SEARCH

SEARCH BY CITATION

Keywords:

  • Hematopoiesis;
  • Vasculogenesis;
  • Vascular endothelial growth factor;
  • Hypoxia;
  • Embryonic development

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Vascular endothelial growth factor (VEGF) and the vascular endothelial growth factor receptors (VEGFRs) regulate the development of hemogenic mesoderm. Oxygen concentration-mediated activation of hypoxia-inducible factor targets such as VEGF may serve as the molecular link between the microenvironment and mesoderm-derived blood and endothelial cell specification. We used controlled-oxygen microenvironments to manipulate the generation of hemogenic mesoderm and its derivatives from embryonic stem cells. Our studies revealed a novel role for soluble VEGFR1 (sFlt-1) in modulating hemogenic mesoderm fate between hematopoietic and endothelial cells. Parallel measurements of VEGF and VEGFRs demonstrated that sFlt-1 regulates VEGFR2 (Flk-1) activation in both a developmental-stage-dependent and oxygen-dependent manner. Early transient Flk-1 signaling occurred in hypoxia because of low levels of sFlt-1 and high levels of VEGF, yielding VEGF-dependent generation of hemogenic mesoderm. Sustained (or delayed) Flk-1 activation preferentially yielded hemogenic mesoderm-derived endothelial cells. In contrast, delayed (sFlt-1-mediated) inhibition of Flk-1 signaling resulted in hemogenic mesoderm-derived blood progenitor cells. Ex vivo analyses of primary mouse embryo-derived cells and analysis of transgenic mice secreting a Flt-1-Fc fusion protein (Fc, the region of an antibody which is constant and binds to receptors) support a hypothesis whereby microenvironmentally regulated blood and endothelial tissue specification is enabled by the temporally variant control of the levels of Flk-1 activation.

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


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Author contributions: K.A.P.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; S.H.L.G.: conception and design, collection and/or assembly of data, manuscript writing; S.M.D.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing; K.C.: provision of study material, BL-CFC assay assistance; A.N.: provision of study material; P.W.Z.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript

Oxygen-mediated signaling is expected to initiate and modulate the development of the major oxygen transport-related tissues. Adaptive responses to hypoxia (O2 deprivation) are required for mammalian placentation [1], facilitating O2 and nutrient delivery to the rapidly growing embryo, and for the normal development and patterning of the cardiovascular system [2]. Furthermore, hypoxia stimulates pathways leading to the proliferation of endothelial cells [3] and hematopoietic progenitors [4, 5].

As a regulator of oxygen homeostasis, hypoxia-inducible factor (HIF) induces a network of genes related to angiogenesis, erythropoiesis, and glucose metabolism [6]. It allows dynamic control of cell survival and function in response to changing environmental stimuli [7]. Although the molecular mechanisms are not completely understood, hypoxia plays a crucial role in establishing hemogenic mesoderm [8] and in the subsequent generation of hematopoietic and endothelial cells. Hypoxic regions have been identified in the yolk sac mesoderm, where HIF-1α and vascular endothelial growth factor (VEGF) colocalize to induce blood vessel formation [9]. These observations point to oxygen-regulated activation of growth factor signaling as an important component of the hemogenic niche associated with the emerging embryonic blood islands. A recent review elaborates on oxygen availability during embryonic development [10].

VEGF is a potent endothelial cell mitogen and has a central role in hematopoiesis, vasculogenesis, and angiogenesis [11, [12]13]. There are two VEGF receptors, vascular endothelial growth factor receptor (VEGFR)-2 (or fetal liver kinase-1 [Flk-1], also known as kinase insert domain-containing receptor [14]), which transduces signals [15], and VEGFR1 (or fm-slike tyrosine kinase receptor-1 [Flt-1] [16]), which has been described as a nonsignaling ligand trap [17, 18]. Flt-1 has a higher affinity (Kd 10–30 pM, where Kd is the equilibrium constant for dissociation [19]) for VEGF than its actively signaling counterpart Flk-1 (Kd 75–760 pM [14, 19]). VEGF and its receptors Flt-1 and Flk-1 play a crucial role in generating mature endothelial and hematopoietic cells. Embryos heterozygous for VEGF die by embryonic day (E) 10 because of malformations in the vascular and blood system [11, 13]. These defects are more pronounced in embryos homozygous for the VEGF deletion [11], suggesting a dose-dependent regulation of fetal vascular development by VEGF.

Homozygous Flk-1 null (Flk-1−/−) embryos fail to form blood islands and vessels [20]. In vivo and in vitro analyses have demonstrated that although the requirement of Flk-1 is cell-intrinsic for endothelial cells, the failure to generate blood cells may be indirect and associated with the inability of the early mesodermal cells to migrate to hematopoietic conducive sites [15, 21]. Recent cell labeling experiments suggest that a continuum of Flk-1 expression is important in the specification of the hemogenic mesoderm toward blood or endothelial cells [22]. Mechanisms to control the activation of Flk-1 may thus play an important role in the embryonic hematopoietic niche. Flt-1 homozygous null (Flt-1−/−) mice generate both endothelial and hematopoietic cells but die from an abnormally high number of endothelial cells that fail to form an organized vascular network [23, 24]. In contrast, deleting only the intracellular domain of Flt-1 (by generating Flt-1 tyrosine kinase-deficient mice [Flt-1TK−/−]) yields apparently normal development [18]. Together these results have led to speculation that the function of Flt-1 is to act primarily as a VEGF “sink” or signaling modulator [25].

Considered in combination, the phenotypes of Flt-1−/−, Flt-1TK−/−, and Flk-1−/− mice suggest that VEGF action during embryogenesis depends on the strength and timing of the activation of its signaling receptor(s), a parameter that can be manipulated by changing ligand or receptor availability. Given that VEGF secretion can be broadly regulated by the microenvironment, we hypothesized that oxygen could act as a developmentally relevant signal to control the VEGF-ligand-receptor signaling threshold (VEGF-LIST) [26]. We further speculated that because of its action as a nonsignaling VEGF sink, Flt-1 expression may modify the VEGF-LIST as a function of oxygen tension. We tested these hypotheses using quantitative analysis of blood and endothelial development from embryonic stem cells and mouse embryos.

We demonstrated for the first time that sFlt-1 is an important mediator of hematopoietic development in a developmental-stage-dependent and oxygen-dependent manner. Early transient Flk-1 signaling occurs in a hypoxic microenvironment because of low levels of sFlt-1 and high levels of VEGF, resulting in the enhanced generation of hemogenic mesoderm and blast-colony-forming cells (BL-CFCs). Sustained or delayed Flk-1 activation resulted in enhanced endothelial cell output. On the other hand, delayed inhibition of Flk-1 signaling resulted in an increase in blood progenitors from the hemogenic mesoderm. Our results demonstrate a mechanism whereby hypoxia, VEGF, and its cell surface-bound and soluble receptors (sFlt-1) collaborate to regulate the development of the hematopoietic and endothelial lineages.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Cells

Mouse R1, VEGF−/− [11], Flt-1−/− [23], and Flk-1−/− [15] embryonic stem cells (ESCs) were maintained in a humidified incubator with 5% CO2 at 37°C as described previously [27]. G4 and Flt-1-Fc mutant ESCs were grown on mitomycin C-treated mouse embryonic fibroblasts (derived from TgN (DR4)1 Jae embryos) and maintained as described above.

ESCs were differentiated in either stirred suspension [28] or liquid suspension [27]. Treatments included 25 ng/ml VEGF (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com; R&D Systems Inc., Minneapolis, http://www.rndsystems.com) and 5 μg/ml SU1498 (Flk-1 tyrosine kinase inhibitor; Sigma-Aldrich).

E7.5 embryos were harvested, treated with 25 mg/ml collagenase with 20% fetal bovine serum in medium for 60–90 minutes at 37°C, and dissociated with a 26-gauge needle before culture on OP9-green fluorescent protein stromal cells. Cells were cultured in standard OP9 medium [29] supplemented with SU1498, VEGF, or mrFlt-1-Fc (R&D Systems) for 5 days prior to CFC and endothelial cell (EC) assessment.

Encapsulation Process and Bioreactor Culture

Mouse ESC aggregates were formed by generating a single-cell suspension at 3 × 105 cells per milliliter in ESC medium and incubating for 1 day [28]. The Cellferm-pro system (DasGip, Julich, Germany, http://www.dasgip.com) was used for stirred suspension culture of encapsulated ESCs under controlled conditions. Vessels were filled with 125–200 ml of ESC medium without leukemia inhibitory factor and inoculated with 5 × 105 to 1 × 106 encapsulated ESCs. Supernatants were collected and analyzed by enzyme-linked immunosorbent assay (ELISA).

Hematopoietic Cell Assays

Embryoid bodies (EBs) were dissociated by incubation (2 minutes, 37°C) in 0.25% trypsin-EDTA (Sigma-Aldrich). Yolk sacs from E8.5 embryos were digested with collagenase as described previously [30]. Individual cells were counted using a hemocytometer and analyzed by either myeloid-erythroid (ME)-CFC assay or flow cytometry [27]. ME-CFC assays were performed as recommended using M3434 medium (StemCell Technologies, Vancouver, BC, Canada, http://www.stemcell.com) with 20,000–100,000 test cells plated into 35-mm Greiner dishes (Greiner Bio-One, Frickenhausen, Germany, http://www.gbo.com/en). CFCs were scored by morphology 7 days after plating according to established criteria [31].

Blast-CFC assays were performed with 50,000–100,000 cells per 35-mm Greiner dish and scored by morphology 4 days after plating [32]. After 4 days in the assay, individual blast colonies were picked and replated into hemogenic-endothelial cell supportive medium [32], and/or total RNA was isolated using the Picopure kit (Ambion, Austin, TX, http://www.ambion.com).

Cells were prepared for flow cytometry as described [27]. The cells or isotype controls were incubated with 1:100 phycoerythrin (PE) anti-mFlk-1 (Avas 12∝1), fluorescein isothiocyanate (FITC) anti-mCD34 (RAM34), PE anti-mCD45 (30-F11), PE rat IgG2a,κ, FITC rat IgG2a,κ, and PE rat IgG2b,κ (BD Pharmingen, San Diego, http://www.bdbiosciences.com/index_us.shtml) prior to addition of 1 μg/ml 7-aminoactinomycin D (Molecular Probes, Eugene, OR, http://probes.invitrogen.com) in the final wash. The cells were analyzed on a Coulter Epics XL using Expo 32 ADC 1.1 software (Beckman Coulter, Fullerton, CA, http://www.beckmancoulter.com) or FlowJo (Tree Star, Ashland, OR, http://www.treestar.com). Positive staining was defined as fluorescence emission at >99.9% of the levels obtained by negative control cells from the same starting population.

ELISA

VEGF and sVEGFR1 concentrations in medium supernatants were measured using Quantikine Immunoassay Kits (R&D Systems) following manufacturer-provided protocols. VEGF ELISAs measure the concentration of VEGF in the supernatant. That includes both free VEGF and VEGF bound to sVEGFR1 (S.M. Dang, personal communication; R&D Systems). Similarly, sVEGFR1 ELISAs measure the concentration of both free sVEGFR1 and VEGF-bound sVEGFR1 (S.M. Dang, personal communication; R&D Systems).

Reverse Transcription-Polymerase Chain Reaction and Quantitative Reverse Transcription-Polymerase Chain Reaction Analysis

Total RNA was prepared using the Picopure kit or Trizol reagent (Invitrogen, Carlsbad, CA, http://www.invitrogen.com). Using Quantitect (Qiagen, Hilden, Germany, http://www1.qiagen.com), 0.25 μg (blast colony) or 1 μg (tissue/ESCs) of RNA was reverse-transcribed. The following primers were obtained from Primer Bank (pga.mgh.harvard.edu/primerbank): ID β1H1 globin-6680179a1, Brachyury-6678203a1, CD34-19526792a1, Flk-1-27777648a1, GAPDH-6679937a1, L32–2131294a1, Nanog-31338864a1, VEGF164-6678563a1. Primers for sFlt-1 and Flt-1 were obtained from Huckle and Roche (2004), 203-EX9-SS, 206-EX14-AS, and 227-INT13-AS [33]. These primers were used to detect sFlt-1 and Flt-1. Conditions for the quantitative polymerase chain reaction (PCR) were incubation at 50°C for 2 minutes and then 95°C for 10 minutes, followed by 40 cycles of 95°C for 30 seconds, 60°C for 60 seconds, and 72°C for 15 seconds using the Applied Biosystems 7900HT Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com). Data were analyzed using methods described by Ramakers et al. [34].

Endothelial Cell Assay

Individual day 4 EBs grown in serum-containing differentiation medium with or without treatments (25 ng/ml VEGF or 5 μg/ml SU1498) were plated on 250 μg/ml collagen IV (Sigma-Aldrich)-coated 96-well plates. After 24 hours, medium was exchanged and the day 5 EBs were treated with 25 ng/ml VEGF or 5 μg/ml SU1498 from day 5 to day 7, as described in the figure legends. Alternatively, day 7 EBs from hypoxic or normorxic bioreactors were plated and grown for 2 days in differentiation medium. Wells were fixed with 4% paraformaldehyde (PFA) on day 7 and stained for platelet-endothelial cell adhesion molecule-1 (PECAM; CD31) as described below. Images were taken on an Olympus SZ2-ILST microscope with a single magnification set between ×0.67 and ×4 (Olympus, Tokyo, http://www.olympus-global.com) with an Infinity 1 USB 2.0 camera (Zarbeco, LLC, Randolph, NJ, http://www.zarbeco.com); positive thresholds were set using ImageJ (http://rsb.info.nih.gov/ij/index.html), and PECAM-positive areas were analyzed quantitatively.

Transgenic Mice

ICR outbred stock mice (Harlan, Indianapolis, http://www.harlan.com) were used for dissections of E7.5–E8.5 cell cultures. Flt1-Fc mice were generated as described by George et al. [35] and mated with Cre deletor (Tg(ACTB-Cre)1Nagy) [36]. Embryos were genotyped by PCR with the following primers: for Cre: P1, GGTTATTGTGCTGTCTCATCA, and P2, ATATCCTGGCAGCGATCGCTA; and for Flt-1-Fc: P1, TGGTTGTAAGCCTTGCATAGTACAGTC, and P2, CTAGCTAGCTTTACCAGGAGAGTGGGAG.

Immunohistochemistry

Embryos were dissected in ice-cold phosphate-buffered saline and fixed overnight in 4% PFA at 4°C. Whole-mount analysis was performed between E7.5 and E9.5 with PECAM (MEC13.3; 1:100; BD Pharmingen) and goat anti-rat Ig-horseradish peroxidase (HRP) conjugate (1:500; Biosource International Inc., Camarillo, CA, http://www.invitrogen.com), using Tyramide-Cy3 amplification (NEL752; PerkinElmer Life and Analytical Sciences, Boston, http://www.perkinelmer.com). Embryos and yolk sacs were visualized using a Leica MZ16FA stereomicroscope (Leica, Heerbrugg, Switzerland, http://www.leica.com) with a QImaging 1300C digital camera (QImaging, Surrey, BC, Canada, http://www.qimaging.com). E7.5-derived ECs grown on OP9s or EB-derived ECs grown on collagen IV, were fixed in 4% PFA for 20 minutes at room temperature or overnight at 4°C. Following overnight blocking in 5% bovine serum albumin, primary PECAM (Armanian hamster MAB1398Z, clone 2H8) was added overnight (1:100; Chemicon, Temecula, CA, http://www.chemicon.com) prior to washing and addition of the secondary, Cy5 AffiniPure goat anti-Armenian hamster (1:1,000; Jackson Immunoresearch Laboratories, West Grove, PA, http://www.jacksonimmuno.com) or goat anti-rat Ig-HRP conjugate (1:500; Biosource), for 3 hours at 4°C. Hoechst was added for 10 minutes to stain the nuclei of cells stained with Cy5, and imaging was completed at room temperature with an ArrayScan VTI platform version 5.5.1.2-0.63x (build 268) automated fluorescent microscope (Cellomics, Pittsburgh, http://www.cellomics.com). Zeiss objectives were used (×10 numerical aperture (NA) ×0.3 or ×20 NA 0.4 Korr; Carl Zeiss, Jena, Germany, http://www.zeiss.com), and image analysis was completed using ImageJ (standardized contrast and threshold set to determine EC area).

Immunoprecipitation and Western Blot

Immunoprecipitation on EB proteins was performed using standard protocols with protein A (Sigma-Aldrich) and Flk-1 antibody (Ab2349; 1:100; Abcam, Cambridge, MA, http://www.abcam.com). Antibodies used in Western blots were mouse phosphorylated tyrosine (1:5,000; gift from Dr. Tony Pawson), rabbit Flk-1 (1:1,000; Abcam), and goat HRP-IgG (anti-mouse or anti-rabbit) (1:10,000; Bio-Rad, Hercules, CA, http://www.bio-rad.com).

Statistical Analysis

All data are reported as mean ± SD except data for quantitative reverse transcription (qRT)-PCR, which are reported as mean ± SEM. Experiments measuring differences between oxygen tensions were assessed using the paired two-tailed Student's t test with n ≥ 3 and α = 0.05 unless noted otherwise in text.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

ESC Blood and Endothelial Cell Output Are Correlated with VEGF Secretion Rates in Opposite Ways

Although Flk-1 expression has been associated with commitment to blood (Flk-1low) or endothelial (Flk-1high) cells [22], it is not clear whether control of the microenvironment could also influence the level of signaling activation of the receptor and consequently the development of endothelial and hematopoietic lineages. As a first step in testing the hypothesis that endogenous regulation of the LIST and subsequent hematopoiesis and vasculogenesis could be manipulated by the microenvironment, we examined the role of hypoxia using a bioreactor system (Fig. 1Ai, 1Aii) to recapitulate the low O2 environment thought to occur during development [37, 38]. Previous reports indicate that hypoxia (≤5% O2) enhances hematopoietic progenitor cell (HPC) output in the EB system [4, 5]. In our studies, ESCs were differentiated under a range of oxygen conditions; after 7 days, HPC frequency was measured using the ME-CFC assay (Fig. 1Bi), and EC development was quantified using PECAM staining of cell outgrowths (Fig. 1Ci). Analysis showed that ME-CFC output peaked at 2%–4% O2 tension (Fig. 1Bii), whereas EC development displayed the opposite trend over the same range (Fig. 1Cii). ELISA was used to demonstrate that, as has been suggested [39, 40], cell-specific VEGF secretion rates are proportional to oxygen (Fig. 1D). Together these results suggest that VEGF-mediated signaling regulates hematopoietic cell and EC development. To demonstrate that VEGF is a key player in the hypoxic effects on CFC/EC output, we also showed that continuous inhibition of Flk-1 with SU1498 significantly decreased both CFCs and ECs in hypoxic conditions. Alternatively, competition for VEGF through the addition of Flt-1-Fc also decreased CFCs in hypoxia relative to the control (Fig. 1E). We next sought to specifically implicate VEGF-mediated signaling in the observed differential fate outcomes.

thumbnail image

Figure Figure 1.. Hematopoietic progenitor cell (HPC) production, endothelial cell (EC) development, and VEGF secretion are functions of oxygen tension. A bioreactor system (Ai) was used to control oxygen tension at 20%, 8%, 4%, 2%, and 1% (Aii). EBs were harvested on day 7, and HPC frequency was assessed by the myeloid-erythroid CFC output on the basis of established morphological characteristics [27] (Bi). HPC frequency was measured as a function of oxygen tension (Bii). Day 7 EBs were plated on collagen IV, fixed 2 days later, and stained with platelet-endothelial cell adhesion molecule-1 (PECAM). PECAM-positive areas were analyzed quantitatively after a positive image contrast threshold was set. Scale bar = 500 μm (Ci). Day 7 EB-derived PECAM-positive areas were calculated and plotted as a function of bioreactor oxygen tension (Cii). Bioreactor medium supernatants were analyzed by enzyme-linked immunosorbent assay (ELISA) on day 7 to determine VEGF concentration. The cell-specific VEGF secretion rate (fg per cell per day) was calculated, and three independent experiments are shown (D). To establish that the VEGF produced in hypoxic conditions was affecting CFC/EC output, differentiating EBs were grown in hypoxia in the presence of Flk-1 inhibitor SU1498 (5 μg/ml) or with competitive Flt-1-Fc (2 μg/ml). CFC/EC output was compared with control (E). Abbreviations: CFC, colony-forming cell; ND, condition not determined; sFlt-1, soluble Flt-1; VEGF, vascular endothelial growth factor.

Download figure to PowerPoint

Mutant VEGF Receptor ESC Lines Demonstrate That Flt-1 Plays a Critical Role in Oxygen-Mediated Modulation of HPCs

As a first step in determining the mechanisms responsible for the observed oxygen-mediated modulation of blood development, we performed experiments using parental or engineered R1 ESC lines with homozygous deletions of VEGF, Flt-1, or Flk-1 and parental or genetically engineered G4 ESC lines that overexpressed Flt-1-Fc [35]. Blood and endothelial development from the VEGF and VEGFR mutant ESCs have not been investigated as a function of oxygen concentration. We reasoned that this analysis would reveal novel interactions and synergies between these molecules and microenvironmental control. In normoxia, VEGF−/− and Flk-1−/− mutants produced similar numbers of CFCs, whereas Flt-1−/− ESCs produced a significantly higher number of CFCs in comparison with wild-type control (Fig. 2Ai). These results suggest that Flt-1 may restrain CFC output in normoxic cultures. Under hypoxic conditions, wild-type control cells significantly increased CFC output (p = .00005), whereas VEGF−/− and Flk-1−/− ESCs exhibited modest increases (Fig. 2Ai). Interestingly, Flt-1−/− ESCs were completely defective in the hypoxic enhancement of CFC output (0.7-fold increase; p = .05). These data implicate Flt-1 in enhancing CFC generation under hypoxic conditions.

thumbnail image

Figure Figure 2.. Flt-1 is the major modulator of CFC production, and sFLt-1 may affect Flk-1 activation as a result of oxygen concentration. The CFC capacity of wild-type (WT) R1 cells and of their mutant derivatives VEGF−/−, Flt-1−/−, and Flk−/− was assessed in hypoxia and normoxia (Ai). On day 7, 105 cells were seeded. The CFC capacity of WT G4 cells and their derivative Flt-1-Fc mutant was assessed in hypoxia and normoxia (Aii). On day 7, 5 × 104 cells were seeded. Significant differences (p < .05) are indicated between hypoxic and normoxic conditions for each cell line (*), between normoxic WT and mutant CFC production (&), and between hypoxic WT and mutant CFC production (#). A representative fluorescence-activated cell sorting plot is shown, demonstrating the Flk-1 isotype control (shaded) and positive Flk-1 expression in hypoxia (solid line) and normoxia (dashed line) (Bi). Shown is the kinetic Flk-1 expression profile in normoxia and hypoxia measured by flow cytometric analysis (Bii). Kinetic VEGF and sFlt-1 concentration profiles (pM) under normoxia and hypoxia were determined by enzyme-linked immunosorbent assay (ELISA) and the calculated ratio of molecules of VEGF/sFlt-1 is shown (n ≥ 3) (C). Abbreviations: CFC, colony-forming cell; sFlt-1, soluble Flt-1; VEGF, vascular endothelial growth factor.

Download figure to PowerPoint

In a second series of studies we used control G4 cells and a derived mutant line, Flt-1-Fc, that overexpresses the protein. There was not a significant difference between the G4 parental line and Flt-1-Fc in normoxia (p = .13), but in hypoxia Flt-1-Fc generated significantly more CFCs compared with both the Flt-1-Fc (p = .002) and G4 (p = .03) lines (Fig. 2Aii). In sharp contrast to the reduced CFC output observed relative to wild-type with the VEGF−/−, Flt-1−/−, and Flk-1−/− mutants, the Flt-1-Fc ESCs produced significantly more CFCs in hypoxia than G4 wild-type controls. Together these data implicate, for the first time, Flt-1 in the oxygen-mediated regulation of blood progenitor cell development.

Hypoxia Influences Flk-1 Activation Via the Secretion of Flt-1 and VEGF

The previous results suggest that oxygen concentration may influence the dynamic interaction between VEGF and its receptors. Gene expression analysis indicated that VEGF, Flt-1, and Flk-1 expression changed with time and that VEGF was affected by oxygen concentration (supplemental online Fig. 1). To assess these differences at the protein level, we first measured Flk-1 cell-surface receptor expression using flow cytometry (Fig. 2Bi). In normoxia, EB Flk-1 expression was detected on day 3, peaked on day 5, and then declined for the remainder of the culture (Fig. 2Bii). The profile of Flk-1 expression at 4% O2 was similar in shape to that at 20% O2; however, Flk-1 expression peaked 1 day earlier. The discrepancies between protein and mRNA measurements for Flk-1 may result from the translation of the gene or from differences in the processing and secretion of the extracellular proteins. This suggests that there is a developmental window that is sensitive to oxygen concentration, during which Flk-1 signaling may be important.

In contrast to Flk-1, which was detected on the surface of the cells, we could find Flt-1 only upon analysis of medium supernatants. ELISA revealed the production of a soluble form of Flt-1 (sFlt-1), strongly supporting its role as an inhibitor because of its lack of intracellular signal transduction capacity. Measured values for the concentrations of sFlt-1 are provided in supplemental online Figure 2. The dynamic competition for the VEGF produced in hypoxia in contrast to normoxic conditions is best shown by the ratio of VEGF/sFlt-1 (the ratio of the ligand and its decoy) (Fig. 2C). The VEGF dimer binds two molecules of receptor [41]; we calculated the concentration of protein as the number of molecules of VEGF or sFlt-1, respectively, in the bulk medium, and assumed that binding occurred in homodimer form. This ratio (VEGF/sFlt-1) was initially high in hypoxia, but the ratio dropped toward 1 by day 5. In contrast, in normoxia the ratio was low until day 4; after this time, the ratio increased and approached 1. These results are consistent with our earlier results that suggest blood development proceeds in a VEGF/Flk-1-independent manner in normoxia and that the increase in blood development during hypoxia proceeds from high VEGF signaling early during culture. Thus, sFlt-1 may influence cell fate during ESC differentiation via the temporal and microenvironment-dependent control of ligand (VEGF) availability.

Mimicking the Flt-1-Mediated Control of ESC Fate Under Normoxic Conditions: Effects on Blood and Endothelial Cell Output

We have proposed that sFlt-1 plays a modulating role in the oxygen-mediated enhancement of blood progenitor generation (and has an inverse effect on endothelial cells). Differences between VEGF and sFlt-1 concentration profiles in hypoxia suggest that CFC generation is enhanced by early activation of Flk-1, followed by sFlt-1-mediated competitive inhibition of Flk-1 activation. To further explore and validate the proposed mechanism, Flk-1 activation was controlled in a time-dependent manner under normoxic conditions using R1 wild-type ESCs. Treatments were provided to either activate (with VEGF) or inhibit (with SU1498) Flk-1 signaling, and generation of both HPCs and ECs was evaluated. Flk-1 activation and SU1498 inhibition was confirmed by immunoprecipitation with Flk-1 and immunoblotting with a phosphorylated tyrosine kinase antibody (supplemental online Fig. 3).

As expected, early VEGF treatment (days 0–5) generated more hemogenic mesoderm compared with the untreated control; HPC frequency and EC growth were 2.3 ± 0.8-fold and 2.3 ± 0.6-fold higher, respectively (Fig. 3A). The observation that blood and endothelial cells develop in close proximity in yolk sac blood islands supports a hypothesis that these cells originate from a common precursor, the hemangioblast [42]. An in vitro equivalent of this precursor can be measured using the BL-CFC assay [32, 43]. We confirmed that hypoxia can enhance BL-CFC output [4] on the basis of colony morphology (Fig. 3Bi, 3Bii) and that the addition of VEGF to normoxic cultures had a similar effect (supplemental online Fig. 4). BL-CFC identity was confirmed with a gene expression panel, the majority showing CD34 and the lack of brachyury expression characteristic of BL-CFC [32, 43], and functional CFC and EC assays (supplemental online Fig. 4B, 4C). Kinetic analysis of BL-CFC activity in our cultures showed a peak at day 4–4.5 of differentiation (data not shown). Together, these data suggest that VEGF-mediated signaling (either direct or mediated by hypoxia) acts to increase the hemogenic mesoderm pool during the first 4–5 days of differentiation.

thumbnail image

Figure Figure 3.. Control of Flk-1 activation affects CFC and EC output in a developmental stage-specific manner. Shown is a comparison of wild-type CFC or EC generation in response to treatment patterns in normoxia. Treatments were normalized with respect to the unsupplemented control medium. Treatments included continuous SU Flk-1 kinase inhibitor (SU0–7); VEGF supplied for the first 5 d of culture (V0–5), continuously (V0–7), or from d5 to d7 (V5–7); and VEGF for 5 d followed by SU (V0–5 SU5–7) as indicated on the x-axis (A). *, significant difference between the treated condition and untreated controls (p < .02; n ≥ 3); #, significant difference between the indicated treatment condition and VEGF d0–5 (p < .05; n ≥ 3). A representative blast colony is shown after 4 days of growth in the blast assay (Bi). Single cells were plated into blast medium at 60,000 cells per 35-mm plate from d2.5–4.5 EBs. Scale bar = 100 μm. Cells from d4 EBs cultured in hypoxia produced more blast-CFCs than in normoxia (Bii). Cells were plated at 100,000 cells per dish, and bars indicate SD of a representative experiment. Shown is representative EC morphology after fixation and platelet-endothelial cell adhesion molecule-1 (PECAM) staining on d7 of the indicated treatments (C). R1 cells were treated for 7 days as indicated in each of the headings (Ci–Cvi), and the Flk−/− cell line was used as a negative control in untreated medium (D). Scale bar = 500 μm. Abbreviations: CFC, colony-forming cell; d, days; EC, endothelial cell; SU, SU1498; V, vascular endothelial growth factor; VEGF, vascular endothelial growth factor.

Download figure to PowerPoint

To isolate Flk-1-mediated differentiation effects on the hemogenic population, we exposed cells treated for 5 days with VEGF either to 2 more days of VEGF or to the Flk-1 signaling inhibitor SU1498 (VEGF d0–5 SU1498 d5–7). HPC frequency was modestly increased in VEGF d0–5 SU1498 d5–7 conditions, whereas the concomitant EC response decreased significantly relative to day 0–5 VEGF treatment (Fig. 3A). In contrast, if VEGF-mediated signaling was provided continuously or only for days 5–7, an increase of ECs and a significant decrease of HPCs were observed (Fig. 3A). This suggests that the level of Flk-1 signaling after the hemogenic mesoderm is formed (i.e., day 5) influences the resultant number of hematopoietic or endothelial cells. The low levels of VEGF-independent blood development that we observed in the VEGF−/− and Flk-1−/− ESC lines were also seen in the wild-type ESCs treated for 7 days with 5 μg/ml specific Flk-1 tyrosine kinase inhibitor SU1498 (IC50 of 0.27 μg/ml [44, 45]) (Fig. 3A).

Striking differences in EC outgrowth morphology were also observed as a function of VEGF activation or inhibition. Baseline levels of EC development were observed in wild-type ESCs cultured in serum (Fig. 3Ci), whereas Flk-1−/− ESCs showed little EC development (Fig. 3D), a result also seen when wild-type ESCs were treated with SU1498 continuously (Fig. 3Cii). Highly branched EC networks arose from VEGF-treated cultures with continuous VEGF supplementation (Fig. 3Civ) or VEGF supplementation from day 5 to day 7 only (Fig. 3Cv). The presence of VEGF early in culture (days 0–5) followed by Flk-1 inhibition by SU1498 resulted in the generation of only compact areas of ECs (Fig. 3Cvi).

Controlling Flk-1 Activation of Primary E7.5-Derived Cells Alters Hematopoietic and Endothelial Outputs in a Manner Similar to That Observed in EB Differentiation

Thus far our results have been limited to ESC differentiation. Although it is an interesting and useful model system, we next sought to determine whether the stage-dependent control of Flk-1 activation could elicit similar fate changes in primary cells. We established ex vivo cultures from wild-type E7.5 embryos. This developmental stage corresponds to day 4–5 EBs [46], and it allowed us to evaluate fate decisions in response to Flk-1 activation or inhibition. CFC development from this cell source consisted primarily of monocyte and erythroid colonies (Fig. 4A). A dose-dependent enhancement in CFC number per embryonic-derived input cell was obtained from the loosely adherent layer of embryonic cells treated with Flk-1 antagonists (Fig. 4Bi, 4Bii). In contrast, exogenous VEGF resulted in a significant dose-dependent suppression of CFC output (Fig. 4Biii). Strikingly, an opposite effect was seen in the analysis of EC development, where the Flk-1 antagonists (mFlt-1-Fc and SU1498) suppressed EC output (Fig. 4Ci, 4Cii), whereas VEGF enhanced EC output (Fig. 4Biii). Visual inspection showed that VEGF enhanced outgrowth and branching from EC colonies, whereas the mrFlt-1-Fc and SU1498 treatments repressed this growth (Fig. 4D). The differences in the branching morphology of the ECs in response to altered Flk-1 signaling are consistent with earlier observations of vessel formation in the EB system [47].

thumbnail image

Figure Figure 4.. Modulation of hematopoietic and endothelial development from primary embryo-derived cells as a function of altering Flk-1 activation and inhibition. Cells from E7.5 embryos were cultured on OP9-green fluorescent protein (GFP) feeders for 5 days. Representative myeloid and erythroid colonies from E7.5 cultures are shown (A). Cells were treated with mrFlt-1-Fc (Bi), SU (Bii), or VEGF (Biii) at various concentrations to modulate development in a manner similar to the EB system. CFC capacity, in comparison with untreated controls, was enhanced by treatment with Flt-1-Fc (0.1, 0.3, and 1.0 μg/ml) or SU (0.25, 0.5, and 1.25 μg/ml). Conversely, VEGF treatment (10, 25, and 40 ng/ml) diminished CFC capacity. *, significant difference between treatment and control conditions (p < .05); n ≥ 3. The adherent layers of the cultures were stained for platelet-endothelial cell adhesion molecule-1 (PECAM) and imaged using quantitative immunofluorescence to quantify the EC coverage. Flt-1-Fc (Ci) and SU (Cii) supplementation decreased EC generation, whereas VEGF (Ciii) treatments enhanced EC formation compared with untreated controls *, significant differences between treatment and control EC coverage (p < .05); n ≥ 3. Two representative fields illustrating the EC treatment response are shown (D). The EC network was enhanced with VEGF treatment and reduced with the Flk-1 antagonists SU and Flt-1-Fc in comparison with untreated control cultures Nuclear dye (Hoechst), blue; OP9-GFP cells, green; PECAM-Cy5+ ECs, red. Scale bar = 100 μm. Abbreviations: CFC, colony-forming cell; Co, serum control condition; E, embryonic day; EC, endothelial cell; SU, SU1498; VEGF, vascular endothelial growth factor.

Download figure to PowerPoint

Overexpression of Flt-1-Fc In Vivo Disrupts Vascular and Hematopoietic Development

We observed that oxygen modulates the timing and extent of sFlt-1 and Flk-1 expression and that the temporal modulation of Flk-1 activation by sFlt-1 can influence blood and endothelial development in vitro and ex vivo. The in vivo relevance of the proposed model is supported by the phenotypes of embryos with deficient Flk-1 signaling, such as Flk-1−/−, Vegfa+/, and Vegfa/ mice [11, 13, 20] and Vegfalo/lo hypomorphs [48]. Controlling the oxygen concentration of developing embryos was not feasible; thus, we perturbed the natural balance of factors in our proposed mechanism and analyzed the effects of expressing the extracellular domain of Flt-1 (Flt-1-Fc [13, 35, 49]) in transgenic mice [33]. We confirmed the presence of the involved factors in E9.5 wild-type and mutant embryos (Fig. 5A).

thumbnail image

Figure Figure 5.. Flt-1-Fc overexpression mimics loss of Flk-1 activation in vivo. Shown is reverse transcription-polymerase chain reaction of VEGF, Flk-1, sFlt-1, and Flt-1 for embryonic day (E) 9.5 WT and Mut. Pl was used as the positive control (A). Whole-mount E9.5 embryos of WT control and mutant (Flt-1-Fc overexpression) are shown (B). The mutants demonstrated reduced growth and did not survive to term. Platelet-endothelial cell adhesion molecule-1 (in red) highlights the absence of a structured developing vasculature in the embryo proper (C) and the yolk sacs (D) of mutant embryos compared with WT littermates. CFCs generated from three individual yolk sacs are shown (*, p < .00001) (E). Histological sections through the embryos depict the reduced caliber of the dorsal aorta in the caudal aspect in mutant embryos in contrast to WT littermates (F). Scale bar = 100 μm. Abbreviations: CFC, colony-forming cell; DA, dorsal aorta; ES, embryonic stem; Mut, mutant embryos; Pl, placenta; sFlt-1, soluble Flt-1; VEGF, vascular endothelial growth factor; WT, wild-type.

Download figure to PowerPoint

The phenotype of the Flt-1-Fc-expressing embryos recapitulated the VEGF−/− phenotype [11], demonstrating a severe reduction of vasculature as observed from whole-mount images (Fig. 5B) or by PECAM expression in the embryo proper (Fig. 5C) and yolk sacs (Fig. 5D). The reduction of hematopoietic development was quantitatively demonstrated with a significant difference in ME-CFCs from E8.5 yolk sacs; wild-type animals generated 88.9 ± 11.5 colonies per 10,000 yolk sac cells, whereas Flt-1-Fc mice generated 2.3 ± 2.4 colonies (p < .0001; Fig. 5E). This is consistent with the significant reduction in CFCs observed in the VEGF−/− and the SU1498-treated ESCs (Figs. 2, 3). In addition, sections of the mutant embryos demonstrated the lack of the dorsal aorta and absence of hematopoietic cells (Fig. 5F). Together, the overexpression of Flt-1-Fc in the early embryo appears to strongly counteract the effect of endogenously produced VEGF, leading to lethality at E9.5.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Our results reveal dual roles for Flk-1 signaling that manifest at different stages of development and in response to changing microenvironments. During early embryogenesis, VEGF production results in Flk-1 activation and signaling that induces expansion of hemangioblasts prior to embryonic day 7.5 or day 5 in hypoxic culture of EBs in vitro. Flk-1 activation is not inhibited at this stage because of the lower expression of sFlt-1 relative to VEGF (summarized in Fig. 6). After this time, a greater number of HPCs are generated because of competitive inhibition of VEGF by sFlt-1 and the resultant decrease in Flk-1 activation. If Flk-1 activation remains high, a substantial network of endothelial cells is promoted.

thumbnail image

Figure Figure 6.. Schematic of the proposed model. As demonstrated, hypoxia increases VEGF production, which positively affects the generation of hemogenic mesoderm. The cell output or fate is then modulated through the competition of sFlt-1 for VEGF and the resultant effects on Flk-1 signaling. High signaling promotes endothelial cells, whereas low signaling promotes hematopoietic progenitors. The competitive balance is affected by VEGF and by sFlt-1 concentrations, as well as Flk-1 expression. Abbreviations: HPC, hematopoietic progenitor cell; sFlt-1, soluble Flt-1; VEGF, vascular endothelial growth factor.

Download figure to PowerPoint

sFlt-1 acts as an effective signaling modulator by regulating the availability of free VEGF in the microenvironment [17, 47], leaving Flk-1 as the primary signaling receptor for VEGF [50, 51]. The development of hemogenic mesoderm and the resulting differentiation into its derivative tissues depend on both the level and timing of Flk-1 signaling. This was clearly demonstrated by the early lethality of Flt-1-Fc-overexpressing transgenic embryos by their deficiency in hematopoiesis and vasculogenesis. We detected increasing levels of sFlt-1 in the medium supernatant during the EB differentiation assay and demonstrated that after day 5 inhibition of Flk-1 signaling favored CFC generation, whereas high levels of signaling enhanced ECs (Fig. 3A). Similar responses were seen from E7.5 cells following 5 days of Flk-1 inhibition or exogenous VEGF treatments on a hematopoietic-supportive stroma (Fig. 4). Thus, sFlt-1 was an effective VEGF sink and reduced Flk-1 signaling in both the EB and ex vivo system, effectively enhancing HPC outputs. In contrast to the almost complete suppression of CFCs from mutant E9.5 yolk sacs (Fig. 5E), differentiating Flt-1-Fc ESCs did not suppress hematopoietic progenitor cell generation and in fact enhanced CFCs in hypoxia. Although we do not fully understand these differences, we can speculate that changes in protein expression and secretion profiles, extracellular matrix (ECM) capture/diffusion effects, and/or cell migration effects will influence the competitive balance of sFlt-1-mediated inhibition and Flk-1 activation. The mutant embryos have enhanced mRNA expression of VEGF (Fig. 5A; qRT-PCR data not shown), and the production of Flt-1-Fc by ESCs may take some time before it is capable of blocking signaling, thus acting more like the VEGF/SU1498 treatments (Fig. 3A) to boost CFC output. Additional in vivo and in vitro studies are required to understand this apparent disparity.

Vasculogenesis is described as the de novo formation of blood vessels [52], whereas angiogenesis is typically described as the generation of new vessels from existing vessels driven by EC proliferation [53]. We and others [54] have noted that VEGF positively affects EC specification from the hemogenic mesoderm through Flk-1 signaling; under sustained VEGF supplementation, cells differentiate robustly into ECs. We suggest that Flk-1 signaling inhibits generation of blood and stimulates EC generation from hemogenic mesoderm. The VEGF signaling pathway is also important in regulating EC proliferation and branching morphogenesis [55]. Early restriction of Flk-1 signaling activity during differentiation formed dense masses of endothelial cells that underwent minimal branching morphogenesis (Fig. 3Ci, 3Cii). These dense masses were also apparent when initially high Flk-1 signaling activity was followed by signaling inhibition (Fig. 3Ciii, 3Cvi). In addition, ECs/angioblasts may develop independently from the hemangioblast [56]. We may have observed this phenomenon when low Flk-1 activation was followed by VEGF treatment (day 5–7; Fig. 3Cv). Cells that developed in a largely VEGF-independent manner (normoxic control) were still capable of differentiating into branched EC networks.

We propose that the inhibition of Flk-1 signaling in hemangioblasts allows these cells to produce blood progenitors. This result is consistent with earlier evidence that VEGF-treated Flk-1+ mesoderm generates ECs at the expense of hematopoietic commitment in an avian model [57]. We demonstrate that hypoxia can control this Flk-1 inhibition in an sFlt-1-dependent manner. Coupling this with earlier observations that Flt-1 expression level is differentially regulated both spatially and temporally during development [58], we suggest that in its soluble form Flt-1 regulates induction of hematopoiesis. It has already been proposed that a continuum of Flk-1 expression underlies blood and endothelial cell specification [22]; to this model we overlay microenvironmentally controlled Flk-1 activation to the same (as yet undefined) signaling threshold-mediated transcriptional mechanisms to specify these cell types. The difference in early (mesoderm initiation) versus late (mesoderm specification) Flk-1 signaling likely results from differences in the identity of the Flk-1-expressing cell populations at these developmentally distinct times. The earliest Flk-1+ EB-derived cells have been shown to express Brachyury and form BL-CFCs [59], an expression pattern and capacity that are lost in time in both the EB and embryo [32, 43]. Flk-1+ cells from the emerging hemangioblast population at E7.5 have the capacity to differentiate into hematopoietic and endothelial cells [60]. We have demonstrated that Flk-1 signaling after day 5 in the EB system or following E7.5 primary cell explant inhibited either HPC specification or HPC expansion. The later scenario is unlikely, as Flk-1 expression is rapidly lost upon commitment to the hematopoietic lineage [57, 60], and Flk-1 activation prior to its loss is necessary for migration of HPCs to sites supporting their survival and proliferation [61].

In hypoxic conditions, VEGF was able to bind Flk-1 with minimal inhibition until day 5, after which VEGF was competitively bound by Flk-1 and sFlt-1. This pattern of Flk-1 activation and inhibition was demonstrated to enhance CFC generation. The localized regulation of these factors may be important in establishing the hemogenic niche. In fact, this mechanism would allow for the independent (microenvironmentally controlled) initiation of hemogenic niches in different parts of the embryo. Alternative splicing of VEGF gives rise to several isoforms, differing in their expression patterns, biochemical biological properties [62], and affinity for ECM [63, 64]. The complexity and regulatory scope of this conceptual model must be extended to include other molecules, such as placental growth factor (which competes with VEGF to bind Flt-1 and thus promotes Flk-1 activation [65]) and other niche-related local control mechanisms. VEGF-independent pathways may also be involved in hypoxia-enhanced HPC generation, as VEGF−/− and Flk-1−/− ESCs were capable of modest enhancement (Fig. 2Ai).

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Our findings demonstrate that sFlt-1 is a vital modulator of VEGF availability and suggest that VEGF gradients and temporal regulation of Flk-1 signaling are important to ESC-derived hematopoiesis. Microenvironmentally regulated [66, [67]68] competition for Flk-1 activation may continue to affect and maintain hematopoietic-supportive microenvironments or, conversely, avascular areas [69] throughout life in both normal and pathological conditions [70, 71].

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

We thank D. van der Kooy for manuscript review, C. Park and M. Lynch-Kattman for advice on the BL-CFC assay, and G.H. Fong for the Flt-1−/− cells. P.W.Z. is a Canadian Research Chair in Stem Cell Bioengineering and A.N. is a Canadian Institutes of Health Research (CIHR) Senior Scientist. This work was accomplished with support from Natural Sciences and Engineering Research Council of Canada, CIHR, National Cancer Institute of Canada, and Ontario Graduate Scholarship in Science and Technology.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
SC-08-0237_Suppl_Fig_1.pdf214KSupplemental Figure 1
SC-08-0237_Suppl_Fig_2.pdf135KSupplemental Figure 2
SC-08-0237_Suppl_Fig_4.pdf192KSupplemental Figure 4
SC-08-0237_Suppl_Fig_Legends.pdf22KSupplemental Figure Legends
SC-08-0237_Suppl_Fig_3.tif1503KSupplemental Figure 3

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.