Department of Medicine, Lillehei Heart Institute, University of Minnesota, Minneapolis, Minnesota, USA
Correspondence: Rita C. R. Perlingeiro, Ph.D., Lillehei Heart Institute, University of Minnesota, 4-124 Nils Hasselmo Hall, 312 Church St SE, Minneapolis, Minnesota 55455, USA. Telephone: 612-625-4984; Fax: 612-624-8118; e-mail: email@example.com
Author contributions: L.B.: designed and conducted experiments, analyzed and interpreted the data, and wrote the paper; M.I.: designed, conducted experiments, and analyzed the data; N.K.: conducted experiments; D.J.G. and M.K.: interpreted the data; R.C.R.P.: supervised the overall project, designed experiments, analyzed the data, and wrote the paper.
Endoglin (Eng), an ancillary receptor of the transforming growth factor beta (TGFβ) signaling pathway superfamily, has been well recognized for its important function in vascular development and angiogenesis since its discovery more than a decade ago. Recent studies show that this receptor is also critical for the emergence of blood during embryonic development, and that at E7.5, endoglin together with Flk-1 identifies early mesoderm progenitors that are endowed with hematopoietic and endothelial potential. These two lineages emerge in very close association during embryogenesis, and because they share the expression of the same surface markers, it has been difficult to distinguish the earliest hematopoietic from endothelial cells. Here, we evaluated the function of endoglin in hematopoiesis as development progresses past E7.5, and found that the hematopoietic and endothelial progenitors can be distinguished by the levels of endoglin in E9.5 yolk sacs. Whereas endothelial cells are Engbright, hematopoietic activity is primarily restricted to a subset of cells that display dim expression of endoglin (Engdim). Molecular characterization of these subfractions showed that endoglin-mediated induction of hematopoiesis occurs in concert with BMP2/BMP4 signaling. This pathway is highly active in Engdim cells but significantly downregulated in the Eng knockout. Taken together, our findings show an important function for endoglin in mediating BMP2/BMP4 signaling during yolk sac hematopoietic development and suggest that the levels of this receptor modulate TGFβ versus bone morphogenetic protein (BMP) signaling. Stem Cells2013;31:1893-1901
Endoglin (Eng or CD105) is a glycoprotein that acts as an ancillary receptor for various members of the transforming growth factor beta (TGFβ) superfamily [1, 2]. In endothelial cells, where Eng is abundantly expressed and has been mostly studied, this receptor plays an important role in the TGF-β-dependent responses by balancing activating and inhibitory signals [1, 3]. Upon binding of TGF-β to the TGF-β type II receptor (Tgfbr2), TGF-β type I receptor (Tgfbr1), also known as activin receptor-like kinase 5 (Alk5), is recruited and phosphorylated. The activated Alk5 phosphorylates and activates downstream effector Smads . In addition to Alk5, a second TGF-β type I receptor has been described, known as Alk1, whose expression is more restricted to endothelial cells . Several studies support the premise that endoglin is necessary for efficient TGF-β/Alk1 signaling [3, 6, 7], and thus acts as a modulator of TGFβ-dependent activation of Alk1 (stimulatory signal through Smad1/5) versus Alk5 (inhibitory signal through Smad2/3) [3, 8]. Endoglin has also been shown to interact with bone morphogenetic proteins (BMPs), including BMP2 and BMP7, through their respective ligand binding receptor kinases Alk3, Alk6, or Alk2, and BMPRII  but has not, to date, been shown to interact with BMP4.
Members of the TGFβ signaling pathway have been implicated in several critical processes of embryonic development. For instance, BMP signaling has been shown to be required for proper mesodermal formation and patterning [9-12], whereas TGF-β signaling is important for vasculogenesis [13-16]. Interestingly, several of these studies also documented impaired primitive erythropoiesis in the yolk sac (YS) of knockout embryos, such as for TGFβ1 , TGFβRII , and BMP4 . Endoglin-deficient (Eng−/−) mice die around E10.5 due primarily to vascular and cardiac defects [17, 18]. Even though it had been noted for more than a decade that YSs from Eng−/− mice display a pale appearance , this had been attributed to be an indirect result of insufficient blood flow. Initial experimental evidence for a role for endoglin in hematopoietic development came from studies using Eng-deficient differentiating embryonic stem cells (ESCs) [19, 20]; however, a hematopoietic role for endoglin in vivo was proven only recently: we have shown that endoglin is required for the emergence of blood progenitors from mesoderm during early embryonic development . We have also demonstrated that in E7.5 wild-type embryos, hematopoietic progenitors are enriched in the Eng+Flk-1+ subfraction, and that this fraction has a dual hematopoietic/endothelial molecular signature . This finding is consistent with and explains why the blood and endothelial lineages emerge in very close association during early embryonic development. Blood and endothelial cells are evident in the YS of the murine embryo at approximately 8 days postfertilization as blood islands: endothelial sacs encapsulating clusters of hematopoietic cells [22-24]. The two lineages also share the expression of several regulatory genes , antigens , and signaling receptors, such Flk-1 and Eng [20, 21, 27]. Thus, it has been challenging to distinguish hematopoietic from endothelial cells during YS development.
Although Eng has been detected in various hematopoietic cell types , including the long-term repopulating hematopoietic stem cell [29, 30], it remains unknown whether this receptor and consequently the TGFβ signaling pathway plays a role in hematopoietic ontogeny beyond primitive hematopoiesis. Here, we demonstrate that hematopoietic and endothelial cells from the YS are temporally distinguishable by the levels of Eng. At E9.5, dim expression of Eng (EngdimFlk-1neg) identifies hematopoietic progenitors within the YS, whereas Engbright marks endothelial cells (EngbrightFlk-1+). Importantly, these phenotypic changes correlate with differential expression of components of the TGFβ/BMP pathway, revealing a crucial role for this signaling pathway in definitive YS hematopoiesis, and proving insight on its molecular regulation.
Materials and Methods
All animals were handled in strict accordance with good animal practice as defined by the relevant national and/or local animal welfare bodies, and all animal work was approved by the University of Minnesota Institutional Animal Care and Use Committee. Eng-deficient embryos were generated by timed mating Eng+/− heterozygotes (kindly provided by Dr. Michele Letarte, University of Toronto); the morning of a positive plug was considered 0.5 dpc. Genotyping was performed as previously described . Time-pregnant wild-type CD1 mice were purchased from Charles River Laboratories International, Inc (Wilmington, MA, http://www.criver.com).
Embryos were removed from decidua and Reichert's membrane and dissected in phosphate buffered saline (PBS) supplemented with 10% fetal bovine serum (FBS). Embryos were staged according to the somite pair counting (6–12 sp for E8.5, and 18–25 sp for E9.5). All Eng−/− embryos analyzed presented the same somite pair counting as the littermate controls (18–23 sp).
Flow Cytometry and Cell Sorting
E8.5–9.5 YSs were dissociated in trypsin 0.25% for 3 minutes at 37°C, and analyzed by flow cytometry using monoclonal antibodies described in Supporting Information Table S1. YS cells were sorted based on Flk-1 and/or Eng expression on a FACS Aria (Becton–Dickinson, San Diego, CA, http://www.bdbiosciences.com) after addition of propidium iodide (BD, Pharmingen, San Diego, CA, http://www.bdbiosciences. com) to exclude dead cells. Data were analyzed using FlowJo software (Tree Star Inc., St Ashland, OR, http://www.treestar.com).
Hematopoietic Progenitor Assay
To determine the myelo-erythroid potential of each population, sorted cells were directly plated in hematopoietic medium for colony activity (M3434; StemCell Technologies, Vancouver, Canada, http://www.stemcell.com). Plated cells were cultured in a humidified incubator (5% CO2) at 37°C. Ery-P and definitive hematopoietic colonies were scored 6 and 10 days after plating, respectively. Alternatively, sorted cells were cocultured on OP9 stromal cells that had been preplated in six-well plates containing α-MEM medium (Gibco, Grand Island, NY, http://www.invitrogen.com) supplemented with 10% FBS (Gemini, West Sacramento, CA, http://www.gembio.com), mSCF (100 ng/ml, Peprotech, Oak Park, CA, http://peprotech.com), mIL3 (1.17 ng/ml, Peprotech), granulocyte colony-stimulating factor (mG-CSF, 100 ng/ml, Peprotech), vascular endothelial growth factor (mVEGF, 10 ng/ml, Peprotech), erythropoietin (hEpo, 2 U/ml, Stem Cell Technologies), angiopoietin-1 (hAng-1, 100 ng/ml, R&D, Minneapolis, MN, http://www.rndsystems.com), 5 × 10−5 M 2-mercaptoethanol, and 1% penicillin/streptomycin. Plated cells were cultured in a humidified incubator at 37°C in an environment of 5% CO2.
Dissected embryos were immediately fixed in 2% paraformaldehyde (PFA) for 30 minutes, incubated in 5%, and 15% sucrose solutions until the embryos decanted, and frozen in 7.5% Gelatin. Cryosections were incubated with primary antibodies, described in Supporting Information Table S1, overnight at 4°C. Reactions were detected following 1-hour incubation with appropriate secondary antibodies (Supporting Information Table S1). Staining was visualized by laser confocal scanning (Zeiss LM510, Germany, hhtp://www.zeiss.com).
Total RNA of E+F+ and E+F− sorted cells from E9.5 wild-type (WT) YSs (pool of ∼120 YSs) were isolated using RNAqueous-Micro kit (Ambion Inc., Grand Island, NY, http://www.invitrogen.com). For transcription analysis of E9.5 Eng−/−, Eng+/−, and Eng+/+ embryos, isolated YSs were submitted to RNA extraction using TRIzol reagent (Invitrogen, Grand Island, NY, http://www.invitrogen.com). qPCR was performed using TaqMan probes (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com). For Western Blot analysis of WT mice, three independent E9.5 pooled YS (∼180 embryos per pool) were sorted for E−F−, E+F+, and E+F− cells, which were then lysated by adding RIPA Buffer (ThermoScientific, Pittsburgh, PA, http://www.thermoscientific.com) and antiproteolytics (Roche, Branchburg, NJ, http://www.roche.com). For Western Blot analysis of Eng−/−, Eng+/−, and Eng+/+ embryos, isolated whole YSs were directly lysated in RIPA buffer, containing antiproteolytics. Protein lysates (30 μg for WT YS sorted cells and 15 μg for Eng−/−, Eng+/−, and Eng+/+ YS) were loaded in each lane, resolved in 8% SDS-PAGE, and transferred to membranes (Immobilon-P; Millipore, Billerica, MA, http://www.millipore.com), which were incubated overnight at 4°C with primary antibodies (Supporting Information Table S1). After incubation with HRP-conjugated secondary antibodies (Supporting Information Table S1), membranes were incubated with SuperSignal West Pico Chemiluminescent Substrate or Pierce ECL Western Blot Substrate (ThermoScientific) and exposed to film. The Western Blots were quantified using ImageJ (http://imagej.nih.gov/ij/index.html).
Differences between Eng−/−, Eng+/−, and Eng+/+ E9.5 samples or E−F−, E+F+, and E+F− sorted cells were assessed by Analysis of Variance (ANOVA).
Hematopoietic Precursor Activity Resides in Distinct Eng+ Subpopulations During Different Stages of YS Development
To determine whether endoglin continues to play an important role in hematopoiesis beyond its emergence at E7.5 , we began by analyzing the expression of this receptor in E8.5 and E9.5 YSs. We observed similar levels of Eng expression at both developmental stages (between 7% and 10%; Fig. 1A) and found that YS hematopoietic precursor activity was restricted to the Eng+ subpopulation at E8.5 and 9.5 (Fig. 1B). No hematopoietic colonies were detected when Eng− cells were directly plated into methylcellulose-based medium containing hematopoietic cytokines (Fig. 1B), even when 10-fold more cells were plated (data not shown). Similar results were obtained when these two cell fractions were cocultured on OP9 stromal cells in the presence of hematopoietic cytokines (Fig. 1C). After 5 days in culture, typical hematopoietic cell clusters were detected only in the Eng+ fraction (Fig. 1C). Fluorescence-activated cell sorting (FACS) analysis confirmed the hematopoietic nature of cell outgrowths, as evidenced by the expression of hematopoietic markers, in particular CD41, CD45, Ter119, and Mac-1 (Fig. 1D) as well as their ability to produce significant numbers of hematopoietic colonies, which consisted mostly of macrophages (Supporting Information Fig. S2). Conversely, Eng− cells displayed very limited capacity to generate hematopoietic progeny under these culture conditions (Fig. 1C, 1D and Supporting Information Fig. S2).
Since we have previously shown that Eng, in conjunction with Flk-1, identifies virtually all hematopoietic precursors at E7.5 , next we assessed whether this was also the case at later developmental stages (Fig. 2A). FACS analyses demonstrated that about half of Eng+ cells (between 4% and 5%) express Flk-1 (E+F+) at both E8.5 and E9.5 YSs (Fig. 2A). Interestingly upon this subfractionation we found differential intensity levels of endoglin, such that high levels of endoglin (Engbright) were restricted to E+F+ cells, whereas dim expression of endoglin coincided with the Eng+Flk-1− (E+F−) subfraction (Supporting Information Fig. S3A). This was further confirmed by immunofluorescence staining using confocal microscopy in staged embryos obtained from CD1 WT mice (Fig. 2C). In both E8.5 and 9.5 YSs, Eng+ cells were found abundantly in the endothelium, colocalized with Flk-1 (Fig. 2C). Because endoglin expression was found to be dim in Eng+Flk-1− cells, this fraction was more difficult to detect by immunofluorescence staining (Fig. 2C, arrowhead in upper panel). These and the Eng− subfractions were then further characterized for the expression of the hematopoietic markers c-Kit, CD41, CD45, and Ter119 [31-33] as well as endothelial markers, including VE-Cadherin, CD31, and Tie-2 . At E8.5, most hematopoietic and endothelial markers were found preferentially expressed in E+F+ cells (Fig. 2B and Supporting Information Fig. S3B). These results are somewhat similar to those at E7.5 ; however, at E8.5 the expression levels for CD41 and Tie2 are basically equivalent between E+F+ and E+F− cell fractions (Fig. 2B). Remarkably, at E9.5 we observed a change in this pattern as hematopoietic markers were found preferentially expressed in the E+F− fraction, while the E+F+ fraction was enriched for endothelial markers (Fig. 2B). The highest level of the pan-hematopoietic marker CD45 was found at E9.5 in the E+F− cell fraction, whereas E+F+ cells displayed similar low levels of this antigen at both time points (Fig. 2B). Almost no Eng+ cells were found to express the erythroid marker Ter119 in both E8.5 and E9.5 (Fig. 2B).
Accordingly, colony activity of E8.5 and E9.5 YSs based on endoglin and Flk-1 expression (E−F−, E−F+, E+F+, and E+F−) demonstrated that at E8.5, the majority of the definitive colony-forming cells were present in the E+F+ cell fraction (Fig. 3, left panel), whereas at E9.5, almost all hematopoietic activity was restricted to the E+F− cell fraction (Fig. 3, right panel). These data suggest that Eng identifies distinct cell populations as development progresses and that in E9.5 YSs, E+F+ and E+F− cell subfractions represent the endothelial and hematopoietic populations, respectively.
Molecular Analysis of Eng+Flk-1− and Eng+Flk-1+ Cells
Based on the distinct characteristics of E9.5 YS E+F− and E+F+ cell subfractions, we next determined whether differences could be observed in terms of expression of components of the TGFβ superfamily. Real-time PCR analyses demonstrated that E+F− cells, which encompass most of the hematopoietic progenitors at E9.5 YS, expressed distinctively high levels of Alk3, the type I receptor for BMP2/4 , as well as target genes of BMP signaling, such as Gata2 and Msx1 for BMP4 , and c-Myc for BMP2  (Fig. 4B, 4D). Since this was not observed in E+F+ cells, these data suggest that the BMP pathway is more active in the E+F− hematopoietic fraction. Regarding the TGFβ1 pathway, while Tgfbr2 was detected in both E+F− and E+F+ cell fractions, Alk1 was highly expressed in E+F+ cells, whereas Alk5 and its downstream target PAI-1  were selectively expressed in E+F− cells (Fig. 4B–4D). These findings make sense as Alk-1 is well known to be primarily expressed in endothelial cells and at specific sites of epithelial-mesenchymal interactions, whereas Alk5 is ubiquitously expressed in most tissues .
To confirm the activation of the TGFβ/BMP signaling pathways in these particular cell subpopulations, we evaluated the phosphorylation level of Smad1/5/8 in cell fractions sorted from E9.5 YSs by Western Blot (Fig. 4E) and immunofluorescence staining (Supporting Information Fig. S3). This revealed that Smad1 is highly phosphorylated in Eng+ cells (both E+F+ and E+F− fractions) when compared with cells that do not express endoglin (E−F−) (Fig. 4E, 4F). This was an expected outcome since Smad1/5 are downstream effectors of BMP/Alk3 and TGFβ1/Alk1 , which were differentially expressed in E+F− and E+F+ fractions, respectively.
Endoglin Is Required for BMP2/4 and TGFβ1 Signaling
Next we assessed the expression levels of components of the TGFβ1/BMP signaling pathways in whole YSs of Eng+/+, Eng+/−, and Eng−/− E9.5 embryos by qPCR. The ligands Bmp2 and Bmp4 were found downregulated whereas Tgfβ1 levels were unaffected in the absence of endoglin (Fig. 5A). Bmpr2 and Alk3 (Fig. 5B, 5C), common receptors for BMP2/4, and their target genes Cyclin D1, c-Myc, Gata-2, and Msx1 (Fig. 5D; for the last two see ) were significantly reduced in knockout YSs when compared with WT and heterozygous embryos. Whereas no changes were observed in the expression levels of Tgfbr2 (Fig. 5C), type I Tgfβ receptors Alk1 and Alk5 were significantly decreased. This was also the case for PAI-1, target gene for the Alk5 pathway. Id2, target gene for both BMP4 and TGFβ1, was also reduced in Eng-deficient mice.
To confirm that lack of Eng leads to reduced activity of TGFβ1/BMP2/4 signaling, we performed Western blot analysis of pSMAD1/5/8 in isolated Eng+/+, Eng+/−, and Eng−/− YSs from E9.5 embryos (four YSs each genotype). As observed in Figure 5E, 5F, the phosphorylation levels of SMAD1 were drastically decreased in knockout embryos when compared with the WT (approximately 2.5-fold) and heterozygotes (approximately twofold). These results demonstrate that Eng is required for proper activation of TGFβ1/BMP2/4 signaling, and consequently expression of important hematopoietic genes is regulated by this pathway.
The importance of TGFβ signaling during embryonic development, in particular for vasculogenesis and mesodermal patterning, has been the focus of attention of many studies. For instance, BMP4 is considered the most important member of the TGFβ superfamily for mesoderm formation  and subsequent commitment toward hematopoietic fate in gastrulating embryos . BMP4  and TGFβ1  have also been described as important regulators of YS vasculogenesis in E9.5 embryos. However, the role of members of TGFβ signaling in midgestational hematopoiesis has been far less explored, and therefore poorly understood.
Here, we demonstrated that endothelial cells (E+F+) and hematopoietic progenitors (E+F−) of the YS express BMP4 which binds to Alk3 and Bmpr2, phosphorylating the downstream effector Smad1/5, and ultimately activating the hematopoietic gene expression program within hematopoietic progenitor cells (E+F−) (Figs. 3, 4, 5, 6). These observations indicate that BMP4 not only is important for the specification of hematopoietic cells in early embryogenesis but also is necessary for proper hematopoiesis within the YS of midgestational embryos. TGFβ1 may also act on E+F− cells but through Alk5/Smad2, as these cells express high levels of Alk5 and PAI-1, a target gene for Smad2 (Figs. 3A–3D, 5A–5D, 5G). However, we did not observe differences in Smad2/3 activation among E+F+, E+F−, and E−F− cell fractions (data not shown). This agrees with previous in vitro data as Eng-deficient differentiating ESCs showed no changes in the levels of pSmad2, whereas levels of pSmad1/5 were significantly reduced . This has been confirmed here in vivo since E9.5 Eng−/− YSs display a 2.5-fold reduction in the levels of pSmad1/5/8 (Fig. 5E, 5F), which corroborate the premise that endoglin-mediated function in hematopoiesis occurs through BMP signaling, and not through TGF-β. The endothelial (E+F+) fraction at E9.5 (Fig. 2C, right panel) is also characterized by high levels of pSMAD1/5/8, most likely resulting from TGFβ1/Alk1 but also possibly from BMP4/Alk3 signaling (Fig. 5G).
Interestingly, lack of Eng in the YS leads to decreases of 3.1- and 1.6-fold in BMP2 and BMP4 expression, respectively. This effect was correlated with the downregulation of BMP2/4 target genes (Fig. 5A–5D), and decreased frequency of hematopoietic progenitors within E9.5 YSs . These data lead us to think that Eng signaling might regulate the expression of BMP2/4 and other members of the TGFβ signaling pathway in YS cells.
We also observed that hematopoietic and endothelial cells from the YS are temporally distinguishable by the levels of Eng. At E8.5, E+F+ cell fraction comprises approximately two thirds of the hematopoietic progenitors within the YS, in particular the most uncommitted precursor GEMM, while the E+F− fraction contains the remaining third (Fig. 3, left panel). As development progresses, hematopoietic activity transits to the E+F− population (Fig. 3, right panel). At E9.5, dim expression of Eng identifies hematopoietic progenitors within the YS, whereas bright endoglin expression marks endothelial cells (Fig. 2 and Supporting Information Fig. S3). EngdimFlk-1− cells were found in clusters adjacent to endothelial cells of the blood islands (BI) (Fig. 2C). The endothelial cells of the BI were EnghighFlk-1+ while the differentiated blood cells themselves were E−F−. Consistent with this, CD41 expression was enriched in the E+F− at E9.5 when compared with the E+F+ cell fraction (42.0% ± 14.6% vs. 12.6% ± 10.2%). These results are supported by a previous study showing that CD41+ cells present within the BI are enriched for hematopoietic progenitors . Interestingly, the localization of CD41+ cells within the YS found in that study resembles the position found here for Eng+ cells. It is important to note that at E9.5, E+F+ cells still contained hematopoietic activity but to a lesser extent. These results suggest that levels of Eng in the plasma membrane should be reduced as hematopoietic commitment occurs, while Flk-1 expression is completely abolished. Others have demonstrated that decreasing VEGF signaling through Flk-1 downregulation is necessary for the development of definitive hematopoiesis [39, 40]. However, this is the first time that such regulation has been described for the TGFβ signaling pathway during embryogenesis.
Altogether, our data demonstrate that endoglin plays a crucial role in regulating the activation of members of the TGFβ superfamily in YS of midgestational embryos. This regulation is also associated with a differential expression of TGFβ members within diverse cell types of the YS, which ultimately controls hematopoiesis.
This project was supported by NIH Grants R01 HL085840-01 and U01 HL100407. M.I. is currently affiliated with the Department of Pediatrics, Harbor-UCLA Medical Center, Torrance, CA.
Disclosure of Potential Conflicts of Interest
The authors declare no potential conflicts of interest.