes, that were positive forHb� and Hb but negative for Hb� (Fig. 4F, 4R–4U). These cellscorrespond to definitive erythrocytes (EryD). The number oferythrocytes generated from either cell fraction was maximal onday 12 but gradually decreased thereafter. Subsequently, a sec-ond wave of erythrocytes appeared around day 21 (Fig. 4W).The proportion of EryD gradually increased around day 18 and

Figure 3. Schematic representation of theassays used to analyze the differentiation ofHCs and ECs from ES cells. (A, B): Thecells were cocultured with OP9 cells in thepresence of 20 ng/ml vascular endothelialgrowth factor (VEGF) and sorted usingVEGFR-2 and CD34 mAbs on day 6. TheFACS plot (B) shows the regions (R3–R7)that were used to define the different popu-lations of cells: VEGFR-2high CD34–, R3;VEGFR-2high CD34�, R4; VEGFR-2low

CD34–, R5; VEGFR-2– CD34–, R6;VEGFR-2low or – CD34�, R7. The sortedcells from each population were then trans-ferred onto fresh confluent OP9 cells. Toanalyze HC development, the sorted cellswere cultured in �-minimum essential me-dium (�-MEM) supplemented with 10% fe-tal calf serum (FCS), 2-mercaptoethanol(2ME), and a mixture of granulocyte colony-stimulating factor, erythropoietin, interleu-kin-3, and stem cell factor (hematopoieticcytokine mixture). The number of adherentHC clusters were counted 4 days after cellsorting, and the floating cells were analyzedevery 2 days. To analyze EC development,the sorted cells were cultured in �-minimumessential medium (�-MEM) supplementedwith 10% FCS, 2ME, and VEGF. Cultureswere analyzed by immunostaining with an

antibody against vascular endothelial cadherin 6 days after cell sorting. For the single-cell deposition assay, individual sorted cells from theVEGFR-2high CD34– and VEGFR-2high CD34� fractions were seeded onto OP9 stromal cells and cultured with a hematopoietic cytokine mixture.Six days later, HC and EC development was evaluated by immunostaining or by the 1,1�-dioctadecyl-1,3,3,3�,3�-tetramethylindocarbocyanine-labeledacetylated low-density lipoprotein incorporation assay. FACS reanalysis showed that the purity of the sorted VEGFR-2high CD34– and VEGFR-2high

CD34� cell populations was 99.0%–99.7% and 93.0%–97.0%, respectively (see the right-hand dotplots for examples). Abbreviations: EC, endothelialcell; ES, embryonic stem; HC, hematopoietic cell; VEGFR, vascular endothelial growth factor receptor.


www.StemCells.com

Figure 4. HC developmentfrom the VEGFR-2high CD34–

and VEGFR-2high CD34� frac-tions. (A, B): Micrographs ofadherent HC clusters generatedon day 10 from the VEGFR-2high CD34– fraction (A) andthe VEGFR-2high CD34� frac-tion (B). (C, F): May-Giemsastaining of floating HCs ondays 12 (C) and 27 (F). (D, E):Micrographs of adherent HCclusters generated on day 27from the VEGFR-2high CD34–

fraction (D) and the VEGFR-2high CD34� fraction (E). (G–M): Alkaline phosphatase de-tection of adherent HCs stainedwith antibodies to CD34 (H),VEGFR-2 (I), CD45 (J),CD11b (K), CD41 (L), andHb� (M). The staining with theisotype control IgG1 is shownin G. (N, R) Immunostainingof hemoglobin (Hb) (fluores-cein isothiocyanate [FITC[)and Hb� (Cy3) in erythrocyteson days 12 (N) and 27 (R).(O–Q, S–U): Immunostainingof Hb (FITC) and Hb� (Cy3) inerythrocytes on days 12 (O–Q)and 27 (S–U). Merged imagesare shown in N, Q, R, and U.Nuclei were labeled withHoechst 33342 (N–U). The an-ti-human Hb Ab, which reactswith embryonic, fetal, andadult erythrocytes, was used todetect all erythrocytes amongthe floating cells during cul-ture. The same staining resultswere obtained from both theVEGFR-2high CD34– andVEGFR-2high CD34� frac-tions. Original magnification,�100 (A, B, D, E, G–M) and�400 (C, F, N–U). Scale bars� 100 �m (A, B, D, E, G–M)and 10 �m (C, F, N–U). (V):The number of adherent HCclusters in the indicated frac-tions. Small clusters (whitebars), which consisted of20–49 round blast-like cells,and large clusters (black bars),which consisted of more than

50 cells, were counted on day 10. (W): Sequential analysis of the number oferythrocytes generated from the VEGFR-2high CD34– (white circles) andVEGFR-2high CD34� (black circles) populations. (X): Sequential analysis ofthe proportion of definitive erythrocytes (EryD) among all the erythrocytesgenerated by the VEGFR-2high CD34– (white columns) and VEGFR-2high

CD34� (black columns) populations. EryD were defined as Hb�-negative, Hb-and Hb�-positive erythrocytes, whereas primitive erythrocytes were Hb-,Hb�-, and Hb�-positive. (V–X): Data represent the mean � SD of triplicatewells, and representative results from one of three independent experiments areshown. Abbreviations: HC, hematopoietic cell; VEGFR, vascular endothelialgrowth factor receptor.


thereafter, constituting up to 20% of erythrocytes, in parallelwith the second wave of erythropoiesis (Fig. 4X). Thus, theVEGFR-2high CD34– and VEGFR-2high CD34� fractions werecapable of both primitive and definitive erythropoiesis, but theVEGFR-2high CD34– cells were less competent to differentiatethan the VEGFR-2high CD34� cells. In contrast, the VEGFR-2low CD34– fraction produced few floating HCs, whereas theVEGFR-2

low or –CD34� and VEGFR-2– CD34– fractions failed to

produce any HCs (data not shown).

EC Development from VEGFR-2high FractionsWe also investigated the capacity of the various fractions sortedat day 6 to differentiate into ECs in the presence of exogenousVEGF. Some GFP-positive cells formed sheet-like or cord-likeclusters that first appeared on the OP9 stromal layer on day 10(Fig. 5A). These clusters took up DiI-Ac-LDL (Fig. 5B) andco-expressed VE-cadherin (Fig. 5C), CD34 (Fig. 5D),VEGFR-2 (Fig. 5E), vWF (Fig. 5F), and CD31 (data notshown), indicating that these cells are ECs. Immunostainingwith a VE-cadherin mAb showed that 6 days after sorting, theVEGFR-2high CD34� fraction generated significantly moreVEGFR-2high VE-cadherin� EC clusters than the VEGFR-2high

CD34– fraction (Fig. 5H). Cluster formation was rare in theVEGFR-2low CD34– or VEGFR-2low or – CD34� fraction, andno clusters were observed in the VEGFR-2– CD34– fraction.Thus, EC production was restricted to the VEGFR-2high cellfractions, and the VEGFR-2high CD34� cells had more angio-genic potential than did the VEGFR-2high CD34– cells.

HC and EC Development from Single VEGFR-2high

CD34– and VEGFR-2high CD34� CellsFinally, we performed a single-cell analysis by using a single-cell deposition system (the Clon-Cyt system) to analyze whetherthe VEGFR-2high CD34– or VEGFR-2high CD34� fractionscontain the common progenitor for both HC and EC lineages.Each well was observed by fluorescence microscopy 24 hoursafter cell deposition, and wells that contained more than oneGFP-positive cell were excluded from further analysis. HC andEC clusters were produced by single cells from the VEGFR-2high CD34– and VEGFR-2high CD34� fractions. Immunostain-

ing with anti-CD45, CD41, Hb�, and VE-cadherin mAbs orby DiI-Ac-LDL incorporation assays confirmed the presenceof HCs or ECs (Fig. 6A– 6C, 6E, 6F). When a mixture ofanti-CD45, CD41, and Hb� mAbs was used for staining wellscontaining HC clusters, all round cells were positive (Fig.6D). The concomitant development of both lineages of cellswas also confirmed by immunostaining with a mixture of thethree hematopoietic lineage mAbs and VE-cadherin mAb orwith the anti-CD34 mAb (Fig. 6G, 6H). When 480 cells fromeach fraction were individually seeded, the potential formono- or bipotential progenitor development was approxi-mately 2-fold higher in the VEGFR-2high CD34� cell popu-lation than the VEGFR-2high CD34� cell population (thefrequencies of HC development alone: 2.5% [12 wells] vs.0.8% [four wells], p � .05; those of EC development alone:15.4% [74 wells] vs. 7.9% [38 wells], p � .05; those of HCplus EC development: 2.3% [11 wells] vs. 1.0% [five wells],p � .13). Nevertheless, the data also strongly suggest thatboth VEGFR-2high fractions contain the common hemoangio-genic progenitors, the “hemangioblasts.”

DISCUSSIONES cells are pluripotent and can differentiate into multiple celltypes, including derivatives of all three germ layers. Althoughtheir high potential for differentiation has been intensively ex-amined in many murine ES cell culture systems [28–32], it isinevitable that the development of particular cells will be con-taminated with that of cells from other lineages. Isolating cellsof interest using FACS is a particularly useful approach toenrich for tissue-specific cells during in vitro ES cell differen-tiation. HC and EC development is a good model for such anapproach because both lineages of cells have many well knowncommon surface markers, as well as lineage-specific antigens[3, 4]. This enhances our ability to select live cells of interest.Indeed, these features have been used to identify the progenitorsof both HC and EC lineages during murine ES cell differenti-ation [8, 31, 32].

The selection of tissue-specific stem cells or progenitors andthe tracing of their fates by further culture are essential forpreclinical research using monkey ES cells that aims to examine

Figure 5. EC development from the VEGFR-2high CD34– and VEGFR-2high CD34� fractions. (A–G): Some green fluorescent protein-positive(GFP�) embryonic stem cell-derived adherent cells formed sheet-like or cord-like endothelial cell (EC) clusters (A) that took up 1,1�-dioctadecyl-1,3,3,3�,3�-tetramethylindocarbocyanine-labeled acetylated low-density lipoprotein (B). Alkaline phosphatase detection of EC clusters stained withantibodies vascular endothelial cadherin (C), CD34 (D), VEGFR-2 (E), and vWF (F). The staining with the isotype control IgG1 is shown in G.Original magnification, �100; scale bars � 100 �m (A–F). (H): The number of EC clusters in cultures of the indicated populations. The datarepresent the mean � SD of triplicate wells, and representative results from one of three independent experiments are shown. Abbreviation: VEGFR,vascular endothelial growth factor receptor.


www.StemCells.com

the efficacy and safety of clinical applications using human EScells [13, 33]. In this study, we have dissected the differentiationpathways by which primate ES cells generate HCs and ECs byanalysis of HC and/or EC cell surface markers. Our results showthat the surface markers associated with HC and EC develop-ment are expressed in a defined order during culture, consistentwith earlier studies of murine ES cell differentiation [8, 31, 32].Furthermore, we show here that it is possible to identify theprogenitors of both lineages and to determine their fates byanalyzing a combination of well known surface markers. Theseobservations will facilitate further investigations on primate EScells, including in vivo studies.

Disruption of the murine homologue of the VEGFR-2(Flk-1) gene in murine embryos or ES cells resulted in a com-bined defect in HC and EC development, which may reflect theloss of a common progenitor, the hemangioblast [5, 6]. Further-more, in vitro differentiation of murine ES cells shows thatVEGFR-2� cells indeed serve as hemangioblasts [7]. In pri-mates, however, there is no direct evidence suggesting thepresence of hemangioblasts during embryogenesis, althoughsome studies have suggested that these cells are present in thefetus and during adulthood [34, 35]. Unlike murine ES cells,several studies have detected VEGFR-2 expression in undiffer-entiated primate ES cells [12, 19, 36–38]. We found that un-differentiated ES cells expressed low levels of VEGFR-2 andthat this surface marker is down-regulated during culture of theES cells in the OP9 coculture system. In addition, whenVEGFR-2low cells were sorted from the cultures on day 6 orfrom undifferentiated ES cells, little HC and EC differentiationwas observed (Figs. 4 and 5). Moreover, markers of the undif-ferentiated state, such as TRA-1–60, Rex-1, and Oct-4, wereexpressed by the VEGFR-2low cells (Fig. 2C, 2D, and data notshown), which suggests that the former fraction still containedundifferentiated components. In contrast, VEGFR-2high cellsemerged on day 6, immediately prior to HC and EC differenti-ation, and we showed here that these cells subsequently gave

rise to both primitive and definitive HCs and ECs. It should benoted that the sorted VEGFR-2high cell populations were notcompletely pure (purity ranged from 93.0%–99.7%). However,it is unlikely that the hemoangiogenic cells came from contam-inating VEGFR-2low or VEGFR-2– cells, since the sortedVEGFR-2low or VEGFR-2– cells were not able to differentiateinto either HC or EC lineages. The results of the single-cellculture assays also strongly suggest that the VEGFR-2high frac-tions contain the common hemoangiogenic progenitors, the he-mangioblasts. Recently, Wang et al. reported the identificationof primitive endothelial-like cells derived from human ES cellsby embryoid body (EB) formation [39]. Their observation thatdevelopment of both lineages can be observed from a single EScell-derived progenitor is in agreement with our own study.Furthermore, such progenitor cells expressed VEGFR-2, but notCD45, as has been observed in mesodermal differentiation ofmurine ES cells [7, 8]. In contrast, the VEGFR-2high hemoan-giogenic progenitors in the report of Wang et al. expressedVE-cadherin, generally considered to be an EC marker [39],whereas the progenitor cells in our study did not. Our resultsdemonstrate that the appearance of hemoangiogenic progenitors,without any HC or EC lineage-specific properties, clearly pre-cedes differentiation into either of these cell lineages. Further-more, this is the first report to demonstrate that both primitiveand definitive HCs, as well as ECs, were generated in theVEGFR-2high fractions. These differences in hematopoietic andendothelial differentiation may be partially due to differences inthe culture conditions (the EB and OP9 coculture system),and/or in the ES cells that were used for these studies.

Sequential FACS analysis with a combination of surfacemarkers revealed that two distinct populations, VEGFR-2high

CD34– and VEGFR-2high CD34� cells, were present on day 6.The potential of the VEGFR-2high CD34� cell to serve as amono- or bipotential progenitor is approximately twice that ofthe VEGFR-2high CD34– cell, although both cell types produceequal proportions of HCs and ECs. Notably, when we analyzed

Figure 6. Single-cell deposition assay showing the hemoangiogenic potential of VEGFR-2high CD34– and VEGFR-2high CD34� cells. (A–D):VEGFR-2high CD34– and VEGFR-2high CD34� cells were sorted, and single cells were seeded onto an OP9 stromal layer in 96-well plates. Six dayslater, hematopoietic cell (HC) development was evaluated by immunostaining with monoclonal antibodies CD45 (A), CD41 (B), or Hb� (C). (D):When a mixture of all these monoclonal antibodies was used for staining wells containing HC clusters, all of them were positive. (E, F): ECdevelopment was evaluated by immunostaining with vascular endothelial cadherin mAb (E) or by the 1,1�-dioctadecyl-1,3,3,3�, 3�-tetramethylindo-carbocyanine-labeled acetylated low-density lipoprotein incorporation assay (F). (G, H): Concomitant development was confirmed by a mixture ofCD45, CD41, and Hb� (blue)/vascular endothelial cadherin (brown) double immunostaining (G) or immunostaining with CD34 mAb (H). Originalmagnification, �100; scale bars � 100 �m.


the expression of HC and/or EC lineage marker genes by theVEGFR-2high CD34– and VEGFR-2high CD34� cell popula-tions on day 6, we found that only the VEGFR-2high CD34�

cells expressed SCL. Scl/ murine embryos show lack ofblood formation and a defect in yolk sac angiogenesis, indicat-ing that this transcription factor is essential for HC and ECdevelopment [40–42]. Furthermore, recent reports on the de-velopmental kinetics of VEGFR-2 and SCL suggest thatVEGFR-2� SCL� cells may be hemangioblasts [43, 44]. Theseobservations together suggest that CD34 is expressed by theVEGFR-2high cells during their differentiation into hemoangio-genic progenitors, concomitant with an upregulation of a set offactors that regulate the development of both lineages.

Recent studies, including our previous work, have reportedthat exogenous VEGF enhances early HC development [12, 45].We also observed that ECs are generated more abundantly in thepresence of VEGF (unpublished data). Unlike other reports withmonkey or human ES cells [37, 39, 45], in our culture system,exogenous BMP-4 fails to induce hematopoietic differentiation,probably because it causes the OP9 stromal cells to differentiateand thereby impairs their interaction with ES cells [12]. Weanalyzed the effect of various concentrations of VEGF on thedevelopment of VEGFR-2high cells by FACS and found that itincreases the proportion of CD34� cells in the VEGFR-2high

cell population in a dose-dependent manner. Taken together, theeffect of VEGF on HC and EC development is mainly due to its

ability to enhance the proliferation and/or differentiation ofVEGFR-2high CD34� cells with a higher hemoangiogenic po-tential during the initial 6-day differentiation induction.

In summary, we have been able to identify and characterizehemoangiogenic progenitors by sequential phenotypic analysisduring primate ES differentiation. Our observations after cellsorting strongly suggest that the VEGFR-2high fraction of cellscontains hemangioblasts. The approach we have taken in thisstudy will contribute to investigations of early developmentalsteps in human biology and, in addition, will provide a cellsource for regenerative medicine applications in the future.

ACKNOWLEDGMENTSWe thank Tanabe Seiyaku Co. Ltd. (Osaka, Japan) for help inpreparing the primate ES cells. This work was supported bygrants from the Science Research on Priority Areas and theCreative Science Research programs. It was also supported bythe Japan Society for the Promotion of Science; by the Ministryof Education, Culture, Sports, Science and Technology of Japan;and by the program for Promotion of Fundamental Studies inHealth Sciences of the National Institute of Biomedical Inno-vation of Japan.

DISCLOSURESThe authors indicate no potential conflicts of interest.

REFERENCES

1 Sabin FR. Studies on the origin of blood vessels and of red corpuscles asseen in the living blastoderm of the chick during the second day ofincubation. Contrib Embryol 1920;9:213–262.

2 Murray PDF. The development in vitro of the blood of the early chickembryo. Proc Natl Acad Sci U S A 1932;68:497–520.

3 Wood HB, May G, Healy L et al. CD34 expression patterns during earlymouse development are related to modes of blood vessel formation andreveal additional sites of hematopoiesis. Blood 1997;90:2300–2311.

4 Garcia-Porrero JA, Manaia A, Jimeno J et al. Antigenic profiles ofendothelial and hemopoietic lineages in murine intraembryonic hemo-genic sites. Dev Comp Immunol 1998;22:303–319.

5 Shalaby F, Rossant J, Yamaguchi TP et al. Failure of blood-islandformation and vasculogenesis in Flk-1-deficient mice. Nature 1995;376:62–66.

6 Shalaby F, Ho J, Stanford WL et al. A requirement for Flk1 in primitiveand definitive hematopoiesis and vasculogenesis. Cell 1997;89:981–990.

7 Choi K, Kennedy M, Kazarov A et al. A common precursor for hema-topoietic and endothelial cells. Development 1998;125:725–732.

8 Nishikawa SI, Nishikawa S, Hirashima M et al. Progressive lineageanalysis by cell sorting and culture identifies FLK1�VE-cadherin� cellsat a diverging point of endothelial and hemopoietic lineages. Develop-ment 1998;125:1747–1757.

9 Tavian M, Coulombel L, Luton D et al. Aorta-associated CD34� hema-topoietic cells in the early human embryo. Blood 1996;87:67–72.

10 Labastie MC, Cortes F, Romeo PH et al. Molecular identity of hemato-poietic precursor cells emerging in the human embryo. Blood 1998;92:3624–3635.

11 Marshall CJ, Moore RL, Thorogood P et al. Detailed characterization ofthe human aorta-gonad-mesonephros region reveals morphological po-larity resembling a hematopoietic stromal layer. Dev Dyn 1999;215:139–147.

12 Umeda K, Heike T, Yoshimoto M et al. Development of primitive anddefinitive hematopoiesis from nonhuman primate embryonic stem cellsin vitro. Development 2004;131:1869–1879.

13 Suemori H, Tada T, Torii R et al. Establishment of embryonic stem celllines from Cynomolgus monkey blastocysts produced by IVF or ICSI.Dev Dyn 2001;222:273–279.

14 Furuya M, Yasuchika K, Mizutani KI et al. Electroporation of cynomol-gus monkey embryonic stem cells. Genesis 2003;37:180–187.

15 Kannagi R, Cochran NA, Ishigami F et al. Stage-specific embryonicantigens (SSEA-3 and -4) are epitopes of unique globo-series gangliosideisolated from human teratocarcinoma cells. EMBO J 1983;2:2355–2361.

16 Luo HY, Liang XL, Frye C et al. Embryonic hemoglobins are expressedin definitive cells. Blood 1999;94:359–361.

17 Sawano A, Iwai S, Sakurai Y et al. Flt-1, vascular endothelial growthfactor receptor 1, is a novel cell surface marker for the lineage ofmonocyte-macrophages in humans. Blood 2001;97:785–791.

18 Yoshino N, Ami Y, Terao K et al. Upgrading of flow cytometric analysisfor absolute counts, cytokines and other antigenic molecules of cyno-molgus monkeys (Macaca fascicularis) by using anti-human cross-reac-tive antibodies. Exp Anim 2000;49:97–110.

19 Sone M, Itoh H, Yamashita J et al. Different differentiation kinetics ofvascular progenitor cells in primate and mouse embryonic stem cells.Circulation 2003;107:2085–2088.

20 Suwabe N, Takahashi S, Nakano T et al. GATA-1 regulates growth anddifferentiation of definitive erythroid lineage cells during in vitro ES celldifferentiation. Blood 1998;92:4108–4118.

21 Rosler ES, Fisk GJ, Ares X et al. Long-term culture of human embryonicstem cells in feeder-free conditions. Dev Dyn 2004;229:259–274.

22 Orkin SH, Zon LI. Genetics of erythropoiesis: Induced mutations in miceand zebrafish. Annu Rev Genet 1997;31:31–60.

23 Breier G. Endothelial receptor tyrosine kinases involved in blood vesseldevelopment and tumor angiogenesis. Adv Exp Med Biol 2000;476:57–66.

24 Venugopal SK, Devaraj S, Yuhanna I et al. Demonstration that C-reac-tive protein decreases eNOS expression and bioactivity in human aorticendothelial cells. Circulation 2002;106:1439–1441.


www.StemCells.com

25 Ema M, Faloon P, Zhang WJ et al. Combinatorial effects of Flk1 andTal1 on vascular and hematopoietic development in the mouse. GenesDev 2003;17:380–393.

26 Casella I, Feccia T, Chelucci C et al. Autocrine-paracrine VEGF loopspotentiate the maturation of megakaryocytic precursors through Flt1receptor. Blood 2003;101:1316–1323.

27 Kobayashi-Osaki M, Ohneda O, Suzuki N et al. GATA motifs regulateearly hematopoietic lineage-specific expression of the Gata2 gene. MolCell Biol 2005;25:7005–7020.

28 Doestchmann TC, Eistetter H, Katz M et al. The in vitro development ofblastocyst-derived embryonic stem cell lines: Formation of visceral yolksac, blood islands and myocardium. J Embryol Exp Morphol 1985;87:27–45.

29 Wiles MV, Keller G. Multiple hematopoietic lineage develop from embry-onic stem (ES) cells in culture. Development 1991;111:259 –267.

30 Keller G, Kennedy M, Papayannopoulou T et al. Hematopoietic com-mitment during embryonic stem cell differentiation in culture. Mol CellBiol 1993;13:473–486.

31 Hirashima M, Kataoka H, Nishikawa S et al. Maturation of embryonicstem cells into endothelial cells in an in vitro model of vasculogenesis.Blood 1999;93:1253–1263.

32 Fujimoto T, Ogawa M, Minegishi N et al. Step-wise divergence of primitiveand definitive haematopoietic and endothelial cell lineages during embry-onic stem cell differentiation. Genes Cells 2001;6:1113–1127.

33 Hanazono Y, Terao K, Ozawa K. Gene transfer into nonhuman primatehematopoietic stem cells: Implications for gene therapy. STEM CELLS2001;19:12–23.

34 Pelosi E, Valtieri M, Coppola S et al. Identification of the hemangioblastin postnatal life. Blood 2002;100:3203–3208.

35 Guo H, Fang B, Liao L et al. Hemangioblastic characteristics of fetalbone marrow-derived Flk-1�CD31– CD34– cells. Exp Hematol 2003;31:650–658.

36 Kaufman DS, Hanson ET, Lewis RL et al. Hematopoietic colony-forming cells derived from human embryonic stem cells. Proc Natl AcadSci U S A 2001;98:10716–10721.

37 Li F, Lu S, Vida L et al. Bone morphogenetic protein 4 induces efficienthematopoietic differentiation of rhesus monkey embryonic stem cells invitro. Blood 2001;98:335–342.

38 Levenberg S, Golub JS, Amit M et al. Endothelial cells derived fromhuman embryonic stem cells. Proc Natl Acad Sci U S A 2002;99:4391–4396.

39 Wang L, Li L, Shojaei F et al. Endothelial and hematopoietic cell fate ofhuman embryonic stem cells originates from primitive endothelium withhemangioblastic properties. Immunity 2004;21:31–41.

40 Robb L, Lyons I, Li R et al. Absence of yolk sac hematopoiesis frommice with a targeted disruption of the scl gene. Proc Natl Acad Sci U S A1995;92:7075–7079.

41 Shivdasani RA, Mayer EL, Orkin SH. Absence of blood formation inmice lacking the T-cell leukaemia oncoprotein tal-1/SCL. Nature 1995;373:432–434.

42 Visvader JE, Fujiwara Y, Orkin SH. Unsuspected role for the T-cellleukemia protein SCL/tal-1 in vascular development. Genes Dev 1998;12:473–479.

43 Chung YS, Zhang WJ, Arentson et al. Lineage analysis of the heman-gioblast as defined by FLK1 and SCL expression. Development 2002;129:5511–5520.

44 D’Souza SL, Elefanty AG, Keller G. SCL/Tal-1 is essential for hema-topoietic commitment of the hemangioblast but not for its development.Blood 2005;105:3862–3870.

45 Cerdan C, Rouleau A, Bhatia M. VEGF-A165 augments erythropoieticdevelopment from human embryonic stem cells. Blood 2004;103:2504–2512.


Documents

Identification and Characterization of Hemoangiogenic Progenitors During Cynomolgus Monkey Embryonic Stem Cell Differentiation