Hemangioblastic Derivatives from Human Induced Pluripotent Stem Cells Exhibit Limited Expansion and Early Senescence§


  • Author contributions: Q.F.: Conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; S-J.L.: Conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; I.K., I.G., D.K., Y.C.: Collection and assembly of data; G.H.: Data analysis and interpretation; K-S.K.: Collection and assembly of data, data analysis and interpretation; R.L.: Conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript.

  • First published online in STEM CELLS EXPRESS February 4, 2010.

  • §

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


Human induced pluripotent stem cells (hiPSC) have been shown to differentiate into a variety of replacement cell types. Detailed evaluation and comparison with their human embryonic stem cell (hESC) counterparts is critical for assessment of their therapeutic potential. Using established methods, we demonstrate here that hiPSCs are capable of generating hemangioblasts/blast cells (BCs), endothelial cells, and hematopoietic cells with phenotypic and morphologic characteristics similar to those derived from hESCs, but with a dramatic decreased efficiency. Furthermore, in distinct contrast with the hESC derivatives, functional differences were observed in BCs derived from hiPSCs, including significantly increased apoptosis, severely limited growth and expansion capability, and a substantially decreased hematopoietic colony-forming capability. After further differentiation into erythroid cells, >1,000-fold difference in expansion capability was observed in hiPSC-BCs versus hESC-BCs. Although endothelial cells derived from hiPSCs were capable of taking up acetylated low-density lipoprotein and forming capillary-vascular-like structures on Matrigel, these cells also demonstrated early cellular senescence (most of the endothelial cells senesced after one passage). Similarly, retinal pigmented epithelium cells derived from hiPSCs began senescing in the first passage. Before clinical application, it will be necessary to determine the cause and extent of such abnormalities and whether they also occur in hiPSCs generated using different reprogramming methods. STEM CELLS 2010;28:704–712


There is considerable excitement that human induced pluripotent stem cells (hiPSCs) can serve as a potentially safe and embryo-free source of patient-specific cells for regenerative medicine. Since 2007, when hiPSC lines were first generated by Yu et al. [1] and Takahashi et al. [2], several methods have been reported for reprogramming somatic cells to pluripotency [3–7]. The ability of hiPSCs to differentiate into derivatives of all three embryonic germ layers is well established, and rapid progress is being made toward controlled in vitro differentiation of hiPSCs into specific cell types, precursors as well as differentiated progenies representing various tissues, such as heart [8], pancreas [9], liver [10], eye including retinal pigment epithelium (RPE) cells [11–14], neuronal [15, 16], and endothelial and hematopoietic lineages [17–20]. Although these studies clearly suggest a similar differentiation potential between hiPSCs and human embryonic stem cells (hESC), it is unclear whether they can be expanded into homogeneous cell populations suitable for use in drug discovery and clinical translation.

We have developed an efficient method to reproducibly generate large numbers of bipotential progenitors—known as hemangioblasts—from multiple hESC lines using an in vitro differentiation system [21, 22]. These blast cells (BCs) express gene signatures characteristic of hemangioblasts, and could be differentiated into multiple hematopoietic lineages as well as into endothelial cells that could repair ischemic vasculature in vivo [21]. Using hESC-derived hemangioblasts/BCs as intermediates, we have also generated functional oxygen-carrying erythrocytes on a large scale from multiple hESC lines, demonstrating the robust expansion capability of these cells [23]. This system provides an excellent model for evaluation and comparison of hiPSC derivatives to their hESC counterparts. In the present study, we successfully generated BCs, endothelial cells, and hematopoietic cells from multiple well-characterized hiPSC lines. We further compared the functional characteristics of BCs derived from hESCs and hiPSCs, and found that hemangioblastic derivatives generated from these hiPSC lines display abnormal molecular and cellular processes compared with their corresponding hESC counterparts. Similarly, RPE cells derived from hiPSCs begin senescing in the first passage, indicating that the observed phenomenon is not limited to hemangioblastic lineages.


Human ESC and iPSC Cultures..

The hESC lines used in this study were H1, H7, and H9 [24]; 9 Harvard Stem Cell Institute hESC lines [25] and 13 hESC lines derived at Advanced Cell Technology [26, 27] (see Table 1 and online supporting information Table 1). The hiPSC lines used were iPS(IMR90)-1, iPS(Foreskin)1-1, iPS(Foreskin)4-1, and 4-3, which were generated using Thomson's four factors (Oct-4, Sox-2, Nanog, and Lin28) with lentiviral vector [1]; rv-hiPS01, rv-hiPS02, rv-hiPS03, and rv-hiPS04, which were derived by using Yamanaka's four factors (Oct-4, Sox-2, c-Myc, and Klf4) with retroviral vector [6]. Cells were cultured on mitomycin-treated feeder mouse embryonic fibroblasts (MEFs) in hESC medium supplemented with 20% Knockout Serum Replacement (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), 10 ng/ml basic fibroblast growth factor (bFGF) (Stemgent, Cambridge, MA, www.stemgent.com, β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO, http://wwwsigmaaldrich.com), non essential amino acids (NEAA), and GlutaMax-I and penicillin/streptomycin (Invitrogen). Cells were fed with fresh medium daily and confluent hESC/hiPSC cultures were split using 1 mg/ml collagenase IV in Dulbecco's Modified Eagle's Medium/F12 (Invitrogen).

Table 1. Differentiation efficiencies of human induced pluripotent stem cells and human embryonic stem cells
inline image

Embryoid Body Formation..

We incubated hESC/hiPSC cultures with collagenase IV at 37°C for up to 10 minutes and collected the large clumps of cells. The clumps were washed once with 5 ml hESC medium. To initiate differentiation, the cell clumps were resuspended in embryoid body (EB) differentiation medium (Stemline II supplemented with 50 ng/ml BMP4 [R&D Systems Inc., Minneapolis, MN, http://www.rndsystems.com], and vascular endothelial growth factor [VEGF] [Invitrogen]), at a density of approximately 2 × 106 cells per well of a 6-well ultralow attachment plate (Corning Life Sciences, Lowell, MA, http://www.corning.com/lifesciences).

Generation of Hemangioblasts/BCs..

Hemangioblasts/BCs were generated as previously described [21, 22]. Briefly, EBs were cultured in differentiation medium for 48 hours, and then half of the medium was replaced with medium containing BMP4 (50 ng/ml), VEGF (50 ng/ml), and bFGF (20 ng/ml). After a 24-hour incubation, the EBs were collected, washed once in 2 ml D-PBS (Invitrogen), and incubated with 1 ml of trypsin/EDTA (Invitrogen) at 37°C for 2 to 3 minutes. The EBs were dissociated by pipetting up and down 10 times using a 1,000 μl pipette or until the cell suspension became homogenous with no visible clumps. An equal volume of MEF medium was added and the cells were centrifuged at 210g for 4 minutes. The cell pellet was resuspended in Stemline II medium to 1-2 × 106 cells/ml, 5 × 104 cells plated in 1-2 ml of blast growth medium (BGM) [21, 22], and then incubated for up to 6 days.

Characterization of Hematopoietic Colony-Forming Capability..

Day 6 hiPSC-BCs and hESC-BCs were purified from blast cell cultures. Next, 1 × 103 blast cells were resuspended in 20 μl of Stemline II medium, mixed well with 0.5 ml of either colony forming cells (CFC) medium (H4436 only, Stem Cell Technologies) (Stemcell Technologies, Vancouver, BC, Canada, http://www.stemcell.com/) or BGM (H4436 plus cytokines) using two quick vortex pulses. Cells in semisolid colony-forming units (CFU) assay medium were plated into wells of a 24-well ultra-low attachment plate for CFU colony growth. Cells were incubated in a carbon dioxide incubator at 37°C for up to 15 days. The formation of hematopoietic colonies was monitored microscopically. On day 15, the number of hematopoietic colonies in each well and their morphology were recorded. For the colony forming unit-megakaryocytic (CFU-MK) assay, BCs were purified and plated according to the manufacturer's instructions using a MK CFU assay kit (Stem Cell Technologies, Cat 04973).

Differentiation of hiPSC-BCs and hESC-BCs into Erythroid Cells..

Erythroid differentiation was carried out as previously described [23]. Briefly, day-5 blast cells from both hiPSCs and hESCs were purified and plated into 24-well plates (30,000 cells per well) with BGM medium and allow to expand for 5 to 7 days or until the culture became confluent. At the end of each expansion, cells were collected from each well and counted. To expand further, equal numbers of cells from both hiPSC-BCs and hESC-BCs were replated in BGM medium. A total of three expansion cycles were performed in this study (each expansion experiment was carried out in triplicate). To further differentiate toward erythroid cells, expanded BGM medium with cells was mixed at 1:1 ratio with Stemline II + EPO (3U/ml, Cell Sciences, Canton, MA, http://www.cellsciences.com/) and cultured for an additional 6 to 8 days. Additional Stemline II + EPO medium was added every 2 to 3 days. The erythrocytes were then spun onto glass slides and stained for benzidine/Giemsa (Sigma) and erythrocyte marker CD235a (DAKO, Glostrup, Denmark, http://www.dako.com) as described previously [23].

Differentiation of Endothelial Cells from hiPSC-BCs and hESC-BCs..

Purified BCs from both hESCs and hiPSCs were plated into wells of fibronectin-coated 24-well plates (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com) with EGM-2 medium (Lonza, Basel, Switzerland, http://www.lonza.com/group/en/company/about.html) at a density of 36,000 cells per well (n = 4). Endothelial cells differentiated from hESCs and hiPSCs were subcultured when they reached confluence, and were continuously cultured for up to 30 days. The morphology and growth rate of endothelial cells were recorded. To test the endothelial clonogenic capability of BCs derived from hESCs and hiPSCs, individual blast colonies were handpicked and plated for endothelial cell colony formation.

Endothelial Markers and Cell Senescence Staining..

The BC-derived endothelial cells were allowed to grow for 10 days in EGM-2 medium. The endothelial cells derived from hiPSCs were trypsinized and plated onto Matrigel (BD Biosciences) for network structure formation as described previously [21]. Expression of endothelial markers CD31 (DAKO), VE-Cad (Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com). and vWF (DAKO) on the hiPSC-derived endothelial cells was confirmed by immunofluorescence staining as described previously [21]. Low-density lipoprotein (LDL) uptake of hiPSC-endothelial cells was evaluated by culturing the cells in the presence of Alexa-Fluor 594 acetylated low density lipoprotein (Ac-LDL) (Invitrogen) at 10 μg/ml in EGM-2 for 8 hours, followed by washing with medium three times. The intracellular fluorescent LDL was examined using fluorescence microscopy. For detection of endothelial cell senescence, cells were fixed between 20 to 30 days after initiation of endothelial differentiation and then stained for endogenous β-galactosidase using a Cell Senescence Kit (Cell Signaling Technology) according to the manufacturer's manual.

Immunoflourescence Staining..

Blast colonies were handpicked using a fine glass pipette and immobilized onto positively charged glass slides through cytospin at 500 revolutions per minute. After fixation in a mixture of methanol and acetone (Fisher Scientific, Pittsburgh, PA, http://www.Fishersci.com/) for 20 minutes, the slides were incubated in 0.1% Triton X-100 in phosphate buffered solution (PBS) for 10 minutes, and washed three times with PBS (5 minutes per wash). For RPE markers, cells were fixed for 10 minutes with 2% paraformaldehyde in PBS, permeabilized with 0.1% NP-40 substitute (Sigma) for 15 minutes, and blocked for 1 hour with 10% goat serum in PBS. For BC markers, antibodies against human CD71 (BD Pharmingen, San Diego, CA, http://www.bdbiosciences.com), CXCR4 (GeneTex), (GeneTex, Irvine, CA, http://www.gentex.com/) and TPO-receptor (R&D Systems Inc., Minneapolis, MN, http://www.rndsystems.com), CD31, CD34 (DAKO), and KDR (R&D Systems) were used. For RPE markers, antibodies to bestrophin (Novus Biologicals, Littleton, CO, http://www.novusbio.com/), MITF and CRALBP (Abcam, Cambridge, U.K., http://www.abcam.com), and Pax6 (Covance, Princeton, NJ, http://www.covance.com) were used. For apoptotic BC identification, antibody binding to a cleaved form of caspase-3 (Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com) was used. Cells were then incubated with primary antibodies in blocking solution overnight at 4°C, washed with PBS (3 times for 5 minutes each), followed by incubation with flourochrome-conjugated secondary antibodies at room temperature for 1 hour. Cell nuclei were stained with DAPI (1 μg/ml) in PBS and the slides were mounted using Aquamount (Polysciences Inc., Warrington, PA, http://www.polysciences. com).

For RPE cell generation, analysis and imaging, see online supporting information Methods.


HiPSCs Are Capable of Differentiating into Hemangioblasts/BCs, Hematopoietic, and Endothelial Cells In Vitro

A series of studies was carried out using well-characterized hiPSCs lines generated by lentiviral expression of the transcription factors Oct-4, Sox-2, Nanog, and Lin28 (Thomson reprogramming method; cell lines IMR90-1, Foreskin-1-1, Foreskin-4-1 and Foreskin-4-3) [1] as well as retroviral expression of Oct-4, Sox-2, Klf4, and c-Myc (Yamanaka reprogramming method; cell lines rv-hiPS01, rv-hiPS02, rv-hiPS03 and rv-hiPS04) [6]. These hiPSC lines expressed the standard markers of pluripotency, and formed teratomas after inoculation in severe combined immunodeficiency (SCID) mice as reported [1, 6], and were all morphologically indistinguishable from hESCs (Fig. 1A, 1B) after transferring and growing in our laboratories. Using our previously optimized methods [21, 22], we first carried out a series of experiments to determine whether the hiPSC lines could generate BCs. The hiPSCs were differentiated into EBs under conditions optimized for the development of BCs, and individual EB cells were plated in blast growth/expansion medium (BGM) for the development of blast colony. EB cells from all hiPSC lines developed blast colonies 6 days after plating. Variable efficiencies were observed for the different hiPSC lines (Table 1), although the blast colonies that developed from all of the lines were much smaller in size, and the cells in the colonies were not as healthy as those generated from hESCs (Fig. 1B, online supporting information Fig. 1A). Nevertheless, as observed for BCs derived from hESCs (Fig. 1A) [21], these BCs from hiPSCs expressed hemangioblast markers CD71, CXCR-4, and Tpo receptor (Fig. 1B), but most of these BCs did not express CD31, CD34, and KDR, as demonstrated by immunocytochemical analyses. After replating in hematopoietic colony-forming media supplemented with a spectrum of cytokines for 10 to 14 days, erythroid (CFU-E), myeloid (CFU-G and CFU-GM), and macrophage (CFU-M, not shown) hematopoietic cell colonies developed (Fig. 1C). BCs also formed CFU-MK with the potential to differentiate into megakaryocytes (Fig. 1C, iv). To determine their endothelial potential, hiPSC-derived BCs were plated on fibronectin-coated plates and cultured in endothelial cell differentiation medium. The cells attached to the surface and grew as adherent cells (Fig. 1D, i), which formed capillary-vascular-like structures on Matrigel (Fig. 1D, ii), expressed high levels of vWF, PECAM-1 (Fig. 1D, iii) and VE-cadherin (Fig. 1D, iv), and took up acetylated low-density-lipoprotein (Ac-LDL, Fig. 1D, iv). These results clearly demonstrate that hiPSCs generated with different combinations of reprogramming factors using different delivery systems (lentivirus and retrovirus) are capable of differentiating into hemangioblasts, hematopoietic, and endothelial lineages under serum-free conditions, which is consistent with previous reports [17–20].

Figure 1.

Differentiation of hiPSCs and hESCs toward hemangioblasts/blast cells and hematopoietic cells. (A): Hemangioblasts/blast cells (ii, ×400) derived from hESCs (i, ×100) were stained with CD71 (×1,000), CXCR4 (×1,000) and Tpo receptor (×1,000). (B): Hemangioblasts/blast cells (ii, ×400) derived from hiPSCs (i, ×100) were stained with CD71 (×1,000), CXCR4 (×1,000) and Tpo receptor (×1,000). (C): Multiple types of hematopoietic CFU developed from hiPSC-blast cells 14 days after replating; ×100 for CFU-E, CFU-GM, and CFU-Mk; ×40 for CFU-G. (D): Hemangioblast/blast cells derived from hiPSCs differentiated into endothelial cells (i, ×200); hiPSC-derived endothelial cells formed vascular-like network on Matrigel (ii, ×40), expressed vWF (red) and CD31 (iii, green, ×400), as well as VE-cadherin (iv, green) with low-density lipoprotein uptake capability (iv, red ×400). Abbreviations: CFU, colony-forming units; CFU-E, colony forming units-erythrocyte; CFU-G, colony forming units-granulocyte; CFU-Mk, colony forming units-megakaryocyte.

hiPSC-Derived BCs Exhibit Apoptotic Phenotype

As summarized in Table 1, we observed that the efficiency of blast colony formation from all tested hiPSC lines was dramatically lower than that from all tested hESC lines. Even with the best cell line (IMR90-1) observed in our testing system, substantially fewer BCs developed compared with all tested hESC lines which displayed excellent reproducibility, although some variations were observed among different hESC lines [21, 22]. In contrast with hESC-derived BCs, their hiPSC counterparts displayed an apoptotic morphology (Fig. 1A, 1B, online supporting information Fig. 1A). Caspase-3, the major protease involved for the execution of apoptosis, is produced as an inactive enzyme protein, and is cleaved by other initiation caspases such as caspase-8 and caspase-9 during the onset of apoptosis [28]. The presence of the cleaved form of caspase-3 in the hiPSC-derived BCs is a clear indicator that the cells were undergoing apoptosis. More than 20% of the BCs generated from hiPSCs were positive for the cleaved form of caspase-3 (Fig. 2A, green) with fragmented nuclear morphology (Fig. 2A, iii, arrow); in distinct contrast, less than 1% of BCs derived from hESCs stained positive for the cleaved form of caspase-3. The adjacent nonapoptotic BC (CD71+, red) has an intact nucleus. As previously reported, hESC-derived BCs always possess remarkable expansion capability in BGM [23]. Figure 1A shows a typical blast colony generated from hESCs, which expands dramatically from day 6 through 10 (online supporting information Fig. 1A), whereas blast colonies derived from hiPSCs (Fig. 1B, online supporting information Fig. 1A) are significantly smaller and expand slightly, if at all, during days 6 through 10.

Figure 2.

Apoptotic blast cells derived from hiPSCs and early senescent phenotype of endothelial cells from hiPSCs. (A): Blast cells (BCs) stained for CD71 (i, red), apoptotic marker cleaved caspase-3 (ii, green), DAPI (iii, blue), and merged image (iv). The arrow indicates an apoptotic BC with fragmented nucleus, ×1000. (v): Percentage of apoptotic BCs derived from hESCs and hiPSCs, p < .01. (B): Endothelial cells differentiated from hESCs (i–iii) and hiPSCs (iv–vi) were examined for acetylated low-density lipoprotein (Ac-LDL) uptake (i and iv, red, ×200) and cell senescence marker β- galactosidase (blue, ii and v, ×100, and iii and vi, ×200). (vii): Endothelial cells from hiPSCs exhibited limited growth/expansion capability, p < .001 (total endothelial cells derived from 36,000 BCs, and expanded for 30 days). (viii): Endothelial cells from hiPSCs committed early senescence, p < .001 (percentage of senescent endothelial cells). Abbreviations: hiPSCs, human induced pluripotent stem cells; hESCs, human embryonic stem cells.

Endothelial Cells Derived from hiPSCs Senesce in Early Growth Phase

Because of the very low efficiency for most hiPSC lines, we focused on the best performed IMR90-1 line for endothelial cell experiments. HiPSC (IMR90-1)-derived BCs were seeded at high density in fibronectin-coated plates with medium optimized for endothelial cell differentiation. The cells quickly attached to the plate (within 3 to 4 hours) and differentiated into endothelial cells (Fig. 1D). However, the cells grew very slowly (Fig. 2B), if at all, and most of them displayed a large, senescent-like phenotype (Fig. 2B, v and vi). More than 50% of the cells expressed β-galactosidase (Fig. 2B, viii) which has been shown to be a reliable marker for cellular senescence [29]. In contrast, hESC-derived endothelial cells exhibited normal endothelial cell morphology and a robust growth rate (Fig. 2B, i–iii). Less than 5% of the hESC-derived cells stained positive for β-galactosidase (Fig. 2B, iii).

HiPSC-Derived BCs Show Dramatically Reduced Expansion and Hematopoietic CFU-Forming Capability

Although BCs from hiPSCs were capable of forming multiple types of hematopoietic CFUs, the efficiency was also dramatically reduced compared with hESC-BCs (Fig. 3A, i-vi). When 103 BCs from either hESCs or hiPSCs were plated for the development of hematopoietic CFUs, 181.3 ± 34.7 and 76.4 ± 12.3 versus 2.9 ± 1.5 CFUs formed after 14 days for BCs from MA01 and H1 hESCs, and IMR90-1 hiPSCs that is the most efficient cell line we observed, respectively (26-fold to 63-fold difference; Fig. 3A, vii, p < .001). Furthermore, colony forming unit-granulocyte, erythrocyte, monocyte and megakaryocyte (CFU-GEMM) very rarely developed from cells derived from hiPSCs, and most CFUs were also much smaller than those generated from hESCs (Figs. 1C, 3A, online supporting information Fig. 1B). Very rarely, large CFU-G developed from hiPSC-derived BCs, as shown in Fig. 1C.

Figure 3.

Reduced colony-forming units (CFU) formation, erythroid cell differentiation and expansion capability of hiPSC-blast cells. (A): A typical well of CFU assay from 1,000 hESC-blast cells (i–iii) and hiPSC-blast cells (iv–vi). (vii): CFU development of blast cells derived from IMR90-1 hiPSCs (n = 10), H1 (n = 5), and MA01 (n = 5) hESCs, p < .001. (B): (i–v): Erythroid cells obtained from hESC-blast cells after three cycles (15 days) of differentiation and expansion. Erythroid cells were stained with benzidine for hemoglobin (brown) and Giemsa (blue, ×200), and CD235a (red, ×1,000). (vi–x): Erythroid cells obtained from hiPSC-blast cells after three cycles (15 days) of differentiation and expansion. Erythroid cells were stained with benzidine for hemoglobin (brown) and Giemsa (blue, ×200), and CD235a (red, ×1,000). (C): Quantitative results of erythrocyte expansion. Total fold increase: hiPSCs = 3.03 versus hESCs = 5,071.74. Abbreviations: BCs, blast cells; CFC, colony forming cells; CFU, colony-forming units; hiPSCs, human induced pluripotent stem cells; hESCs, human embryonic stem cells; vs., versus.

We previously demonstrated that erythroid cells can be reproducibly generated from hESC-BCs with high efficiency [23], suggesting that hiPSCs could potentially serve as a source for generating O Rh(-) universal blood. Since most hiPSC lines showed very low efficiency of developing BCs, we then focused on the best performing hiPSC line, IMR90-1 for erythroid expansion. Using these methods, 3 × 104 hiPSC(IMR90-1)- and hESC (MA01)-BCs were subjected to three cycles of erythroid cell expansion/differentiation. As shown in Figure 3B and 3C, a more than 5 × 103-fold increase in the number of cells was obtained from hESC-BCs; these cells morphologically resembled erythroblasts and stained positive for hemoglobin and CD235a (Fig. 3B). Although positive expansion was obtained for the first cycle (the initial 3 to 5 days), hiPSC-derived BCs showed negative expansion during both the second and third cycles of expansion (threefold expansion after 15 days, Fig. 3C).

RPE Cells Derived from hiPSCs Begin Senescing in the First Passage

To determine whether the observed phenomenon is limited to hemangioblastic lineages, we generated RPE cells from both hESCs and hiPSCs as previously described [30]. The pigmented clusters developed from IMR90-1 and iPS(Foreskin)1-1 cells were isolated manually and digested with collagenase, the cells were plated on gelatin-coated tissue culture plates in media optimized for RPE cells. Using this strategy, more than 100 RPE lines were established using more than two dozen hESC lines, including WiCell-, Harvard-, and Advanced Cell Technology-derived lines (Table 1 and online supporting information Table 1). We observed that hiPSCs initially differentiated into RPE cells in a manner similar to hESCs (Fig. 4). The cells isolated from these cultures showed the usual expression of RPE markers (online supporting information Fig. 2A, 2B), as demonstrated by immunostaining and reverse transcription polymerase chain reaction analysis. However, after RPE cells were isolated from the differentiating cultures, noticeable differences between hESC- and hiPSC-RPE cells were observed. The initial hESC-RPE cultures (passage 0) usually had very few cells, but these cells grew robustly; one well of a four-well plate at passsage 0 produced approximately 9 to 12 million cells at passage 2 (even if started from a few hundred cells isolated from differentiated hESC cultures). When RPE cells proliferate in culture, they lose their pigmented polygonal morphology and turn into spindle-shaped cells. After the cells reach confluence, they regain their RPE phenotype. The hESC-derived cells were routinely passaged at least 2 or 3 times before we observed the presence of a few abnormal, large, and often fibroblast-like cells that still had RPE markers. Although abnormal-looking cells began to appear during passage 3, the hESC-derived RPEs continued to proliferate and regain their RPE phenotype until at least passage 5 (online supporting information Fig. 2C). In contrast, hiPSC-RPE cells grew very slowly and often failed to fill the original well of the four-well plate. In some cases, hiPSC-derived RPE cells failed to regain normal RPE morphology at passage 1 (similar to that observed with hESC-RPEs at much later passages). In striking contrast with hESC-derived RPEs, abnormal senescent-like cells (large vacuolated cells with irregular morphology) were observed in all hiPSC-cultures at passage 1. β-galactozidase staining confirmed that these cells were senescent (Fig. 4B, iii).

Figure 4.

Derivation and characterization of retinal pigment epithelium (RPE) cells from human induced pluripotent stem cells (hiPSCs) and human embryonic stem cells (hESCs). (A): RPE cells derived from hESCs. (i): Differentiated RPE clusters before isolation, ×40; (ii): RPE cells at passage 0, ×200; (iii): RPE cells at passage 2 stained with β- galactosidase, ×200. (B): RPE cells derived from hiPSCs. (i): Differentiated RPE clusters before isolation, ×40; (ii): RPE cells at passage 0, ×200; (iii): RPE cells at passage 2 stained with β-galactosidase (blue), ×200.


Consistent with previous published reports [12–14, 17–20], we show that well-established hiPSC lines are capable of generating hemangioblasts/BCs, endothelial, and hematopoietic cells, as well as other lineages, such as RPE cells, that share a high similarity to their hESC counterparts. A major hallmark of hESC-derived cells is the high recovery and proliferative capability of the cells, including, for instance, the ability of the latter to generate functional oxygen-carrying red blood cells on a large scale (10 to 100 billion cells per six-well plate hESCs) [23]. However, in contrast with hESC-derivatives, hemangioblasts/BCs and RPEs generated from hiPSCs displayed limited expansion capability and apoptosis morphology. When compared side by side, hiPSC-BCs demonstrated extremely limited hematopoietic colony-forming and erythroid lineage expansion and differentiation capabilities; endothelial lineage-specific differentiation studies of hiPSC-BCs revealed reduced endothelial clonogenic capability, limited growth rate, and early senescence. Although many hESC lines studied showed consistent, reproducible BC-generation capability, limited success was observed with most hiPSC lines studied here with the exception of IMR90-1. We therefore believe that vigorous screening of all established hiPSC lines using well established differentiation protocol combined with phenotypical characterization of derivatives will ultimately help narrow down the list of true hiPSC lines and also provide powerful technical criteria for deriving new breeds of hiPSC lines.

The molecular mechanisms responsible for these differences remain elusive. Further investigation will be necessary to determine the nature and extent of these abnormalities. The lineage-specific differentiation of hESCs has been well studied, highly efficient, and reproducible methodologies have been reported for differentiating hESCs toward all three germ layers [31]. Although hiPSC lines have been derived from a variety of somatic cell types using diverse reprogramming strategies, only limited reports on mesoderm, ectoderm, and definitive endoderm lineage-specific differentiation are available. Using EB formation in the presence or absence of serum, Lengerke et al. [20] and Ye et al. [19] demonstrated hematopoietic development of hiPSCs derived by retroviral overexpression of reprogramming factors. Similarly, using coculture with BM stromal cell line OP9 and cell sorting, Choi et al. [17] reported successful hematopoietic/endothelial lineage-specific differentiation using hiPSC lines derived using lentiviral forced expression of Oct4, Sox2, Nanog, and Lin28 in human fibroblasts. However, none of these studies provided information on the proliferative capability of the progenitor cells. Choi et al. [18] recently generated mature myelomonocytic cells through coculture of hiPSCs with OP9 feeder cells, and the yield of differentiated cells from hiPSCs was found to be comparable with that obtained from H1 hESCs. However, only short-term (2 days) expansion data were presented. Other hiPSC lineage-specific differentiation studies, including generation of endothelial cells [32], hepatocytes [10], pancreatic insulin-producing β-cells [9], adipocytes [33], and retinal pigmented epithelial cells [11, 13, 14], focused primarily on the characterization of the differentiated cell population rather than progenitor cells during early lineage specification. One study, however, did compare the proliferation of hESC- and hiPSC-derived cardiomyocytes (partially purified from contracting EBs) through BrdU incorporation. Reduced cell proliferation was indeed observed, particularly in cardiomyocytes from late EBs [8].

There have been numerous reports, both in the scientific literature and the media, suggesting that iPSC may be identical to ESCs. Global gene expression profiles of hiPSCs and hESCs were compared for 32,266 transcripts [2]. Although the results showed a similar global gene expression profile, there were 1,267 genes (approximately 4% of the total interrogated genes) with a greater than fivefold difference in expression level between hiPSCs and hESCs. It is unclear whether this difference in gene expression will have any serious consequences in normal cellular function, or if it is simply the result of cell culture conditions. A genome-wide study to compare mouse and human iPSCs with ESCs was published recently by Chin et al. [34]. This is by far the most comprehensive study using comparative genome hybridization arrays to uncover subkaryotypic genome alterations and gene expression changes; coding RNA and micro RNA profiling, to determine changes in expression of small noncoding RNAs, and histone modification profiling, to determine whether epigenetic changes correlate with gene expression differences. Based on a vast amount of data, they concluded that early- and late-passage hiPSCs are not identical to their embryonic-derived counterparts [34]. A significant number of hiPSC lines were found to have abnormal expression of imprinted genes, which were found to be stable in hESCs [35]. Doi et al. [36] recently showed that differential methylation of tissue-specific CpG island shores could distinguish hiPSCs, hESCs, and fibroblasts, clearly demonstrating the differences between hiPSCs and hESCs. Furthermore, vaccination with hESCs generated a broad immunologic and clinical response against cancer cells in mice, whereas vaccination with hiPSCs failed to confer tumor protection in the animals [37]. These results suggest that there is, at least, a differential expression of oncofetal antigens between hESCs and hiPSCs. Although substantial evidence points toward the fact that hiPSCs and hESCs are not identical, several groups have reported the generation of viable, fertile live-born progeny by tetraploid complementation using murine iPSCs, indicating that these iPSCs maintain a pluripotent potential that is very close to murine ESCs [38–40]. Furthermore, by using a humanized sickle cell anemia mouse model, Hanna et al. [41] showed that mice can be rescued after transplantation with hematopoietic progenitors obtained in vitro from autologous iPS cells. These studies clearly suggest the differences between iPS cells derived from humans and mice.

The mechanisms behind the slower growth and early cellular senescence observed here are unknown. However, recent studies clearly indicate the involvement of p53/p21 pathways in the somatic cell reprogramming process [42]. Lin28, one of the transcription factors used to generate the hiPSCs used in the present study, has also been shown to be tumorogenic [43, 44]. Although the proviral insertion of transgenes of reprogramming transcription factors are reported to be silenced in the full reprogrammed cells [1, 6], it remains to be seen whether they could be reactivated upon initiation of certain lineage differentiation protocols. We have observed the nuclear localization of LIN28 in hiPSC-derived BCs, which has not been observed in hESC-derived BCs (unpublished data). It will, therefore, be important to determine whether the abnormalities observed in the present study apply to hiPSCs generated using other transcription factors and reprogramming strategies, especially hiPSC lines derived without exogenous vector and transgene sequences [3–6].


We wish to thank Loyda Vida (University of Illinois at Chicago), Timothy Kelley and Hyungjoon Kim (Stem Cell and Regenerative Medicine International), and Jennifer Shepard (Advanced Cell Technology) for their technical assistance; and Dr. J. Thomson for hips cell lines.


Q.F., S.J.L., and Y.C. are employees of Stem Cell and Regenerative Medicine International. I.K. and R.L are employees of Advanced Cell Technology. K.S.K. is an advisor of Stem Cell and Regenerative Medicine International. I.G., D.K., and G.R.H. declare that they have no competing financial interest in this study.