Differentiation of Human Embryonic Stem Cells in Serum-Free Medium Reveals Distinct Roles for Bone Morphogenetic Protein 4, Vascular Endothelial Growth Factor, Stem Cell Factor, and Fibroblast Growth Factor 2 in Hematopoiesis

Authors


Abstract

We have utilized a serum- and stromal cell-free “spin embryoid body (EB)” differentiation system to investigate the roles of four growth factors, bone morphogenetic protein 4 (BMP4), vascular endothelial growth factor (VEGF), stem cell factor (SCF), and basic fibroblast growth factor (FGF2), singly and in combination, on the generation of hematopoietic cells from human embryonic stem cells (HESCs). Of the four factors, only BMP4 induced expression of genes that signaled the emergence of the primitive streak-like population required for the subsequent development of hematopoietic mesoderm. In addition, BMP4 initiated the expression of genes marking hematopoietic mesoderm and supported the generation of hematopoietic progenitor cells at a low frequency. However, the appearance of robust numbers of hematopoietic colony forming cells and their mature progeny required the inclusion of VEGF. Finally, the combination of BMP4, VEGF, SCF, and FGF2 further enhanced the total yield of hematopoietic cells. These data demonstrate the utility of the serum-free spin EB system in dissecting the roles of specific growth factors required for the directed differentiation of HESCs toward the hematopoietic lineage.

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

Introduction

Differentiated cells of the hematopoietic and endothelial lineages represent the earliest mesodermal derivatives that emerge during gastrulation in the vertebrate embryo, when cells from the epiblast migrate through the primitive streak and give rise to mesoderm and definitive endoderm [1, [2]–3]. Studies from our laboratory and others have demonstrated that in vitro differentiation of embryonic stem cells recapitulates aspects of this developmental sequence [4, [5], [6], [7]–8]. In a step wise manner, differentiating ESCs express the same genes that are associated with early postimplantation embryonic stages and, predictably, the earliest hematopoietic precursors generated from mouse ESCs emerge shortly after the onset of the in vitro primitive streak-like stage [7, 9].

Several laboratories have derived hematopoietic precursors and their progeny from differentiating human embryonic stem cells (HESCs) utilizing a variety of culture conditions [10, [11], [12], [13], [14], [15], [16]–17]. These studies employed serum-containing medium and/or coculture with stromal/feeder cell layers, such as the murine bone marrow cell line S17, the yolk sac endothelial cell line C166, murine colony-stimulating factor 1-null OP9 cells, or human bone marrow stroma [10, 12, 15, [16]–17]. In instances where the contributions of growth factors and cytokines to hematopoiesis have been investigated, the observed effects may have been contributed to by inductive or inhibitory influences from fetal calf serum or from the presence of stromal cells [10, 11, 13, 14].

Hematopoietic cells generated from the in vitro differentiation of HESCs represent a potential alternative supply of blood cells and blood products, which are currently sourced from an increasingly restricted pool of eligible blood donors. In order for this potential to be realized, it will be necessary to define medium components and protocols for the efficient directed differentiation of HESCs toward the hematopoietic lineages. Furthermore, if HESC-derived hematopoietic cells are to reach the clinic, it will be important that the transplanted cells are generated in the absence of serum or stromal cells in order to guarantee freedom from contamination by adventitious animal or human pathogens.

Our laboratory has taken the initial steps toward this goal, recently reporting a serum- and stromal cell-free method (“spin embryoid bodies [EBs]”) in which a combination of cytokines was used to induce robust and reproducible hematopoietic differentiation from HESCs [18]. Here, we have exploited this system to analyze the roles of four cytokines, bone morphogenetic protein 4 (BMP4), vascular endothelial growth factor (VEGF), stem cell factor (SCF), and basic fibroblast growth factor (FGF2), on the early hematopoietic differentiation of HESCs. These factors were evaluated because these molecules or their receptors have been shown to be necessary for mesoderm patterning or blood cell formation during embryogenesis [19, [20], [21], [22], [23], [24], [25], [26]–27].

Our data show that BMP4 was sufficient to induce the expression of primitive streak genes and genes marking hematopoietic mesoderm but that the addition of VEGF was required for the efficient generation of hematopoietic progenitor cells. The addition of SCF and FGF2 to this factor combination further enhanced the total yield of hematopoietic cells in our serum- and stromal cell-free HESC differentiation system.

Materials and Methods

HESC Culture and Differentiation

Human embryonic stem cells (HES3) [28] were maintained on irradiated primary mouse embryonic fibroblast (PMEF) feeder cells in Dulbecco's modified Eagle's medium/Hams F12 medium (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) supplemented with 20% knockout serum replacer (Invitrogen) and recombinant human (rhu)FGF2 (4 ng/ml; Peprotech, Rocky Hill, NJ, http://www.peprotech.com) by mechanical passaging. In order to generate cells for differentiation experiments, colonies were expanded onto tissue culture flasks preseeded with irradiated PMEFs at a density of 2 × 104 per cm2 using the same medium as above with 8 ng/ml of rhuFGF2 and passaged twice weekly, disaggregating cells with 0.25% trypsin-EDTA (Invitrogen)/2% chicken serum (Serum Australis, Manilla, NSW, Australia, http://www.serumaustralis.com.au) as previously described [18]. To prevent the emergence of karyotypically abnormal clones, cultures were only maintained as bulk cultures for 15–25 passages. The karyotypes of HESC lines used in these experiments were regularly monitored, and no abnormalities were detected.

The day before differentiation, HESCs were passaged onto tissue culture flasks seeded with irradiated PMEFs at low density (1 × 104 per cm2). To initiate differentiation, the cultures were harvested with trypsin-EDTA or TrypLE Select (Gibco, Grand Island, NY, http://www.invitrogen.com), washed with phosphate-buffered saline (PBS), and centrifuged (480g for 3 minutes). Following centrifugation, the pellet of single cells was resuspended in a serum-free chemically defined medium (CDM) initially described by Johansson et al. [29] and modified as described [18]. The remaining incompletely defined components in this medium that still contain animal or nonrecombinant products are bovine serum albumin (A3311, mouse embryo tested; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), insulin-transferrin-selenium (ITS-X; Gibco), and synthetic lipids (Gibco). One hundred μl of CDM containing 2,500 HESCs supplemented with growth factors was added per well in low attachment 96-well round-bottom plates (Nunc, Rochester, NY, http://www.nuncbrand.com) and centrifuged at 480g for 5 minutes to induce aggregation and placed in a humidified incubator (5% CO2 in air at 37°C). CDM was supplemented with the following recombinant human growth factors singly or in combination: rhuBMP4, 1–50 ng/ml; rhuVEGF-A, 10 ng/ml; rhuSCF, 25 ng/ml; and rhuFGF2, 10 ng/ml (from R&D Systems Inc., Minneapolis, http://www.rndsystems.com, or Peprotech). Recombinant factor concentrations were determined by titrating the doses of factors against an experimental outcome, generally using flow cytometric evaluation of cell surface marker expression or intracellular flow cytometry for MIXL1 expression (data not shown).

Within 24 hours in culture, embryoid bodies formed in each well. After 10 days, the EBs were transferred to 96-well adherent flat-bottom plates containing 100 μl of CDM supplemented with recombinant human thrombopoietin at 20 ng/ml and rhuSCF at 25 ng/ml (R&D Systems or Peprotech). EBs were harvested for analysis at different time points, disaggregated with trypsin-EDTA or TrypLE Select, and passed through a 23-guage needle to generate a single cell suspension.

Flow Cytometric Analysis of Differentiated HESCs

The following antibodies were used in combination for flow cytometric analysis of the harvested disaggregated EBs: (a) CD45-fluorescein isothiocyanate (FITC), CD34-phycoerythrin (PE), CD41a-allophycocyanin (APC); (b) CD34-FITC, CD15 PE, CD33-APC; and (c) CD31-FITC, CD117-PE, CD34-APC (BD Biosciences, San Diego, http://www.bdbiosciences.com). Antibodies were added according to predetermined optimal concentrations and incubated for 30 minutes at 4°C. Cells were washed in PBS and centrifuged at 480g for 5 minutes at 4°C, and cell pellets were resuspended in PBS containing propidium iodide (to exclude nonviable cells) before analysis on a FACSCalibur (BD Biosciences). Overall, the viability of differentiating HESCs was above 70% in all BMP4-containing media.

Hematopoietic Colony Assay of Differentiated HESCs

Triplicate assays were performed in 24-well tissue culture-treated plates (Nunc) with 10,000–50,000 HESCs added per well in 0.5 ml of MethoCult (Stem Cell Technologies, Vancouver, BC, Canada, http://www.stemcell.com) supplemented with the following recombinant human growth factors: granulocyte-macrophage–colony stimulating factor at 20 ng/ml, SCF at 50 ng/ml, interleukin (IL)-3 at 20 ng/ml, erythropoietin at 3 U/ml, and IL-6 at 20 ng/ml (from R&D systems and Peprotech). Plates were incubated at 37°C in a humidified atmosphere in a 5% CO2 incubator and scored for colony formation between 10 and 14 days.

Quantitative Real-Time Polymerase Chain Reaction

Total RNA from undifferentiated and differentiated HESCs was prepared using RNeasy reagents according to the manufacturer's instructions (Qiagen, Hilden, Germany, http://www1.qiagen.com). First-strand cDNA was reverse transcribed with random hexamer priming using Superscript III reagents (Invitrogen). Quantitative real-time polymerase chain reaction was performed using TaqMan gene expression probes and TaqMan reagents purchased from Applied Biosystems (Foster City, CA, http://www.appliedbiosystems.com) and the 7500 Fast Real-Time PCR System absolute thermal cycler and software (Applied Biosystems). The comparative cycle threshold (Ct) method was used to analyze data, with gene expression levels compared with a reference gene. Initially, three reference genes were compared, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), UBIQUITIN C, and hypoxanthine phosphoribosyl transferase (HPRT), in response to recent studies indicating that commonly used reference genes could exhibit unacceptably large variations in expression among different tissue types and between undifferentiated and differentiated embryonic stem (ES) cells [30, [31]–32]. As shown in supplemental online Figure 1A–1D, we analyzed 38 of our cDNA samples comparing GAPDH, UBIQUITIN C, and HPRT as alternative reference genes. These data show that GAPDH is a far more sensitive measure of amplifiable cDNA than UBIQUITIN C (∼11.5-fold more sensitive) or HPRT (∼135-fold more sensitive) for our experiments (Fig. S1A). There was an excellent correlation between threshold cycle numbers among GAPDH, UBIQUITIN C, and HPRT, with the best correlation (r2 = .9638) between GAPDH and UBIQUITIN C. This correlation was maintained across a wide range of cDNA input concentrations. Indeed, the samples analyzed spanned a range of >3,400-fold in relative cDNA quantity from the most concentrated (GAPDH Ct = 16.55) to the most dilute (GAPDH Ct = 28.30) (Fig. S1B). When we compared the expression of differentiation specific genes normalized against each of the reference genes, no significant differences were discernible among results obtained using GAPDH, UBIQUITIN C, or HPRT (Fig. S1C, S1D). For these experiments, the Ct for expression was calculated for each gene and for the relevant reference gene (REF). Since gene expression is inversely proportional to the Ct, the expression for a given target gene relative to REF may be given by the formula:

equation image

In this paper, this calculated value was multiplied by 10 for samples normalized to HPRT, by 100 for samples normalized to UBIQUITIN C, or 1,000 for samples normalized to GAPDH for the purposes of presentation. For these reasons, and because of its greater sensitivity, real-time polymerase chain reaction (PCR) data shown in the body of the manuscript were normalized using GAPDH as the reference gene.

Results

BMP4 Induces the Primitive Streak Gene MIXL1 in Differentiating HESCs

Four growth factors (BMP4, FGF2, VEGF, and SCF) that have been implicated in the differentiation of HESCs to hematopoietic lineages were evaluated singly and in combination. Initial experiments focused on the ability of single factors to induce the transcription of MIXL1 [33, 34], a gene that is expressed in the embryonic primitive streak, a transient structure harboring precursors of mesodermal and endodermal lineages that is an obligate intermediate for blood cell formation [2]. In the absence of growth factors, or in medium supplemented by VEGF, FGF2, or SCF, no increase in MIXL1 transcription was observed during HESC differentiation (Fig. 1A and supplemental online Fig. S1). In contrast, supplementing the serum-free medium with BMP4 increased both the viability and number of cells in the EBs and promoted cyst formation, consistent with its ability to improve viability in differentiating mouse ESCs [7] (Fig. 1B) and led to the transient induction of MIXL1 mRNA (Fig. 1A and supplemental online Fig. S1). Cells expressing MIXL1 protein were detected by flow cytometry only in cultures containing BMP4. Furthermore, these cells coexpressed OCT4, consistent with the phenotype of a primitive streak-like population that we have described previously (Fig. 1C) [35].

Figure Figure 1..

BMP4 induces MIXL1 expression in differentiating human embryonic stem cells (HESCs). (A): MIXL1 expression detected by quantitative real-time polymerase chain reaction in HESCs differentiated in serum-free cultures supplemented with no growth factors, 10 ng/ml VEGF, 25 ng/ml SCF, 10 ng/ml FGF2, or 10 ng/ml BMP4. MIXL1 gene expression relative to GAPDH as a reference gene is shown on the vertical axis. Similar results were obtained using UBIQUITIN C or HPRT as reference genes. (See Materials and Methods and supplemental online Fig. S1 for more details.) Results shown are the mean ± SEM (n = 4, asterisk, p < .03 for BMP4 vs. other conditions). (B): HESC spin embryoid bodies (EBs) at day 10 of differentiation demonstrate increased EB size and cyst formation in the presence of BMP4 (×50 magnification). (C): Percentage of cells expressing OCT4 and both OCT4 and MIXL1 proteins (OCT4/MIXL1) in day 6 HESC cultures supplemented with no GF, VEGF, SCF, FGF2, or BMP4. The results shown are the mean ± SD (n = 6, asterisk, p < .01 for BMP4 vs. other conditions); 97.5% ± 1.7% of the undifferentiated HESCs expressed OCT4 and 2.1% ± 0.2% also expressed MIXL1. Abbreviations: BMP4, bone morphogenetic protein 4; FGF2, basic fibroblast growth factor 2; GF, growth factors; SCF, stem cell factor; VEGF, vascular endothelial growth factor.

BMP4 Is Sufficient to Induce Primitive Streak and Early Hematopoietic Mesoderm Gene Expression

We extended these studies to examine whether other primitive streak or hematopoietic genes were also induced by BMP4 stimulation. Cultures of differentiating HESCs were exposed to increasing concentrations of BMP4 in the presence or absence of the KDR ligand, VEGF, and gene expression was examined after 3 and 5 days. VEGF was chosen because it is absolutely required for blood cell formation in the mouse embryo [23, 36] and we had previously documented that BMP4 induced Flk1 (the mouse ortholog of KDR) in mouse ESCs [7]. These experiments confirmed that BMP4 induced expression of the primitive streak genes MIXL1, BRACHYURY, and GOOSECOID (Fig. 2A–2C and supplemental online Fig. S2) and that the levels of expression and kinetics of induction were not influenced by concomitant exposure to VEGF. Furthermore, the expression of two genes associated with development of hematopoietic mesoderm, GATA2 and RUNX1, was also induced by BMP4 and was independent of VEGF (Fig. 2D, 2E). As described for differentiating mouse ESCs, BMP4 also induced expression of KDR (Fig. 2F). High levels of KDR expression during the early phases of differentiation were consistent with expression of KDR protein detected by flow cytometry on undifferentiated HESCs (data not shown). However, in the case of the stem cell marker CD34 and the hematopoietic transcription factor SCL/TAL 1, the addition of VEGF to BMP4 cultures enhanced gene expression at day 5 of differentiation (Fig. 2G, 2H). These data argued that VEGF contributed to the survival, expansion, and/or further differentiation of the early hematopoietic mesoderm induced by BMP4. Consistent with this hypothesis, immunophenotypic analysis of EBs differentiated for 20 days revealed that the percentage of hematopoietic cells increased in a BMP4 dose-dependent manner and that the inclusion of VEGF acted synergistically to further increase hematopoietic cell frequency by greater than fourfold (supplemental online Fig. S3).

Figure Figure 2..

Bone morphogenetic protein 4 (BMP4) induces primitive streak and hematopoietic mesoderm genes in a dose-dependent manner (A–H). Differentiating human ESC cultures were supplemented with 1, 10, or 50 ng/ml of BMP4 (B1, B10, B50) ± 10 ng/ml VEGF (V10), and expression of the indicated genes relative to glyceraldehyde-3-phosphate dehydrogenase (see Materials and Methods and supplemental online Fig. S1) was analyzed by quantitative real-time polymerase chain reaction after 3 and 5 days. Expression profiles for undifferentiated cells and cells differentiated in VEGF alone or in the absence of added growth factors are also indicated. The experiment shown is representative of three independent experiments performed (see supplemental online Fig. S2A, S2B for additional experimental results). Abbreviations: B, bone morphogenetic protein 4; GF, growth factors; Undiff, undifferentiated cells; V, vascular endothelial growth factor.

Consistent with a prior study that reported the induction of trophectoderm by BMP4 [37], we also observed BMP4 dependent induction of expression for chorionic gonadotropin (CGb) after 5 days (supplemental online Fig. S4). Expression of another trophectoderm gene, luteinizing hormone (LH), also increased slightly with differentiation, but this was not BMP4 dependent. Since expression of both trophectoderm genes clearly postdated the induction of primitive streak genes (compare with Fig. 2, expression of BRACHYURY, MIXL1, and GOOSECOID), it is likely that the primitive streak inducing effects of BMP4 were direct and not via a trophectoderm intermediate.

We subsequently compared the expression of GATA2, RUNX1, CD34, and SCL in HESCs when the cells were differentiated in cultures supplemented with single factors or combinations of BMP4, VEGF, SCF, and FGF2. Analysis of gene expression by real-time PCR showed very low levels of GATA2, RUNX1, CD34, or SCL in undifferentiated HESCs (Fig. 3). In experiments analyzed at day 11 of differentiation, similar levels of GATA2 and RUNX1 expression were observed in cultures supplemented either by BMP4 or by any of the BMP4-containing factor combinations tested (Fig. 3). In contrast, much lower levels were observed in the cultures differentiated in the other single factors. As anticipated by the studies shown in Figure 2, levels of CD34 and SCL in the presence of BMP4 were intermediate between the very low levels seen in VEGF, SCF, or FGF2 stimulated cultures and the higher expression in the BMP4-containing multifactor combinations, although the differences did not reach statistical significance (Fig. 3). When we examined early neurectodermal gene expression, we observed significant induction of SOX1 expression by day 10 in the absence of growth factor stimulation, consistent with the default neural differentiation of ES cells (supplemental online Fig. S5). Although similar levels of SOX1 induction were observed in cultures supplemented with VEGF, SCF, or FGF2, the inclusion of BMP4 overrode the neural induction, and high levels of SOX1 were not observed in any BMP4 containing medium. In our cultures, PAX6 was only expressed at very low levels in all the factor combinations.

Figure Figure 3..

Expression of hematopoietic genes in differentiating human embryonic stem cells (HESCs). Expression of GATA2, RUNX1, CD34, and SCL detected by quantitative real-time polymerase chain reaction in undifferentiated HESCs and HESCs differentiated for 11 days in serum-free cultures supplemented with BMP4 (10 ng/ml), VEGF (10 ng/ml), SCF (25 ng/ml), and FGF2 (10 ng/ml) singly or in the indicated combinations (mean ± SEM, n = 3). Abbreviations: B, bone morphogenetic protein 4; BV, bone morphogenetic protein 4 and vascular endothelial growth factor; BVF, bone morphogenetic protein 4, vascular endothelial growth factor, and basic fibroblast growth factor; BVS, bone morphogenetic protein 4, vascular endothelial growth factor, and stem cell factor; BVSF, bone morphogenetic protein 4, vascular endothelial growth factor, stem cell factor, and basic fibroblast growth factor; F, basic fibroblast growth factor; noGF, no growth factors; S, stem cell factor; Undiff HESC, undifferentiated human embryonic stem cell; V, vascular endothelial growth factor.

FGF2 Increases Cell Yield During HESC Differentiation

The influence of the different cytokine combinations on cell yield during differentiation was examined. After 2 days, cell numbers were higher in BMP4-VEGF-SCF-FGF2 (BVSF) medium compared with BMP4-VEGF (BV) medium (p < .02 at day 2, p < .03 at day 3, Fig. 4A). Analysis after 5 days of differentiation showed that both BVSF and BMP4-VEGF-FGF2 (BVF) media generated over twice as many cells as BV or BMP4-VEGF-SCF (BVS) media (BVSF vs. BV and BVS, p ≤ .05 from day 5; BVF vs. BV and BVS, p ≤ .05 from day 10, Fig. 4B). There was a tendency for the BVSF combination to give higher total cell counts than BVF, but this difference did not reach statistical significance. There was no difference in viability observed among the cytokine combinations tested (data not shown), raising the possibility that the differences in final cell number reflected differences in cell proliferation rather than cell survival.

Figure Figure 4..

Basic fibroblast growth factor (FGF2) increases cell yield during human embryonic stem cell (HESC) differentiation. Viable cell counts of differentiating HESCs (using trypan blue exclusion) were calculated at the indicated day by harvesting embryoid bodies from a plate of 72 wells initially seeded with 2.5 × 103 cells per well (i.e., a starting number of 180,000 cells per plate) in the presence of the indicated growth factors. (A): Increased cell numbers were observed in BVSF medium from day 2 of differentiation (n = 5, p values as shown for BVSF vs. BV). (B): Over a longer period of observation, the yield of differentiated cells was higher in factor combinations containing FGF2 (BVSF vs. BV and BVS, p < .05 from day 5; BVF vs. BV and BVS, p < .05 from day 10; n = 15). Abbreviations: BV, bone morphogenetic protein 4 and vascular endothelial growth factor; BVF, bone morphogenetic protein 4, vascular endothelial growth factor, and basic fibroblast growth factor; BVS, bone morphogenetic protein 4, vascular endothelial growth factor, and stem cell factor; BVSF, bone morphogenetic protein 4, vascular endothelial growth factor, stem cell factor, and basic fibroblast growth factor.

The Combination of BMP4, VEGF, SCF, and FGF2 Generated the Greatest Number of Hematopoietic Cells from Differentiating HESCs

In order to correlate the growth factors required for expression of hematopoietic genes with those required for the generation of hematopoietic progenitors, the frequency of colony-forming cells (CFCs) in cultures supplemented with BMP4 alone was compared with the CFC frequency observed in the factor combinations. Whereas HESCs differentiated for 10 days in the presence of BMP4 formed colonies in methylcellulose at a frequency of 29.5 ± 11 (mean ± SEM) CFCs per 105 cells, differentiating HESCs cultured in VEGF-containing combinations consistently generated colonies at a three- to fourfold higher frequency (Fig. 5A, 5B). When the total yield of CFCs obtained per plate of EBs was compared for the various growth factor combinations, it was evident that the inclusion of FGF2 in the BVF and BVSF resulted in a greater total number of colonies than was observed in the BV and BVS conditions at day 10 (Fig. 5C, 5D) and day 20 of differentiation (Fig. 5E–5H). The hematopoietic colony yield was always higher in BVSF than in BVF medium, but this difference did not reach significance until day 20 of differentiation (Fig. 5H). Colony morphology and stained cytospin preparations of individual picked colonies indicated the presence of similar proportions of mixed, macrophage, and erythroid colony types cultured under all permissive conditions (Fig. 5I–5K and supplemental online Table S6).

In order to further dissect the role of BMP4 alone versus the BV factor combinations in the generation of hematopoietic cells, flow cytometry was used to characterize the hematopoietic composition of the differentiating HESC populations. Cultures supplemented with BMP4 alone (or any other single cytokine tested) failed to produce substantial percentages of CD34+ (progenitor and endothelial cell marker) cells at day 10 or CD45+ (leukocyte marker) or CD33+ (myeloid cell marker) cells by day 20 (Fig. 6A–6C and supplemental online Fig. S3). On average, a 10-fold increase in the proportion of CD34+, CD45+, and CD33+ cells were observed in all medium combinations containing BMP4 and VEGF (Fig. 6A–6C). Although, in the example shown, BV medium did not increase the percentage of CD45+ and CD33+ cells to the same extent as the other cytokine combinations, over a number of experiments there was no statistically significant difference in the percentage of CD45+ and CD33+ cells generated among the four multifactor combinations (Fig. 6A–6C). When the total number of CD34+, CD45+, and CD33+ cells generated was calculated, it was observed that the combination of BVSF in the culture medium yielded the greatest number of hematopoietic cells, with an average of over 800 CD45+ cells generated per EB (Fig. 6D–6F).

Figure Figure 5..

The combination of bone morphogenetic protein 4 (BMP4), vascular endothelial growth factor, stem cell factor, and basic fibroblast growth factor is required for efficient hematopoietic colony formation from human ESCs. (A, E): The frequency of hematopoietic CFCs (mean ± SEM, n = 5) in the presence of BMP4 and/or the indicated factor combinations from day 10 and day 20 cultures. (C, G): The total number of hematopoietic CFCs (mean ± SEM, n = 5) per plate (72 wells initially seeded at 2.5 × 103 cells per well) generated from day 10 and day 20 cultures in the presence of BMP4 and/or the indicated factor combinations. (B, D): The data from (A) and (C), respectively, showing fold expansion, which represents the colony frequency for each factor combination normalized to the frequency (B) or number (D) of colonies counted at day 10 in the BMP4 alone cultures for each experiment. In (B), p < .05 for all combinations versus BMP4 and, in (D), p < .05 for all combinations versus BMP4. Asterisks denote p < .05 for BVF and BVSF versus BV or BVS. (F, H): The data from (E) and (G), respectively, showing fold expansion, which represents the colony frequency for each factor combination normalized to the frequency (F) or number (H) of colonies counted at day 20 in the BV cultures for each experiment. In (H), p ≤ .03 for BVSF versus all other combinations. Examples of (I) multilineage, (J) myeloid, and (K) erythroid colonies from methylcellulose cultures established at day 10 of differentiation (×50 magnification) and cytospins of representative colonies stained with May-Grünwald-Giemsa (×1,000 magnification). Abbreviations: B, bone morphogenetic protein 4; BV, bone morphogenetic protein 4 and vascular endothelial growth factor; BVF, bone morphogenetic protein 4, vascular endothelial growth factor, and basic fibroblast growth factor; BVS, bone morphogenetic protein 4, vascular endothelial growth factor, and stem cell factor; BVSF, bone morphogenetic protein 4, vascular endothelial growth factor, stem cell factor, and basic fibroblast growth factor; CFC, colony-forming cell.

Figure Figure 6..

The combination of BMP4, vascular endothelial growth factor (VEGF), stem cell factor (SCF), and basic fibroblast growth factor (FGF2) is required for maximal generation of hematopoietic cells. (A–C): Flow cytometric analysis from a representative experiment at day 10 and day 20 of differentiating human embryonic stem cells (HESCs) cultured with BMP4 or the indicated factor combinations, showing results of staining for CD34+, CD45+, and CD33+ cells. The graphs at the end of each row indicate the percentage of cells expressing each antigen (mean ± SEM, n = 6) with the multifactor combinations (legend shown at the bottom of the figure). (D–G): Number of cells expressing (D) CD34+, (E) CD45+, (F) CD33+, and (G) CD117+, CD31+ and KDR+ as a function of time, in cultures supplemented with BMP4, VEGF, SCF, and FGF2 combinations as indicated. The absolute cell numbers (viable cells) generated per plate containing 72 wells of differentiating HESCs seeded at 2.5 × 103 cells per well are shown on the y-axis (n = 10). Values were determined by multiplying the number of viable cells per time point (determined by trypan blue exclusion) by the percentage of cells positive for each marker detected by flow cytometry. Asterisks denote p values, .01 < p < .05, compared with BV. Abbreviations: BMP4, bone morphogenetic protein 4; BV, bone morphogenetic protein 4 and vascular endothelial growth factor; BVF, bone morphogenetic protein 4, vascular endothelial growth factor, and basic fibroblast growth factor; BVS, bone morphogenetic protein 4, vascular endothelial growth factor, and stem cell factor; BVSF, bone morphogenetic protein 4, vascular endothelial growth factor, stem cell factor, and basic fibroblast growth factor; D, day; SSC, side scatter; Undiff, undifferentiated.

We also examined the differentiating cultures for expression of CD117 (c-KIT), the receptor for the cytokine SCF, and CD31 and KDR, proteins present on both early hematopoietic progenitors and endothelial cells (Fig. 6G). CD117 was expressed on approximately 32% of undifferentiated HESCs (data not shown), and a second wave of expression was observed at day 10, when most of the CD117+ cells also coexpressed CD34 (data not shown). The greatest yield of CD117+ cells was in cultures supplemented with BVSF (Fig. 6G), and similar results were observed for CD31 and KDR (Fig. 6G). Consistent with the observations of others [13], we found that CD34, CD31, and KDR were frequently coexpressed (data not shown).

These experiments demonstrate that BMP4 was sufficient to initiate hematopoietic gene expression and permit a proportion of cells to develop to the progenitor cell stage but was unable to support the generation of significant numbers of mature blood cells. This work also showed that VEGF was necessary for robust generation of CFCs, and the combination of four factors that included FGF2 and SCF increased the total yield of hematopoietic progenitors and mature cells.

Discussion

We have utilized a serum-free medium coupled with a spin EB differentiation system to dissect the contributions of four growth factors, alone and in combination, to the development of hematopoietic cells from HESCs. The role of BMP4 in inducing the expression of the primitive streak genes MIXL1, BRACHYURY, and GOOSECOID was consistent with its recognized role in mesoderm induction in a range of vertebrate species including frog, zebrafish, and mouse [35, 38, 39]. Conversely, VEGF, SCF, or FGF2 alone were unable to induce the primitive streak genes, nor did they augment the effects of BMP4 with respect to this outcome. These results were not surprising, since none of these three factors plays a role in the commitment of cells to mesoderm during embryogenesis.

The ability of BMP4 to induce hematopoietic gene expression and to permit the development of colony-forming cells was consistent with similar findings with mouse ES cells from our own laboratory and from the work of others [7, 40, 41]. Inclusion of both BMP4 and VEGF in the medium enhanced hematopoietic differentiation, as evidenced by increased expression of CD34 and SCL genes, increased frequency of progenitor cells, and increased numbers of immature and mature hematopoietic cells detected by flow cytometry. In previous studies, the exact roles played by these factors in hematopoietic differentiation of HESCs have been clouded by the presence of serum in the differentiation medium [42]. However, prior studies in mouse ESCs differentiated in serum-free media demonstrated that hematopoietic commitment was induced by BMP4 and that expansion of blood cell precursors required the inclusion of VEGF, consistent with the data generated in our current study [40, 41]. Indeed, the study by Park et al. utilizing mouse ESCs harboring a CD4 reporter gene at the Scl locus demonstrated that BMP4 alone induced Scl expression in a small percentage of cells, but that the combination of BMP4 and VEGF was required for robust induction of SCL+ cells [41]. These data highlight the conservation of mechanisms underlying hematopoietic differentiation of mouse and HESCs.

Previous work has shown that FGF2 complements factors produced by PMEFs in maintaining undifferentiated HESCs [43, [44], [45]–46]. In the mouse, the FGF pathway is required for mesoderm formation, dorsoventral patterning, and cell migration during gastrulation [26, 27, 47, 48], and Fgfr1-deficient mouse ESCs displayed a marked impairment in hematopoietic colony formation in vitro [49]. Our data showed that FGF2, included from the onset of differentiation, had little effect on hematopoietic progenitor cell frequency but increased the total yield of hematopoietic cells approximately fourfold for day 10 CFCs compared with BMP4 alone and three- to sixfold for day 20 CFCs compared with the combination of BMP4 and VEGF. This was accompanied by a sustained increase in total cell number from 2 days after the initiation of differentiation.

We observed that CD117 (c-KIT), the receptor for SCF, was expressed on undifferentiated HESCs, consistent with data correlating c-kit expression and pluripotency in mouse ES cells [50]. The possibility of an autocrine loop was suggested by a recent demonstration that mouse ESCs secreted biologically active SCF into the culture medium [51]. In contrast, another study that examined c-kit-null mouse ESCs did not uncover a defect in maintenance of undifferentiated cells, but rather demonstrated a specific Bcl2-mediated antiapoptotic function for kit signaling in differentiating ESCs following leukemia inhibitory factor withdrawal [52]. However, our studies demonstrated only a subtle benefit that derived from the inclusion of SCF in the differentiation medium during the first 10 days of differentiation. This was manifest as an increase in the total number of hematopoietic progenitors and mature hematopoietic cells generated in BVSF medium compared with BVF medium. This difference was not evident in the comparison between BVS and BV media, demonstrating that the SCF effect was context dependent and required the inclusion of FGF2. Taken together, these data suggest that FGF2 and SCF may synergize in promoting proliferation and preventing cell death in HESCs differentiating toward the blood lineages.

Many of the potential applications of HESC derived blood cells will require large quantities of mature cells or their progenitors produced in a serum- and stromal-free culture system. In order to achieve this outcome, it will be necessary to establish differentiation protocols that optimize each step of the process. In this respect, our current study provides a baseline for the further refinement of conditions required for the serum-free differentiation of hematopoietic cells from HESCs in vitro.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.

Acknowledgements

The authors are grateful to Robyn Mayberry and Kathy Koutsis for provision of HESCs and Elizabeth Ng for technical assistance. This work was supported by research grants from the Australian Stem Cell Centre, the Juvenile Diabetes Research Foundation, and the National Health and Medical Research Council (NHMRC) of Australia. A.G.E. is a Senior Research Fellow of the NHMRC.

Ancillary