The SHB Adapter Protein Is Required for Normal Maturation of Mesoderm during in Vitro Differentiation of Embryonic Stem Cells*

Definitive mesoderm arises from a bipotent mesendodermal population, and to study processes controlling its development at this stage, embryonic stem (ES) cells can be employed. SHB (Src homology 2 protein in β-cells) is an adapter protein previously found to be involved in ES cell differentiation to mesoderm. To further study the role of SHB in this context, we have established ES cell lines deficient for one (SHB+/-) or both SHB alleles (SHB-/-). Differentiating embryoid bodies (EBs) derived from these ES cell lines were used for gene expression analysis. Alternatively, EBs were stained for the blood vessel marker CD31. For hematopoietic differentiation, EBs were differentiated in methylcellulose. SHB-/- EBs exhibited delayed down-regulation of the early mesodermal marker Brachyury. Later mesodermal markers relatively specific for the hematopoietic, vascular, and cardiac lineages were expressed at lower levels on day 6 or 8 of differentiation in EBs lacking SHB. The expression of vascular endothelial growth factor receptor-2 and fibroblast growth factor receptor-1 was also reduced in SHB-/- EBs. SHB-/- EBs demonstrated impaired blood vessel formation after vascular endothelial growth factor stimulation. In addition, the SHB-/- ES cells formed fewer blood cell colonies than SHB+/+ ES cells. It is concluded that SHB is required for appropriate hematopoietic and vascular differentiation and that delayed down-regulation of Brachyury expression may play a role in this context.

Definitive mesoderm and endoderm both arise from the primitive streak during gastrulation. An entity named mesendoderm representing a transitory state in mesodermal and endodermal development has been identified (1). Several factors have been shown to be involved in the commitment of mesodermal and endodermal lineages during this stage of development (2)(3)(4)(5), but the precise mechanism by which they exert their action remains elusive.
Differentiation of embryonic stem (ES) 3 cells to embryoid bodies (EBs) has been a useful tool in understanding early events during development due to the experimental complications of in vivo studies whenever a gene exerts multiple effects. Studies on in vitro differentiation of ES cells have corroborated the existence of mesendoderm (3). Another example of the usefulness of the in vitro EB system relates to studies of endothelial and hematopoietic ontogenesis. Differentiation of these lineages is initiated in specific structures composed of mesodermal cells (called blood islands) present in the yolk sac endoderm (6). Blood islands consist of primitive hematopoietic blood cells surrounded by a layer of endothelial cells and support primitive hematopoiesis between embryonic days 7 and 14.5. Endothelial cells in the yolk sac form the primitive vascular plexus in a process named vasculogenesis, and subsequently a more mature vascular system evolves by angiogenesis. The presence of several factors universal for both endothelial and blood cells, such as VEGFR-2, Tal-1, and CD34 (7,8), has suggested a common progenitor for primitive hematopoietic and endothelial cells that has been given the name hemangioblast (9). Despite numerous investigations, the mechanisms responsible for differentiation of endothelial and blood cells are still not completely understood. In addition to studies on differentiation of mesodermal lineages, EBs have been employed for studies of endodermal and ectodermal differentiation (10 -13).
SHB is a ubiquitously expressed adapter protein (14). Its N terminus contains proline-rich motifs (15), the central part includes a phosphotyrosine binding domain (16,17) and several tyrosine phosphorylation sites (17,18), and the C terminus contains an Src homology 2 (SH2) domain (15). The SHB SH2 domain mediates its interactions with several receptors such as the VEGFR-2 (19), FGFR-1 (15,20) (21), and platelet-derived growth factor receptor (15). The R522K mutation in the SHB SH2 domain renders it inactive. The other SHB domains bind to different cytosolic proteins and thus transduce signals downstream of the receptors. SHB is implicated in many cellular processes such as differentiation of endothelial cells (22) and T cell signaling after CD3 stimulation (21). Moreover, SHB is involved in apoptotic signaling in fibroblasts, islets of Langerhans, and endothelial cells (23)(24)(25). EBs expressing the R522K SHB mutant displayed a lower mRNA content of several liver and pancreatic markers (26). In addition, SHB was found to be essential for EB blood vessel formation (27), as EBs expressing R522K SHB failed to form normal blood vessels.
To address the role of the SHB adapter protein in development, we have established knock-out ES cell lines deficient for one or both SHB alleles. These ES cell lines were differentiated in vitro and aimed at understanding the impact of SHB on hemangiogenesis. The data suggest an important involvement of SHB in the development of mesoderm.

MATERIALS AND METHODS
Gene Targeting-The murine embryonic stem cell line GSI-1 derived from 129SvJ mice was inactivated at both (SHB Ϫ/Ϫ ) or one (SHB ϩ/Ϫ ) of the SHB alleles. These were inactivated in two subsequent steps. The first targeting vector contained three loxP sites. The first and the second loxP sites flanked the neomycin resistance gene, and the second and third loxP sites bordered the first exon (Fig. 1A). One of the clones in which homologous recombination had not occurred was used as an SHB ϩ/ϩ control. The clone in which homologous recombination had taken place and all three loxP sites were present (loxPSHB) was treated with Cre recombinase by in vitro transfection. One of the clones, which had lost both the neomycin resistance gene and the first SHB exon, was used as a maternal clone for the second transfection.
The second targeting vector contained an insertion of the neomycin gene into the first exon without loxP sites (Fig. 1A). Two sister clones to SHB Ϫ/Ϫ , in which homologous recombination had not occurred were used as SHB ϩ/Ϫ . Clones were screened by PCR with one primer outside the sequence of the construct (supplemental Fig. 1).
ES Cell Culture-ES cells were cultured in the presence of murine embryonic fibroblasts in Dulbecco's modified Eagle's medium/glutamax (Invitrogen), penicillin/streptomycin, 15% heat-inactivated fetal bovine serum (Invitrogen), 1ϫ nonessential amino acids (Invitrogen), sodium pyruvate (Invitrogen), monothioglycerol (Sigma), and recombinant leukemia inhibitory factor (Chemicon International, Temecula, CA) as described previously (26). Cells were grown at 37°C in 5% CO 2 . For ES cell differentiation, fibroblasts and leukemia inhibitory factor were omitted. ES cells were trypsinized (5 min at 37°C) and aggregated by the hanging drop technique (1200 cells/20 l of medium for 2 days (28)) to form EBs. These were grown in suspension for one more day and then plated on adherent tissue culture dishes. EBs were harvested 4, 6, 8, 12, and 21 days after the initiation of differentiation. Alternatively, EBs (after formation by the hanging drop technique for 2 days followed by 2 days of suspension culture) were grown as attachment cultures in glass chambers for an additional 7 days (yielding a total differentiation time of 11 days) and then stained for CD31 or CD41 to identify vascular structures or CD41-positive cells.
Real-time Reverse Transcription (RT)-PCR Analysis-Total RNA was isolated using the RNAeasy mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Gene expression was quantified by real-time RT-PCR using a onestep reaction using the SYBR Green RT-PCR kit (Qiagen, Hilden, Germany). PCR conditions were 50°C for 20 min and 95°C for 15 min followed by 45 cycles at 94°C for 15 s and annealing for 25 s at the various temperatures indicated followed by extension at 72°C for 15 s.
Alternatively, gene expression was quantified by a two-step reaction. M-MuLV reverse transcriptase (Finnzymes, Helsinki, Finland) was used to convert 1 g of total RNA into cDNA. For PCR reactions, the SYBR Green Taq Ready Mix TM (Sigma) was used. PCR conditions were initial denaturation (94°C for 30 s) followed by 45 cycles at 94°C for 10 s, and annealing for 25 s at the various temperatures indicated followed by extension at 72°C for 15 s. Primer sequences are shown in supplemental Table 1. The PCR reactions were run on a Light Cycler instrument (Roche Diagnostics). Transcription levels were normalized against ␤-actin.
Immunohistochemical Staining of EBs with CD31-Blood vessels were stained according to a protocol described previously (29). Briefly, EBs grown on glass chambers were washed with Tris-buffered saline (TBS) and fixed with 200 mM Tris, pH 6.6, 150 mM NaCl, 37 mM ZnCl 2 , 23 mM zinc-acetate, 3.2 mM calcium-acetate, and 0.2% Triton X-100 (4°C overnight). On the next day, EBs were washed with TBS and treated with 3% H 2 O 2 in methanol to block endogenous peroxidases. EBs were washed with TBS, and unspecific interactions were blocked in 0.1 M Tris-HCl/0.15 M NaCl/blocking buffer. The tissue was incubated with primary antibody, rat anti-mouse CD31 (BD Biosciences) at 1/1000, washed with TBS-Tween, and incubated with biotinylated secondary antibody (1/1000), goat anti-rat (Vector Laboratories Inc., Burlingame, CA), and again washed with TBS-Tween. EBs were incubated with streptavidin-horseradish peroxidase (PerkinElmer Life Sciences), washed with TBS-Tween and H 2 O, and the chromogen substance was added (AEC substrate kit, rat; Vector Laboratories Inc., Burlingame, CA). Alternatively, EBs were stained for CD41 (BD Biosciences), except that immunofluorescence (Alexa goat anti-rat 568; Molecular Probes) was adopted to visualize CD41-positive cells.
In Vitro Hematopoietic Differentiation-ES cells were differentiated according to a published procedure (30) using components from Stem Cell Technologies, Inc. (Vancouver, Canada). The procedure is depicted in supplemental Fig. 2. Two days prior to differentiation, ES cells were trypsinized and cultured in IMDM ES medium (Iscove's modified Eagle's medium with L-glutamine/HEPES (Invitrogen)), penicillin/ streptomycin, 15% heat-inactivated fetal bovine serum (Invitrogen), 1ϫ nonessential amino acids (Invitrogen), sodium pyruvate (Invitrogen), monothioglycerol (Sigma), and recombinant leu-kemia inhibitory factor (Chemicon International, Temecula, CA). After 2 days in IMDM ES medium, the ES cells had reached 25-50% confluence and were then trypsinized and washed twice with IMDM 5% fetal calf serum. ES cells were then transferred to Petri dishes at a density of 1500 cells/ml in IMDM differentiation medium (Iscove's modified Eagle's medium/L-glutamine/ HEPES (Invitrogen)), penicillin/ streptomycin, 15% heat-inactivated fetal bovine serum (Invitrogen), monothioglycerol (Sigma), ascorbic acid (Sigma), monothioglycerol (Sigma), L-glutamine (Sigma), and protein-free hybridoma medium (Invitrogen) to form EBs. These were grown at 37°C in 5% CO 2 for 7 days. Next, the media containing EBs were collected in Falcon tubes and the EBs allowed to sediment. Supernatants were removed, and EBs were trypsinized (3 min at 37°C), after which the trypsin was inactivated by serum and EBs dispersed several times by an 18-gauge needle and 2-3 times through a 23-gauge needle. ES cells were counted and mixed by vortexing with methylcellulose medium (MethoCult, StemCell Technologies Inc., Vancouver, Canada) and plated with a syringe at a concentration of 10 4 cells/ml. The MethoCult M3434 medium contains Iscove's modified Dulbecco's medium with methylcellulose for solid support, fetal bovine serum, bovine serum albumin, recombinant human insulin, human transferrin, 2-mercaptoethanol, l-glutamine, recombinant mouse stem cell factor, recombinant mouse interleukin-3, recombinant mouse interleukin-6, and recombinant human erythropoietin to support hematopoietic colony growth. Cell colonies were scored after an additional 7-10 days of culture, thus giving a total differentiation time of 14 -17 days.
Statistical Analysis-For realtime RT-PCR, ␤-actin-normalized cycle threshold values (as calculated by the Roche LightCycler version We screened for correct recombination by PCR with the pairs of primers indicated in supplemental Fig. 1, and the sequences amplified have been indicated. The sequence of one primer always originates from outside of the construct and that of the second primer from the unique part of the construct. Only the positive clones revealed bands of 4200 and 2500 bp from the left and right arms, respectively. The presence of the third loxP site was tested by cleavage at a unique PmeI site in the 2500-bp PCR product (data not shown). The first exon and neomycin gene were deleted after transfection with a vector expressing Cre recombinase; clones with deletion of the first exon were detected by PCR. Positive clones revealed a 340-bp PCR product. The second targeting vector (neoSHB) contained an insertion of the neomycin gene into the first exon. The correctly targeted locus revealed a 1400-bp-long PCR product. B, schematic representation of the SHB locus in the ϩ/ϩ, Ϯ, and Ϫ/Ϫ ES cells after homologous recombination. Correct insertion of the loxPSHB targeting vector (Fig. 1A) was followed by transfection with Cre recombinase to delete the first SHB exon (SHB ϩ/Ϫ ). These cells were then transfected with the second targeting vector and correct homologous recombination yielded SHB Ϫ/Ϫ . C, Western blot analysis. Complete loss of SHB expression in the SHB Ϫ/Ϫ clone and partial loss of expression in the SHB ϩ/Ϫ clone is shown. p44 extracellular signal-regulated kinase was used as a loading control.
3.5 software) were compared using a paired Student's t test based on absolute or relative differences. All real-time RT-PCR results are shown using logarithmic scales due to the exponential nature of the assay. In all other cases, a two-tailed Student's t test was performed using PractiStat software.

RESULTS
Targeted Inactivation of the SHB Gene-To analyze the function of SHB, we performed targeted inactivation of ES cells using the loxP-SHB construct (Fig. 1A). The first transfection produced only one clone of 408 in which homologous recombination had occurred with all three loxP sites present. Homologous recombination occurred in another 9 clones, but in these, only the two loxP sites bordering the neomycin gene were present and were thus useless for deletion of the first exon. The second transfection generated only one clone of 480 containing the correct insertion of the neomycin gene, i.e. in the second allele of the first exon. Another two clones contained insertion in the first allele, whereas the second allele was untouched. For SHB ϩ/Ϫ cells, two sister clones with random insertion of the targeting construct were chosen. The mutations resulted in loss of SHB protein expression (Fig. 1C).
Expression of Mesodermal Markers in SHB Ϫ/Ϫ EBs-Previous studies have suggested an involvement of SHB in vascular formation (27). We have presently analyzed the expression of two markers for early mesoderm at appropriate stages of spontaneous EB differentiation. The markers studied were Brachyury and Mix1. Brachyury and Mix1 are expressed in mesoderm, mesendoderm, and to a lesser extent in endoderm (31). Brachyury gene expression was rapidly down-regulated in the control EBs, whereas the SHB Ϫ/Ϫ clone exhibited a significant delay in this response (Fig. 2). A similar pattern of delayed downregulation of Mix1 gene expression was observed in the SHB-deficient clone, although the effect failed to reach statistical significance. The expression of later mesodermal markers relatively specific for the hematopoietic, vascular, and cardiac lineages was also studied. These were the hematopoietic markers Tal1 (32), CD41 (33,34), and CD45 (35), the vascular markers CD31 (36, 37) and VEGFR-2 (7,38), and the cardiac marker cardiacactin (39). In the control EBs, all of these, except CD31, showed increased expression with time, which in the cases of Tal1, CD41, and cardiac-actin, peaked on day 8 (Fig. 2). The SHB Ϫ/Ϫ clone showed reduced expression of most of these markers on day 6 or 8 of differentiation. On day 12 of differentiation, expression of Tal1 and CD41 peaked in the SHB Ϫ/Ϫ clone and was higher than that of the control (Fig. 2). When comparing the SHB ϩ/Ϫ clone with the control and the SHB Ϫ/Ϫ clones on day 12 of differentiation, it was observed that the loss of one SHB allele resulted in a response that was very similar to that of the complete SHB knock-out (Fig. 3). Particularly was expression of VEGFR-2, which has been shown to be essential for hematopoietic and vascular development (7), reduced in both the homozygous and heterozygous SHB knockout ES cells (Fig. 3) at this time point. In addition, FGFR-1 expression was determined. SHB interacts with FGFR-1, which is a receptor that also plays a role for vascular and hematopoi- etic development (40). The expression of FGFR-1 was downregulated at day 12 in the SHB Ϫ/Ϫ clone (Fig. 3). The combined data suggest decreased differentiation to mesodermal lineages in SHB Ϫ/Ϫ EBs as a consequence of delayed down-regulation of markers associated with early mesodermal development.
CD41 Protein Expression in SHB Ϫ/Ϫ EBs-The marker CD41 is commonly expressed in cells destined to undergo hematopoiesis (33), and despite the delayed up-regulation of expression, it showed an increased RNA content after 12 days of differentiation in the SHB Ϫ/Ϫ clone in comparison to the SHB ϩ/ϩ clone (Fig. 3). To corroborate this finding, the expression of the CD41 protein was studied also on the protein level by immunofluorescence microscopy of EBs differentiated for 11 (4 ϩ 7) days. CD41-positive cells derived from the SHB Ϫ/Ϫ and SHB ϩ/Ϫ clones were stained more intensely than those derived from the SHB ϩ/ϩ clone (Fig. 4). This indeed supports the view that CD41 mRNA is increased on day 12 in differentiating SHB Ϫ/Ϫ EBs and that this also results in increased protein expression.
Impaired Blood Vessel Formation in SHB Ϫ/Ϫ EBs after VEGF Stimulation-EBs differentiating in glass chambers constitute an excellent model for the study of vasculogenesis and angiogenesis. During vasculogenesis, attached EBs form a primitive vascular plexus. Primary vessels promote the formation of a mature blood vessel network with capillaries in a subsequent process called angiogenesis. Treatment of EBs with VEGF-A increases the migration of endothelial cells from the EB core into the marginal zone around the EB where the blood vessels form a peripheral vascular plexus (29) (Fig. 5B).
Our current intention was to study the potential of SHB Ϫ/Ϫ ES cells to form blood vessels, because SHB previously has been shown to influence EB blood vessel formation. SHB Ϫ/Ϫ or SHB ϩ/Ϫ EBs did not exhibit any significant morphological changes in their vasculatures compared with those of the SHB ϩ/ϩ EBs when cultured under basal conditions (Fig. 5A). These EBs formed a vasculature that was primarily concentrated to the central core of the EB with sparse vascular structures extending toward the periphery. However, after culture in the presence of VEGF, the SHB Ϫ/Ϫ and SHB ϩ/Ϫ EBs showed less developed peripheral plexa than the SHB ϩ/ϩ EBs with fewer capillary structures (Fig. 5B). Similar results were obtained with the other SHB ϩ/Ϫ clone (data not shown). This finding is in the line with the previous study suggesting a role of SHB in regulating the development of vasculature in differentiating EBs (27).
SHB Ϫ/Ϫ ES Cells Are Defective in Colony Formation Assay-Our RT-PCR data for the hematopoietic precursor markers CD41 and Tal-1 suggest a delayed up-regulation of gene expression but also an increased potential of SHB Ϫ/Ϫ ES cells to differentiate into blood cells as their expression was elevated on day 12 of differentiation. To test whether SHB is important for in vitro erythropoiesis, EBs from the SHB Ϫ/Ϫ , SHB ϩ/Ϫ , and SHB ϩ/ϩ clones were analyzed for their capacity to form definitive blood cell colonies. Cell colonies were scored after days 7-11 of differentiation in methylcellulose preceded by a 7-day EB differentiation period allowing a total of 14 -17 days of differentiation. SHB Ϫ/Ϫ EBs generated fewer blood cell colonies than wildtype EBs. In three of four experiments, the SHB ϩ/Ϫ EBs also formed fewer blood cell colonies than the SHB ϩ/ϩ clone (Fig.  6). This suggests that the main effect of SHB knock-out is delayed expression and decreased differentiation to the hematopoietic lineage.

DISCUSSION
In the present study, we have investigated the potential of SHB Ϫ/Ϫ ES cells to differentiate in vitro. ES cells commonly differentiate in vitro in three-dimensional structures known as EBs, which similar to the embryo, contain cells from all three embryonal layers, i.e. ectoderm, endoderm, and mesoderm, but lack the complexity typical for the normal development of the organism. Murine ES cells can complement but not replace developmental studies on murine embryos.
Inactivation of the first exon in the SHB gene causes loss of SHB protein expression. The homologous recombination was not very efficient in the SHB locus. The low yield in obtaining a SHB Ϫ/Ϫ clone after the second transfection could be due to the possibility that one allele is more permissive to the genetic replacement than the other one. Another explanation is that total SHB deficiency can cause less cell proliferation, and thus some SHB Ϫ/Ϫ clones will not expand sufficiently to be identified.
There is a close relationship between endoderm and mesoderm during development. Both of these germ layers originate in the primitive streak in a tissue referred to as mesendoderm (1). Also later during development, mesoderm and endoderm demonstrate interactions necessary for organ differentiation (4,41). Recently, the existence of mesendoderm was described also in vitro (3). Our combined data support the role of SHB in mesodermal differentiation. Down-regulation of the early mesodermal marker Brachyury was paralleled by a delayed increase in the expression of several later mesodermal markers, such as VEGFR-2, Tal1, CD41, and cardiac-actin. Furthermore, expression of CD45, which is a hematopoietic marker expressed subsequently to CD41, was not increased in the SHB knock-out cells. The markers represent three different mesodermal lineages, and thus the data would suggest a generalized defect in mesodermal differentiation beyond early Brachyury expression.
Expression of VEGFR-2 was significantly decreased in both SHB ϩ/Ϫ and SHB Ϫ/Ϫ EBs. VEGFR-2 is known as a factor involved in hematopoiesis and vasculogenesis in vivo and in vitro (29,38,42). When examining the SHB ϩ/Ϫ and SHB Ϫ/Ϫ EBs for their ability to form blood vessels, it was noted that both clones were able to form blood vessels under basal conditions. These results are in contradiction with our previous study using the SH2 domain inactive mutant of SHB, R522K-SHB (27). R522K-SHB ES cells failed to form blood vessels regardless of what growth factor was added. The more prominent phenotype of the R522K-SHB ES cells may have several explanations. First, R522-SHB competes not only with the SHB protein but also FIGURE 4. In situ detection of CD41 ؉ by immunofluorescence after 4 ؉ 7 days of differentiation. EBs were formed during preculture for 4 days and then cultured in chamber slides for 7 days before staining with anti-CD41. Original magnification in the left panels was 100ϫ. Boxes indicate enlarged areas shown in the corresponding right panels. with other members of this protein family, such as SHD, SHE, SHF, and SHG. Second, in the SHB Ϫ/Ϫ cells, these other members may compensate for certain SHB functions. Third, R522K-SHB may interfere with another pathway in which the wild-type SHB protein is not involved.
Both SHB ϩ/Ϫ and SHB Ϫ/Ϫ EBs form less developed vascular plexa after VEGF stimulation. SHB is involved in the differentiation of endothelial Immortomouse brain endothelial cells (22) and in VEGF-induced migration of porcine aortic endothelial cells by binding to phosphorylated tyrosine 1175 (1173 in mouse) in the VEGFR-2 (19). The data thus suggest impaired signaling downstream of VEGFR-2 in SHB-deficient cells during blood vessel formation.
CD41 is known as a marker for hematopoietic progenitors (34,43). The present finding that SHB-deficient EBs showed a delayed increase in expression that nevertheless reached higher levels of expression compared with control at day 12 could suggest an enhanced potential to differentiate to blood cells. This notion was not confirmed when performing a colony formation assay. Thus, the dominating feature is the delayed increase in CD41 expression that prevents further differentiation.
Increased CD41 expression without concomitant differentiation has previously been observed and was explained by the observed existence of two populations of CD41-positive cells, i.e. CD41 ϩ "bright" and CD41 ϩ "dim" (44). Thus, CD41 ϩ dim cells were shown to have the capacity to differentiate into hematopoietic progenitors contrary to CD41 ϩ bright cells, which showed a low capacity to differentiate. It is possible that the CD41-positive cells in the SHB Ϫ/Ϫ EBs resemble primarily the CD41 ϩ bright population of cells that fails to produce significant amounts of hematopoietic progenitors. Interestingly, the erythrocyte marker ␤-globin major, was decreased in R522K-SHB EBs. 4 FGF-2 promotes the formation of slender blood vessels extending from the center to periphery in cultured EBs (29). Nevertheless, FGFR-1 knock-out EBs also produce vascular structures (40). We have not observed any morphological changes in SHB Ϫ/Ϫ EBs after the addition of FGF-2 (data not shown). The fact that both FGF-2 addition and FGFR-1 knock-out increase vessel formation in EBs could explain the lack of effect of SHB Ϫ/Ϫ in this context, especially because SHB may mediate certain aspects of FGFR-1 signaling (20).
In conclusion, the present data suggest that the SHB adapter protein is important for the formation of vascular structures and hematopoietic differentiation. This regulatory function operates at a stage later than initial mesodermal specification. Modulating SHB activity could serve a role in altering the developmental potential of differentiating ES cells. FIGURE 6. Colony formation assay after 7 days of EB differentiation followed by a 7-10-day colony formation period. A, morphology of myeloid-like colonies. B, quantification of formation of blood colonies (means Ϯ S.E.; n ϭ 3-4; *** indicates p Ͻ 0.001).