Progress toward clinical application of ESC-derived hematopoietic cellular transplantation will require rigorous evaluation in a large animal allogeneic model. However, in contrast to human ESCs (hESCs), efforts to induce conclusive hematopoietic differentiation from rhesus macaque ESCs (rESCs) have been unsuccessful. Characterizing these poorly understood functional differences will facilitate progress in this area and likely clarify the critical steps involved in the hematopoietic differentiation of ESCs. To accomplish this goal, we compared the hematopoietic differentiation of hESCs with that of rESCs in both EB culture and stroma coculture. Initially, undifferentiated rESCs and hESCs were adapted to growth on Matrigel without a change in their phenotype or karyotype. Subsequent differentiation of rESCs in OP9 stroma led to the development of CD34+CD45− cells that gave rise to endothelial cell networks in methylcellulose culture. In the same conditions, hESCs exhibited convincing hematopoietic differentiation. In cytokine-supplemented EB culture, rESCs demonstrated improved hematopoietic differentiation with higher levels of CD34+ and detectable levels of CD45+ cells. However, these levels remained dramatically lower than those for hESCs in identical culture conditions. Subsequent plating of cytokine-supplemented rhesus EBs in methylcellulose culture led to the formation of mixed colonies of erythroid, myeloid, and endothelial cells, confirming the existence of bipotential hematoendothelial progenitors in the cytokine-supplemented EB cultures. Evaluation of four different rESC lines confirmed the validity of these disparities. Although rESCs have the potential for hematopoietic differentiation, they exhibit a pause at the hemangioblast stage of hematopoietic development in culture conditions developed for hESCs.
Nonhuman primates, such as the rhesus macaque, have >90% DNA homology to humans and have long been used as models for studies on human behavior, reproductive biology, embryology, and various disease states [1, , , , , , , , –10]. Given its similarities to humans, the rhesus macaque remains the primary in vivo model in which many clinically relevant questions regarding ESC-derived hematopoietic progenitor transplantation may be answered. Problems concerning microenvironmental induction/regulation of stem cell growth, specific allogeneic immune responses, and tumorigenesis cannot be satisfactorily addressed in xenogeneic or small animal allogeneic hosts. Furthermore, the homing and proliferation of human hematopoietic progenitors in the xenogeneic murine hematopoietic microenvironment may be greatly affected by disparate receptor/ligand and cytokine interactions. In addition, the limited proliferative demand placed on the transplanted cells as a result of the short life span of the mouse clouds the assessment of long-term hematopoietic stem cell (HSC) activity. Lastly, the absence of a reproducible allogeneic immune system prevents extrapolation of such xenogeneic transplantation studies to clinical applications. Hence, there is a crucial need to pursue study of ESC-derived hematopoietic progenitor transplantation in a well-characterized large animal allogeneic transplantation model that closely mimics the human hematopoietic system.
However, despite success with human ESCs (hESCs), similar efforts to induce conclusive hematopoietic differentiation of rhesus ESCs (rESCs) using the same techniques have been unsuccessful. A few studies suggest limited hematopoietic development; however, the temporal emergence of CD45-positive cells and characteristic colony forming cells from differentiating rESCs has not been observed [11, –13]. Similarly, studies of hematopoietic differentiation of cynomolgus ESCs grown on OP9 stroma revealed the development of primitive erythroid colonies expressing embryonic fetal and adult globin genes without documentation of CD45 expression [14, 15]. Elaborate protocols involving multiple steps of subculture in conditions specific to cynomolgus ESC differentiation are needed to generate cultures rich in CD45 expression . Ultimately, multilineage chimerism has not been achieved following transplantation into allogeneic or xenogeneic hosts [14, 17, 18]. Thus, the mechanisms underlying hematopoietic differentiation, expansion, and self-renewal are not well defined in monkey ESCs and, curiously, seem to differ from those of hESCs.
Given the importance of nonhuman primate ESCs, the current study pursues the differences in hematopoietic differentiation of rESCs compared with hESCs. Our results clearly demonstrate differential requirements for hematopoietic differentiation between these closely related species. Specifically, we show that rESCs differentiate to the bipotential stage of hematoendothelial development when using culture conditions developed for the hematopoietic differentiation of hESCs. Subsequent hematopoietic commitment is limited by an unclear mechanism. These findings emphasize the need to further study the differentiation of nonhuman primate ESCs as they relate to hESCs to gain essential information from this clinically relevant model.
Materials and Methods
Embryonic Stem Cells
The undifferentiated human embryonic stem cell line H9 (WiCell Research Institute, Madison, WI, http://www.wicell.org) was maintained by coculture with irradiated murine embryonic fibroblasts (MEFs) in Dulbecco's modified Eagle's medium (DMEM)/Ham's F-12 medium (F12) supplemented with 20% fetal bovine serum (FBS), 1% nonessential amino acids (NEAA), 1 mM l-glutamine (all from Invitrogen, Carlsbad, CA, http://www.invitrogen.com), 0.1 mM β-mercaptoethanol (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), and 4 ng/ml human basic fibroblast growth factor (bFGF) (R&D Systems Inc., Minneapolis, http://www.rndsystems.com) [19, 20].
The R366.4, R420, and R456 rhesus macaque ESCs derived from in vivo-flushed blastocysts  (kindly provided by James Thomson, Madison, WI) and the ORMES7 rhesus macaque ESCs isolated from in vitro-produced embryos  (kindly provided by Don Wolf, Portland, OR) were maintained in an undifferentiated state by coculture with irradiated MEFs  in DMEM/F12 (Invitrogen) supplemented with 20% FBS (HyClone, Logan, UT, http://www.hyclone.com), 1 mM glutamine, 0.1 mM β-mercaptoethanol, and 1% NEAA (Invitrogen).
Undifferentiated CJ7 murine ESCs (kindly provided by Dr. Stuart Orkin, Boston, MA) were maintained by coculture with irradiated MEFs in gelatin-coated flasks in DMEM supplemented with 15% FBS (HyClone), 1 mM sodium pyruvate, 1% penicillin/streptomycin, 2 mM l-glutamine, 1% NEAA (all from Invitrogen), and 100uM 1-thioglycerol (Sigma-Aldrich).
Maintenance of Undifferentiated rESC and hESC Feeder-Free Cultures
Initially, R366.4, R420, R456, ORMES-7, and H9 ESCs were maintained as undifferentiated cells by passage on irradiated MEFs (rESCMEF and hESCMEF). The cells were then adapted to feeder-free culture by allowing them to expand on Matrigel (BD Biosciences, San Diego, http://www.bdbiosciences.com)-coated plates as previously described by Xu et al.  and Carpenter et al. . rESC and hESC cultures growing on Matrigel (rESCMAT and hESCMAT) were maintained in MEF-conditioned media supplemented with bFGF at 4 ng/ml (R&D Systems).
Preparation of MEF-Conditioned Media
MEFs were harvested and irradiated at 40 Gy and seeded at 55,000 cells per cm2 in media containing 80% Knockout DMEM (KO-DMEM), 20% Knockout serum replacement, 1 mM l-glutamine, 0.1 mM β-mercaptoethanol, and 1% NEAA (all from Invitrogen). MEF-conditioned medium (MEF-CM) was collected and supplemented with bFGF at 4 ng/ml. hESCMAT and rESCMAT cultures were fed with MEF-CM daily. Cultures were passaged before they became confluent by incubation in 200 units/ml collagenase IV for 5 minutes at 37°C, dissociated, and then seeded back onto Matrigel-coated plates.
Coculture on OP9 Stromal Layer
rESCMAT and hESCMAT cells were seeded on confluent OP9 layers as previously described . Briefly, OP9 cells were maintained in α-minimal essential medium (αMEM) containing 15% FBS (HyClone). The cells were allowed to reach confluence at least 3 days prior to the plating of rESCMAT and hESCMAT cells. On the day of plating, the medium was changed to αMEM supplemented with 15% FBS (HyClone), 1 mM l-glutamine (Invitrogen), 50 μg/ml ascorbic acid (Sigma-Aldrich), 20% BIT9500 (Stem Cell Technologies, Vancouver, BC, Canada, http://www.stemcell.com), and 450 μM monothioglycerol. For the cytokine-supplemented OP9 cocultures, the following cytokines were added to the medium: 150 ng/ml stem cell factor (SCF), 150 ng/ml Flt-3 ligand (Flt-3L), 10 ng/ml interleukin (IL)-3, 10 ng/ml IL-6, 50 ng/ml granulocyte colony-stimulating factor (G-CSF), and 20 ng/ml bone morphogenetic factor (BMP-4) (all from R&D Systems). The cells were fed every fourth day and harvested on days 4–16 using collagenase IV.
Embryoid Body Culture
Undifferentiated hESCs and rESCs that were adapted to feeder-free growth on Matrigel-coated plates were harvested at confluence with collagenase IV. To promote EB formation, the cells were transferred to six-well low-attachment plates for an overnight incubation in KO-DMEM supplemented with 20% FBS (HyClone), 1% NEAA, 1 mM l-glutamine, and 0.1 mM mercaptoethanol (all from Invitrogen). The next day, cultures were fed fresh differentiation media alone (control) or were fed differentiation media supplemented with the following growth factors and cytokines: 150 ng/ml SCF, 150 ng/ml Flt-3L, 10 ng/ml IL-3, 10 ng/ml IL-6, 50 ng/ml G-CSF, and 20 ng/ml BMP-4 (all from R&D Systems). The media was changed every 4 days by transferring the EBs into a 15-ml tube and letting the aggregates settle for 5 minutes. The supernatant was aspirated and replaced with fresh differentiation media.
Undifferentiated murine CJ7 ESCs were maintained in feeder-free culture on gelatin-coated plates in the presence of LIF. For EB formation cells, were harvested using trypsin and were seeded in low-attachment plates in DMEM containing 15% FBS, (HyClone), 1 mM l-glutamine, 1% sodium pyruvate, 0.75% bovine serum albumin (fraction V), 450 μM monthioglycerol (all from Invitrogen), and 20% BIT9500 (Stem Cell Technologies). The next day, cultures were given fresh differentiation media alone (control) or differentiation media supplemented with 50 ng/ml SCF, 10 ng/ml IL-3, 10 ng/ml IL-6, 5 ng/ml G-CSF, 5 ng/ml granulocyte-macrophage colony-stimulating factor (GM-CSF), 10 ng/ml vascular endothelial growth factor (VEGF), 10 ng/ml thrombopoietin (TPO), and 10 ng/ml erythropoietin (EPO) (all from R&D Systems). The cells were fed every fourth day and harvested after 16 days of EB culture.
Cells were washed with media and treated with trypsin and collagenase IV (Invitrogen) for 20 minutes in a 37°C incubator followed by washes with media and passage through a 70-μm cell strainer. The cells were resuspended at approximately 2 × 105 cells per milliliter and stained with the following fluorochrome-conjugated monoclonal antibodies: anti-human CD43 (L10), anti-human CD31 (WM-59), anti-human CD38 (HIT2), anti-nonhuman primate CD41 (HIP8), anti-human CD117 (YB5.B8) (all from eBiosciences, San Diego, CA, http://www.ebioscience.com); anti-human CD45 (D058–1283), anti-human (HI30), anti-human CD34 (563) (BD Biosciences); anti-human FLK-1 (89106), anti-human SSEA-4 antibody (MC-813-70), and anti-human Oct3/4 (240408) (all from R&D Systems). Human-rhesus cross-reactivity was confirmed using rhesus tissue samples. Nonviable cells were excluded with 7-aminoactinomycin D (BD Biosciences). Live cell analysis was performed on a FACSCalibur flow cytometer with Cell Quest software.
For intracellular staining, cells were harvested using trypsin and collagenase IV, washed with phosphate-buffered saline (PBS), and fixed with 2% paraformaldehyde for 20 minutes on ice. The cells were washed with PBS and once with SAP buffer (PBS containing 2% fetal calf serum and 0.2% saponin). The cells were then stained with goat anti-Oct-3/4 (R&D Systems) in SAP buffer for 30 minutes at 4°C and then washed twice with SAP buffer. The cells were then incubated with a fluorescein isothiocyanate (FITC)-conjugated, isotype-specific anti-rat IgG2b secondary antibody (BD Pharmingen, San Diego, http://www.bdbiosciences.com/pharmingen) for 30 minutes at 4°C prior to washing and analysis on a FACSCalibur flow cytometer with Cell Quest software. For negative controls, the primary antibody was omitted.
Clonogenic Hematopoietic Progenitor Assay
Embryoid bodies were dispersed into single-cell suspensions using 1 mg/ml collagenase IV and 0.05% trypsin/EDTA. Viable cells were quantified, plated (3.0 × 105 cells per milliliter), and assayed in humidified chambers for hematopoietic CFCs using Human Methylcellulose Complete media (R&D Systems) containing 50 ng/ml SCF, 3 U/ml EPO, 10 ng/ml GM-CSF, and 10 ng/ml IL-3.
Individual colonies growing on methylcellulose were picked using a pulled-tip glass micropipette. The colony was placed in a 1.5-ml centrifuge tube with 1 ml of PBS. Cell clumps were dissociated by incubation with 0.05% trypsin for 5 minutes. The cells were then washed and resuspended in 300 μl of medium, mounted on Cytoclips (Thermo Electron Corporation, Waltham, MA, http://www.thermo.com), and centrifuged at 800 rpm for 5 minutes. Cells were fixed and stained with Wright-Giemsa reagents (Hema 3 stain; Fisher Scientific International, Hampton, NH, http://www.fisherscientific.com) according to the manufacturer's instructions.
For immunofluorescence staining, the cytospins were washed with twice with PBS and fixed for 10 minutes in PBS containing 2% paraformaldehyde. The fixed cells were washed with PBS and incubated initially with biotin-conjugated anti-VE-cadherin (16B1; eBioscience) and FITC-conjugated anti-CD45 (D058-1283; BD Biosciences) for 1 hour, washed five times, and incubated with streptavidin-Alexa Fluor 546 (Invitrogen) for another 45 minutes. Following the second staining step, the cells were washed and mounted using Antifade gold (Invitrogen). Images were acquired using an MRC 1024ES confocal microscope (Bio-Rad, Hercules, CA, http://www.bio-rad.com). Cells incubated with streptavidin-Alexa Fluor 546 alone and IgG1-FITC served as controls. For acetylated low-density lipoprotein staining, the colonies were plucked from methylcellulose and replated on a Matrigel-coated slide flask (Nalgene Nunc International, Rochester, NY) for 2 days. Aseptically, diI-acetylated low-density lipoprotein (Ac-LDL) (Biomedical Technologies, Stoughton, MA, http://www.btiinc.com) was diluted to 10 μg/ml in medium, added to the live cells, and allowed to incubate for 4 hours at 37°C. The medium was removed, and the cells were washed several times with medium. The cells were mounted using mounted using Antifade gold (Invitrogen) and visualized using standard rhodamine excitation emission filters via confocal microscopy.
Total RNA was isolated from undifferentiated rESCs and 16-day rhesus EB cultures using the RNAqueous-4PCR Kit (Ambion, Austin, TX, http://www.ambion.com). RNA was treated with RNase-free DNase at the last step of the reaction. cDNA was synthesized using the Bio-Rad iScript cDNA synthesis kit. Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was performed using iQ SYBR Green Supermix reagents and an iCycler thermal cycler and software (Bio-Rad). Rhesus-specific primers were as follows: 5′-GAAACCGCAAGGCATCTG-3′ (forward) and 5′-CCCACAATTCCCGCTACC-3′ (reverse) for GATA-1; 5′-CCACAGCCCTAGTATGAAAG-3′ (forward) and 5′-TCACCGCATACAGAATCTAAG-3′ (reverse) for GATA-2; 5′-AAAACAAAGGGCACAGCATC-3′ (forward) and 5′-GAGACCAACGCAATTCATCA-3′ (reverse) for SCL′; 5′-ATGCACGGCATCTGGGAATC-3′ (forward) and 5′-GTCACTGTCCTGCAA-GTTGCTGTC-3′ (reverse) for FLK-1; 5′-CCCTCTCCTGGGAGCATT-3′ (forward) and 5′-AAAGAGAGGAAGGCTCTGGTG-3′ (reverse) for RUNX-1; and 5′-ATCCCCCAATTCTCTGGAAC-3′ (forward) and 5′-ATTGGGGAACTCCAGACACA-3′ (reverse) for PU.1. All primers were tested and optimized for specificity with rhesus samples and nonreactivity with SYBR Green reagents using non-reverse-transcribed cDNA. Briefly, for each reaction, 12.5 μl of the SYBR Green PCR Master Mix (Bio-Rad) was mixed with 10 μM each primer (for each gene of interest or glyceraldehyde-3-phosphate dehydrogenase [GAPDH]) and reverse-transcribed cDNA. The thermal cycling conditions comprised a hot start step at 95°C for 3 minutes. Cycle conditions were 30 seconds at 95°C and 30 seconds at 58°C. Each sample underwent 40 cycles of amplification. All qRT-PCRs were confirmed for specificity of a single PCR product by analysis on 2% agarose gels. Comparative quantification of each target gene was performed based on cycle threshold (CT) normalized to GAPDH using the ΔCT method . The relative expression of each normalized target gene was compared with the GAPDH-normalized expression of the target gene. Fold change expression from undifferentiated rESCs was calculated as 2−ΔΔCT, where ΔΔCT = (ΔCT of differentiated rESCs) − ΔCT (undifferentiated rESCs).
Undifferentiated rESCs Demonstrate a Phenotype Similar to That of hESCs When Expanded on MEF or Matrigel
In preliminary studies, we observed that the presence of even a small number of MEFs significantly inhibited the hematopoietic differentiation of rESCs. Therefore, undifferentiated rESCs (R366.4, R456, R420, and ORMES-7) and hESCs (H9) were expanded on irradiated MEFs (rESCMEF and hESCMEF) and adapted to feeder-free growth on Matrigel-coated plates (rESCMAT and hESCMAT). The rESCMAT maintained a normal karyotype after nearly 20 passages on Matrigel (Fig. 1A–1C). Each of the human and rhesus ESC lines has been shown to induce the development of teratomas following transplantation into immunodeficient mice [19, 21, 27, –29]. Figure 1D–1G provides sample histologic images of a single teratoma derived from the injection of undifferentiated ORMES-7 rESCs. Morphologic evidence is shown for cartilaginous (mesoderm), intestinal (endoderm), and neural (ectodermal) differentiation.
Following expansion on Matrigel, the undifferentiated ESCs were harvested and analyzed for the expression of antigens associated with pluripotency (c-Kit, SSEA-4, and Oct3/4) [30, –32], with early hematoendothelial differentiation (CD31, CD34, and FLK-1) , and with hematopoietic lineage commitment (CD38, CD41, and CD45) [34, 35]. The results showed that undifferentiated rESCs and hESCs were completely devoid of CD45 expression but expressed a low frequency of CD34+ and FLK-1+ cells, suggesting some heterogeneity among both rESCs and hESCs (Fig. 2A, 2B). Both human and rhesus ESCs maintained on Matrigel displayed a slightly higher frequency of the FLK-1+ subset of cells in the culture but an otherwise similar differentiation profile compared with ESCs maintained on MEFs. As expected, the Oct-4, c-Kit, and SSEA-4 were highly expressed in the undifferentiated ESCs.
rESCs Differentiating on OP9 Stroma Lack Phenotypic and Functional Hematopoietic Properties
Recently, Vodyanik et al.  demonstrated the generation of hematopoietic progenitor cells following coculture of hESCs with OP9 stroma. We compared the hematopoietic differentiation of rESCs with that of hESCs on OP9 stroma. As shown in Figure 3, rESCs demonstrated a limited capacity for hematopoietic differentiation compared with hESCs in OP9 stromal culture. Although CD34+ cells could be detected at low levels, no CD45+ cells were seen (<0.3%), despite nearly 3 weeks of OP9 coculture (Fig. 3A). Subsequent plating of the differentiating rESCs in methylcellulose culture led to the development of extensive networks of endothelial cells (data not shown). Conversely, hESCs grown under identical conditions resulted in dramatic hematopoietic differentiation, demonstrated by a high frequency of FLK-1+ (46%), CD45+ (21%), CD34+(47%), CD41+ (26%), and CD43+ cells with downregulation of c-kit (Fig. 3B, 3D). Thus, unlike hESCs, coculture of rESCs with OP9 did not result in clear hematopoietic differentiation but instead exhibited a bias toward endothelial differentiation.
Cytokine Supplementation of EB Cultures Leads to Improved Hematopoietic Differentiation of rESCs to a Lesser Degree Than in hESC Cultures
Chadwick et al.  demonstrated the hematopoietic differentiation of hESCs in cytokine-supplemented EB cultures. We compared the differentiation of rESCs and hESCs in EB culture with and without cytokine supplementation. Rhesus EB cultures without cytokine supplementation demonstrated low levels of CD34+ (<0.5%) and essentially undetectable levels of CD45-, CD41-, CD43-, and CD38-expressing cells despite more than 3 weeks in culture (Fig. 4A). In contrast, human EBs revealed modest hematopoietic differentiation even in the absence of cytokines CD34 (6.7%), CD31 (5.6%), and CD45 (4.14%) (Fig. 4B).
Rhesus EBs formed in the presence of cytokines (BMP-4, SCF, Flt-3 ligand, IL-3, IL-6, and GM-CSF) demonstrated a decrease in SSEA-4 frequency (60%–12%), with a concomitant rise in the levels of CD41+ and CD34+ cells. In addition, for the first time, detectable frequencies of CD45+ cells were observed beginning at EB12 and continuing to EB22 (0.18%–0.98%) in the presence of cytokines (Fig. 4E, 4G). Human EBs revealed a robust hematopoietic differentiation with higher frequencies of CD34-positive (10.0 ± 1.7%), CD45-positive (20.0 ± 7.2%), and CD31-positive (18 ± 4.7%) cells in the presence of cytokines (Fig. 4F, 4H).
rESCs Demonstrate a Lower Capacity for Hematopoietic Differentiation in EB Culture Compared with hESCs or mESCs
Using similar EB culture conditions, we compared the hematopoietic differentiation of rhesus EBs to both murine and human EBs. As shown in Figure 5A, both human and murine EBs demonstrated robust hematopoietic differentiation, as approximately one fourth of the cultured cells expressed CD45. Significantly lower levels of CD45 expression were observed in the differentiating rhesus EBs compared with the either the human or murine EBs. Surprisingly, the differentiation profile of H9 human EBs was closer to that of the CJ7 murine EBs than to those derived from the four different rESC lines tested (R366.4, R420, R456, and ORMES-7). To evaluate the possibility that the rESCs simply require more time in culture to demonstrate significant hematopoietic lineage commitment, rhesus EB cultures were maintained with cytokine supplementation for up to 7 weeks. As shown in Figure 5B, higher levels of CD34+ cells were observed in all three lines of rESCs. However, these increases in CD34 expression were not associated with a hematopoietic lineage commitment, as the levels of CD45 decreased from 3.63% to 0.66% in R420 cells, whereas they remained relatively unchanged in the other two cell lines. In addition, there was a decrease in CD31 frequency in the R456 (from 4.4% to 0.7%) and R420 (from 3.48% to 0.32%) EBs. Thus, allowing the EBs to differentiate for longer periods of time enhanced CD34 expression but did not augment the levels of CD45+ cells.
Next, we examined the effect of cytokine supplementation on the differentiation of rESCs during OP9 coculture. As shown in Figure 5C, cytokine supplementation of OP9 stroma cocultures resulted in slightly lower frequencies of CD34+ and CD41+ cells and higher levels of FLK-1+ cells. The increases in FLK-1 expression suggested enhanced development of hematopoietic mesoderm; however, CD45+ cells remained at an undetectable level. Lastly, rESCs subjected to 4 weeks of OP9 coculture in the absence of cytokines demonstrated rapid expansion of CD34+ cells (76% of mixed culture at 28 days), yet CD45 expression still remained undetectable (Fig. 5D).
Rhesus EBs Express a Transcriptional Profile Consistent with Hematoendothelial Differentiation and Display Both Hematopoietic and Endothelial Differentiation in Semisolid Clonogenic Culture
To further characterize their hematopoietic development, rhesus EBs were studied for their expression of transcription factors critically associated with hematoendothelial development and subsequent lineage commitment (SCL/Tal-1, GATA-1, GATA-2, PU.1, and RUNX1). SCL/Tal-1 has been shown to play a fundamental role in the earliest stages of hematopoietic and endothelial development. In addition, blast colony-forming cells (BL-CFCs) (putative murine hemangioblasts) fail to form from SCL−/− ESCs [37, 38]. GATA-1 is critical to the maturation of primitive and definitive erythroid cells, megakaryocytes, eosinophils, and mast cells [39, –41]. Conversely, GATA-2 plays an indispensable role in the formation of hemangioblasts and HSCs and is transcriptionally repressed by GATA-1 [42, , , –46]. Although GATA-2 is necessary for the formation of hemangioblasts, overexpression at this stage leads to failed hematopoietic development. PU.1 has a diverse role in the hematopoietic lineage commitment of erythrocytes, macrophages, megakaryocytes, and B lymphocytes. Various levels of PU.1 expression are associated with differentiation of each of these lineages. Interaction with GATA transcription factors is essential in this regard, and GATA-2 overexpression inhibits PU.1 transcription. Lastly, RUNX1 (AML1) expression is essential for the development of definitive fetal liver hematopoietic progenitors and profoundly influences lineage commitment of adult hematopoietic progenitors [47, , , –51]. The development of definitive HSCs in the embryonic environment has been shown to be associated with both the timing and the level of RUNX1 expression .
The expression of these transcription factors was measured in three lines of undifferentiated rESCs and compared with day 16 rhesus EBs cultured in cytokine-enriched media. As shown in Figure 6A, all three lines of EBs demonstrated upregulation of factors associated with early hematoendothelial development, as evidenced by increases in GATA-1, GATA-2, SCL, and FLK-1 expression and a dramatic fall in Oct-3/4 expression. However, absence of hematopoietic lineage commitment was demonstrated by a lack of upregulation of either RUNX1 or PU.1 in two of the three lines studied. Interestingly, PU.1 appeared to be upregulated in the third group of rhesus EBs (R420), which also demonstrated the highest frequency of CD45+ cells, as shown in Figure 5A.
When plated in methylcellulose, robust hematopoietic colony formation was observed from both murine and human EBs (data not shown). However, as shown in Figure 6B, cells from rhesus EBs formed colonies of mixed erythroid, myeloid, and endothelial cell types, signaling the existence of bipotential hematoendothelial progenitors in the cultures reminiscent of BL-CFCs. The loosely adherent cells in the hematoendothelial colonies displayed erythroid or macrophage morphology on examination of Wright stains (Fig. 6C, 6D). To better characterize these mixed colonies, the cells were stained for antigens associated with endothelial (VE-cadherin and Ac-LDL) or hematopoietic (CD45) lineage commitment. As shown in Figure 6E–6G, small, rounded cells characteristic of hematopoietic morphology expressed only CD45 without VE-cadherin. Conversely, cells exhibiting endothelial morphology expressed VE-cadherin brightly without CD45 expression. Triple staining with CD45, VE-cadherin, and Ac-LDL is shown in Figure 6H–6K. Similar morphology was seen in cells with high uptake of Ac-LDL. Taken together, these findings indicate that exposure of the rhesus EBs to the current cytokine cocktail results in the abundant development of bipotential hematoendothelial progenitors with only limited commitment to definitive hematopoietic lineages. This “pause” in hematoendothelial differentiation illustrates a point of divergence from patterns of hematopoietic differentiation seen in murine and human EBs under similar conditions.
Progress toward clinical application of ESC-derived hematopoietic progenitor cell transplantation requires rigorous evaluation in a clinically relevant animal model, such as monkeys. However, in contrast to hESCs, efforts to induce conclusive hematopoietic differentiation from rESCs have been unsuccessful [11, 13, 53, –55]. These peculiar differences in hematopoietic differentiation between human and nonhuman primate ESCs have not been consistently studied and are therefore poorly understood. Characterizing these functional differences will likely clarify the critical regulatory steps involved in the hematopoietic differentiation of ESCs. The need for a better understanding of rESC-derived hematopoiesis as it relates to human ESC biology provides the major impetus for this study.
In the present study, rESCs were adapted to feeder-free growth on Matrigel. The rESCMAT, like the hESCMAT , retained a normal karyotype despite nearly 20 passages of feeder-free growth. In OP9 stromal coculture, rESCs demonstrated no CD45 expression but developed CD34+ cells, which gave rise to endothelial cell networks in subsequent methylcellulose culture. In the same conditions, hESCs exhibited convincing hematopoietic differentiation. In cytokine-supplemented EB culture, rESCs demonstrated improved hematopoietic differentiation, as evidenced by significant levels of CD34+ and CD41+ cells and detectable levels of CD45+ cells. However, these levels remained dramatically lower than those for hESCs under identical culture conditions. Subsequent plating of cytokine-supplemented rhesus EBs in methylcellulose culture led to the formation of mixed colonies of erythroid, myeloid, and endothelial cells, confirming the existence of bipotential hematoendothelial progenitors in the cytokine-supplemented EB cultures. Evaluation of four independently isolated rESC lines confirms the validity of these disparities.
Previous comparisons of murine and human hematopoietic ontogeny have revealed extensive homology between these otherwise disparate species. Surprisingly, ESCs derived from the more closely related rhesus macaque behave differently than their human counterparts in the same culture environment. These observations have been reported in previous studies of ESCs derived from the rhesus macaque and other nonhuman primates. Li et al.  showed evidence for hematopoietic differentiation when undifferentiated rESCs were cocultured with the S17 stroma line in media supplemented with various cytokines, including BMP-4, VEGF, SCF, Flt-3 ligand, IL-3, IL-6, G-CSF, Epo, and GM-CSF. Cells cultured for 2 weeks in this potent mixture displayed increased expression of CD34 and formed cobblestone-area-forming colonies (CAFCs) in secondary culture. Although these findings suggest some hematopoietic development, the CD34+ cells failed to form hematopoietic colonies in standard methylcellulose assay. Although the authors did not include studies to quantify the frequency of CD45+ cells, they did demonstrate the formation of erythroid, myeloid, and megakaryocytic lineages, which would be expected to express CD45 . A gene-expression analysis of the CAFCs revealed a significant in increase in the activity of endothelial regulatory genes, such as VE-cadherin and von Willebrand factor. Consequently, the authors concluded that hematopoietic differentiation culture of rESCs results in cells capable of both endothelial and primitive hematopoietic lineage development .
Recent reports have shown the generation of hematopoietic cells from cynomolgus ESCs (cyESCs), but the methodology used differs significantly from that used previously for hESCs. Hiroyama et al.  cultured the CMK6 cyESC line with multiple mouse-derived stroma cell lines (OP9, C2C12, and C3H10T1/2) to induce hematopoietic differentiation. The authors then selectively enriched the floating cells in the presence of VEGF and insulin-like growth factor-II (IGF-II) . VEGF and IGF-II were used to induce further hematopoietic differentiation in an elaborate four-step culture protocol . The studies by Umeda et al. [14, 15] also describe the culture of the same CMK6 cyESCs on OP9 stromal layers. Their studies parallel our findings when similar conditions are used, demonstrating an abundance of CD34+ cells and a very low frequency of CD45+ cells (<1%) in their cultures. Similarly, they found that removing the cells from the stromal microenvironment facilitated the enrichment of the hematopoietic progenitors [14, 15]. These findings imply the absence of an essential growth factor or the existence an inhibitor in OP9 coculture that limits hematopoietic commitment in cyESCs. This effect is not observed in OP9 coculture with hESCs as shown in Figure 3B.
In addition, a study of hematopoietic differentiation in common marmoset ESCs (cmESCs)  demonstrates strikingly similar findings to the present study in rESCs. In that report, four separate cmESC lines were differentiated for up to 18 days as EBs or on OP9 stroma with media containing GM-CSF, IL-3, IL-6, TPO, Flt-3L, VEGF, bFGF, BMP-4, and activin-A . No CD45 expression or hematopoietic colony formation was observed in OP9 coculture. Very limited colony formation was observed in EB culture. Despite the absence of hematopoiesis, CD34 and FLK-1 expression increased with differentiation. The authors similarly noted the disparity in their results compared with prior studies in hESCs . They concluded that cmESCs may possess a lower rate of intrinsic hematopoietic differentiation compared with hESCs. Interestingly, the authors found that when the cmESCs were transduced to overexpress Scl/Tal1 under the control of an EF1-α promoter, a dramatic increase in the hematopoietic colony formation was observed. This was also associated with a significant rise in Gata-1 expression, possibly signaling a switch toward hematopoietic commitment.
Similar conclusions were reached in the present study and are expanded to include multiple rhesus ESC lines in direct comparison with both human and murine ESCs. For the first time, the development of detectable quantities of CD45+ cells confirms the existence of lineage-committed hematopoietic cells. However, their frequency is extremely low relative to both murine and human EB culture. The presence of FLK-1 and CD34 expression points to an arrest of hematopoietic commitment at the bipotential stage of hematoendothelial progenitor development, and subsequent differentiation reveals a bias toward endothelial differentiation. As shown in Figure 5B, allowing the rhesus EBs to differentiate for nearly 7 weeks did not change the above findings. Lastly, subculture of day 12 EB cultures by dissociation of the EBs and replating decreased the levels of CD45-positive cells and enhanced the endothelial differentiation (data not shown).
The observed differences between human and rhesus ESCs did not relate to the subtle differences in their initial procurement. Three of the rESC lines described in this study (R366.4, R420, and R456) were isolated from the inner cell mass of blastocysts produced in vivo. In contrast, one of the rhesus ESC lines (ORMES-7) and the human ESC line (H9) were derived from in vitro-derived rhesus or human blastocysts [19, 57, –59]. The subsequent steps in the isolation and expansion of all ESCs derived from either species were identical. Since the results were consistent among all four lines of rESCs, we conclude that the disparities in hematopoietic differentiation between the rhesus and human ESC cultures were not simply due to the technique of blastocyst derivation.
Possible explanations for the observed arrest in the hematopoietic lineage commitment of differentiating rESCs include either the absence of an essential factor needed to augment hematopoietic differentiation or the existence of a specific inhibitory signal. In the case of OP9 stromal coculture, no trophic factors were added beyond those found in the fetal calf serum or provided by the stromal cells themselves. Although OP9 stroma is derived from the macrophage colony-stimulating factor-deficient osteopetrotic mouse and has been shown to secrete SCF and IL-7 , the underlying molecular interactions that provide HSC support remain unknown [61, 62]. Absence or low amounts of stromal-derived factor-1 production by OP9 stroma  and the endogenous levels of its receptor, CXCR4, might influence hematopoietic differentiation of rESCs on OP9 stroma and explain the differential stromal requirement between rESCs and hESCs. Future experiments designed to characterize the nature of the support provided by the OP9 stroma to human ESCs are currently under way and may provide clues regarding the key inductive or inhibitory factors influencing hematopoietic differentiation of rESCs.
In the EB culture of rESCs, a supportive stromal layer is not used, thus eliminating any effects from direct contact or soluble factors arising from murine fibroblasts. However, an autologous stromal cell population also develops within the EBs that may impair hematopoietic differentiation . Soluble factors, adhesion molecules, or other cell-surface proteins on the autologous stromal cells may provide an inhibitory signal to the developing ESC-derived hematopoietic cells. Although the percentage of CD45+ cells is low, rhesus EBs seem to follow a pattern of phenotypic hematopoietic development similar to that of human EBs (Fig. 4). It remains possible that a stage-specific inhibitory signal is provided by the autologous stroma. Potential inhibitory signals known to have negative regulatory effects on hematopoiesis include tumor necrosis factor-α (TNF-α) and transforming growth factor-β (TGF-β). These cytokines can be produced by both the autologous stroma, resulting in paracrine inhibition of hematopoietic differentiation [65, , –68]. Future experiments designed to block TNF-α and TGF-β function are planned to assess the role of these factors in the hematopoietic differentiation of rhesus EBs.
Growing evidence shows that fibroblast growth factor (FGF) can positively regulate hematopoiesis by acting on various cellular targets, including stromal cells, early and committed hematopoietic progenitors, and possibly some mature blood cells [37, 69, , , , , –75]. It synergizes with other hematopoietic cytokines and antagonizes the negative regulatory effects of TGF-β, thus potentially playing a central role in hematopoiesis. The FGF family acts through high-affinity tyrosine kinase receptors designated fibroblast growth factor receptor 1 (FGFR1) (flg-1), FGFR2 (bek), FGFR3, and FGFR4. FGF-mediated signaling is critical for the proliferation of the hemangioblast . Loss of FGFR-1 expression in murine EB cultures affects hemangioblast differentiation, resulting in an attenuation of hematopoietic development . Absence of FGFR-1 results in an increase in brachyury; a decrease in GATA-1, CD45, and Tal-1; and no change in Runx1, suggesting a lack of hematopoietic commitment from the murine EB-derived hemangioblasts . Thus, endogenous levels of FGFR within the rhesus EBs could specifically alter the differentiation of developing hemangioblasts. We are currently studying a number of candidate populations of hematoendothelial progenitors for their hematopoietic differentiation capacity following isolation and subculture.
In summary, this direct comparison of hematopoietic differentiation between hESCs and rESCs illustrates critical differences in the biology of cultured ESCs between these closely related species. In light of the importance of a nonhuman primate model to clinical application of ESC-based therapy, these findings emphasize the need to develop reliable rESC-specific protocols for hematopoietic differentiation. An improved understanding of the factors regulating hematopoietic differentiation from rESCs will facilitate subsequent transplantation studies in allogeneic hosts.
The authors indicate no potential conflicts of interest.
We thank Daniel Webster for assistance with the qRT-PCR assays, Rachel Lewis for technical assistance with the rESC cultures, Maxim Vodyanik for assistance with the hESC/OP9 coculture technique, and James Byrne for technical assistance with the in vivo teratoma formation. This work was supported in part by a grant to the University of Wisconsin Medical School under the Howard Hughes Medical Institute Research Resources Program for Medical Schools (A.F.S.) and by a Faculty Research Fellowship Award from the American College of Surgeons (A.F.S.).