|Home | About | Journals | Submit | Contact Us | Français|
During human development, signals that govern lineage specification versus expansion of cells committed to a cell fate are poorly understood. We demonstrate that activation of canonical Wnt signaling by Wnt-3a promotes proliferation of human embryonic stem cells (hESCs) -precursors already committed to the hematopoietic lineage. In contrast, non-canonical Wnt signals, activated by Wnt11, control exit from the pluripotent state and entry towards mesoderm specification. Unique to embryoid body (EB) formation of hESCs, Wnt11 induces development and arrangement of cells expressing Brachyury that co-express E-cadherin and Frizzled-7 (Fzd7). Knockdown of Fzd7 expression blocks Wnt-11 dependent specification. Our study reveals an unappreciated role for non-canonical Wnt signaling in hESC specification that involves development of unique mesoderm precursors via morphogenic organization within human EBs.
Lineage specification is a highly coordinated phenomenon delineated by temporal changes in gene expression at a single cell level that respond to changes at a multi-cellular level. Such coordinated events are orchestrated by key morphogenic signaling pathways (Kimelman, 2006) including the highly conserved Wnt family members (Kimelman, 2006; Logan and Nusse, 2004). Unlike invertebrate and other non-human models, this has been more difficult to understand in the human system where early developmental events cannot be experimentally manipulated. Human embryonic stem cells (hESCs) provide an invaluable approach to model fundamental processes of development, and provide a unique opportunity to define cellular mechanisms by which complex development events are modulated and organized by inductive signaling molecules such as Wnts.
Wnts are a family of 18 secreted glycoproteins that act as ligands for the seven-pass transmembrane Frizzled receptors (Fzd), and co-receptor LRP5/6 (LDL receptor related protein) (Wu and Nusse, 2002). Fzd signaling activatesβ–catenin-dependent (canonical) and –independent pathways (non-canonical) (van Noort et al., 2002). Wnt/β-catenin signaling, mediated by Wnt ligands such as Wnt3a, regulates the ubiquitin-proteasome destruction complex (Axin, APC, GSK-3, CK-1), resulting in the stabilization and translocation of β-catenin to the nucleus where it regulates gene expression (Moon, 2005; van Noort et al., 2002). β-catenin-independent Wnt signaling is mediated by Wnt ligands such as Wnt11, and acts through kinases such as c-Jun NH2-terminal kinase (JNK) and the calcium dependent kinases (CaMKII) and PKC (Kuhl et al., 2000; Nateri et al., 2005; Topol et al., 2003; Westfall et al., 2003). Although the Wnt signaling contributes to multiple developmental events during embryogenesis and in homeostasis of adult tissues, the roles of canonical and non-canonical Wnt pathways are poorly understood, and have yet to be studied in early human development.
Considerable species variation exists at very early stages of development, especially during gastrulation and morphogenesis. Accordingly, the functions of soluble factors such as Wnts can be cell- and species-dependent. For example, while the embryos of Wnt3 null mice fail to develop mesoderm, inhibition of the canonical signaling in zebrafish and xenopus results only in the axis truncation (Liu et al., 1999) (Humphrey et al., 2004; Poon et al., 2006; Tavian and Peault, 2005; Xu et al., 2002). This species variation also precludes precise extrapolations from mouse to humans in assigning specific roles for Wnts in development. Consistent with the importance of Wnt3a in mesoderm development, there is emerging evidence that temporal activation of Wnt/β-catenin signaling is crucial for cardiac and hematopoietic fate during murine and zebrafish embryogenesis (Naito et al., 2006; Ueno S., 2007). Similarly in the adult system, Wnt3a mediated signals that are important for achieving a balance of proliferation, differentiation, and self-renewal of the hematopoietic stem cell (HSC) originate from the mesoderm (Kirstetter et al., 2006; Reya et al., 2003; Scheller et al., 2006; Willert et al., 2003). In contrast, the functional roles of the β-catenin-independent Wnt pathways are less clear, but may involve regulating cell movements and frequent antagonization of the β-catenin pathway (Kuhl, 2002). β-catenin-independent Wnt signaling has been implicated in ventral cell fate choices, epithelialization in the early quail mesoderm, and cardiomyocyte differentiation in a number of species, such as quail, xenopus, zebrafish, and mouse ESCs (Eisenberg and Eisenberg, 1999; Eisenberg et al., 1997; Kuhl et al., 2000; Naito et al., 2006; Pandur et al., 2002; Ueno S., 2007). In humans, non-canonical Wnt pathways have been associated with adult stages of development that control the diversification of blood cell types and augment regenerative potential of human HSCs capable of repopulating immune-compromised NOD-SCID mice (Brandon et al., 2000; Murdoch et al., 2003), an observation also made in the mouse (Nemeth et al., 2007).
The exact role of canonical vs. non-canonical Wnts in human mesoderm and blood development remains to be defined. Capitalizing on the ability of hESCs to give rise to hemogenic precursors and primitive blood cells (Wang et al., 2004), we utilized hESCs as a robust in vitro model to examine the function of canonical and non-canonical Wnt activation during embryonic mesodermal and hematopoietic cell fate determination. We reveal a distinct temporal nature of canonical and non-canonical signaling to promote early human hematopoietic development, and propose that these two Wnt pathways mediate their effects via distinct cellular mechanisms to augment human blood cell fate that was not predicted by other invertebrate and non-human models of blood development.
Initial experiments were performed to provide evidence that Wnt ligands, Wnt3a and Wnt11, elicit signals along the expected canonical and non-canonical Wnt cascades within hESCs. The biological activity of the Wnts produced from L-cells was characterized on cells transfected with the TCF optimal promoter-GFP reporter construct (Suppl. Figures S1A–C). In addition, microarray analysis revealed the Wnt pathways are active in hESCs, and that these pathways are regulated within the hematopoietic-derived hESC compartment, which were cultured under feeder-free conditions to ensure transcript detection was not disrupted by feeder cells (Suppl. Figure S1D, (Wang et al., 2004). To evaluate the affects on the biochemistry of target proteins integral to canonical Wnt signaling, the phosphorylation status of β-catenin was assessed under control vs. Wnt ligand treatment in hESCs. Higher total levels of β-catenin were observed under Wnt3a treatment, while reduced levels were seen with Wnt11 stimulations compared to control condition (Suppl. Figure S1E). In addition, phosphorylated β-catenin (Ser-33/37, Thr-41) levels were lower with Wnt3a stimulation, but higher under Wnt11 conditions compared to control (Suppl. Figure S1E). Furthermore, accumulation of the non-phosphorylated form of β-catenin was observed with Wnt3a stimulation, and was reduced with Wnt11 co-treatment (Suppl. Figure S1F). These data are consistent with the notion that non-canonical Wnt signaling inhibits its canonical counterpart in hESCs, and that Wnt3a and Wnt11 can modulate β-catenin-dependent and -independent Wnt signaling respectively.
Since Wnt3a/canonical signaling has been implicated in many aspects of hematopoietic differentiation (Kirstetter et al., 2006; Reya et al., 2000; Scheller et al., 2006; Trowbridge et al., 2006), we hypothesized that the effect of Wnt3a may be temporal in nature, similar to its role during development in non-human species (Ang et al., 2004; Na et al., 2006; Nohno et al., 1999). To examine this, we characterized the effect of canonical Wnt3a signaling in the hESC system where hierarchical stages of blood development have previously been characterized (Figure 1A) (Chadwick et al., 2003; Wang et al., 2004). This developmental scheme of hematopoiesis from hESCS can be divided into two stages: Stage I and II. Stage I (Days 0–7) encompasses hemogenic lineage specification phase, characterized by the initial appearance of the bipotential hemogenic cells to hemogenic and endothelial precursors (CD45−CD31+ cells), and by the absence of hematopoietic (CD45+) cells and the lack of progenitor capacity. Stage I is followed by the commitment phase, Stage II (Days 7–15), characterized as the period where committed hematopoietic progenitors are detected by Day 10 and peak at Day 15.
Treatment with Wnt3a during Stage I or Stage II of EB development revealed that Wnt3a increased both the hemogenic and hematopoietic compartments only when present during the later commitment stage II (7–15 days) of blood development from hESCs. Restricted to Stage II, Wnt3a induced a 2.4-fold increase in hemogenic cell frequency and a 9.2-fold increase in total hemogenic precursors (Figures 1B and C). Similarly, Wnt3a induced a 2.8 and 3.8-fold increase in frequency of hematopoietic cells and total hematopoietic progenitors respectively, but only when present at Stage II of the hEB differentiation (Figures 1E and F), and not Stage I (treated for 1, 3 or 7 days) (Suppl. Figure S2A). In addition, Wnt3a had no effect at Stage I of hematopoietic development under serum-free conditions, indicating that serum does not mask canonical Wnt3a effectiveness (Suppl. Figures S2B and C). Treatment with the canonical Wnt inhibitor, Dkk1, reduced both the hemogenic and hematopoietic progenitors (CD34+CD45+) that were induced by Wnt3a during this phase of blood development (Figures 1D and G). Evaluation of the molecular activity of Wnt3a showed an increase in stable β-catenin levels that was also Dkk1 sensitive (Figure 1H). These data indicate that the effect of Wnt3a is regulated by canonical Wnt signaling that uniquely targets only committed cells contained within Day 7 to Day 15 hEBs (Stage II) to augment hematopoietic differentiation, and does not affect specification of blood fate from the pluripotent state that occurs at Stage I (Days 0–7) (Suppl Figure S2A).
To understand the cellular mechanism by which Wnt3a augments hematopoiesis during latter stages of hEB development when cells were already destined to the hematopoietic fate, we analyzed both cell death and proliferation of committed cells. Augmentation of hemogenic and hematopoietic potential from hESCs following Wnt3a treatment at Stage II could not be explained by affects on cell survival (data not shown). Using markers of proliferation, proliferation cell nuclear antigen [PCNA] and 5-bromo, 2-deoxyuridine (BrdU) incorporation, we observed a 2.3-fold increase in number of PCNA positive hEBs upon Wnt3a stimulation as compared to control (Figures 2A–D). Confocal images of single cells isolated from Day 15 BrdU treated hEBs show robust BrdU staining of the hemogenic [CD31+] and hematopoietic cells [CD45+] upon Wnt3a (Suppl. Figure S3A–C). Quantitative analysis of proliferative status upon Wnt3a stimulation revealed that the overall cell proliferation, enumerated as a proportion of BrdU labeled versus non-labeled cells, was increased in the Wnt3a Stage II treated hEBs (66.5 ± 6%) compared to Stage I (57.4 ± 0.1%) treated hEBs or untreated (21.4 ± 2.7%) hEBs (data not shown). Importantly, the BrdU incorporation (7ADD for DNA content) within the target hemogenic compartment (CD31+CD45−) was 4.9-fold higher during Stage II treatment (Figures 2E and E′). Similarly, we observed a 3.1-fold increase in frequencies of cycling hematopoietic cells (CD45+) (Figures 2F and F′), and a 5-fold increase in the levels of the canonical Wnt target gene Cyclin D1, associated with cellular proliferation (Figure 2G). These data indicate that the cellular mechanism by which canonical Wnt (Wnt3a) signaling promotes hematopoietic output from hESCs is mediated by cell cycle induction of previously committed hemogenic and hematopoietic cells.
Given the temporal proliferative effects of Wnt3a of hESC-derived blood development, we evaluated the effects of non-canonical Wnt signals using Wnt11, based on established responses to Wnt11 in hESCs (Suppl. Figure S1). In contrast to Wnt3a treatment, analysis of Day 15-hEB differentiation revealed that Wnt11 increased blood formation during the initial stage of hematopoietic development from hESCs (Stage I). The frequency of hematopoietic cells and hematopoietic progenitor numbers were upregulated by 2-fold and 2.5-fold respectively when Wnt11 was present during the specification phase (Stage I) (Figures 3A and B). Similar to Wnt11, stimulation with Wnt5a (another non-canonical Wnt pathway ligand), during Stage I of hEB differentiation, also increased hematopoiesis (Suppl. Figures S4A–D). Additionally, Wnt11 treatment during Stage I of differentiation increased primitive hematopoietic programs, as indicated by zeta and epsilon globin expression, which were 16-fold and 4.6–fold higher respectively, compared to control conditions, while definitive hematopoiesis (beta globin) was unaffected (Suppl. Figure S4G). Although Wnt11 was shown to function as an inhibitor of β-catenin (Figure 3D), co-treatment of Wnt11 with Dkk1 (200 ng/ml; as optimized dose of DKK1 Suppl. Figure S2D) did not affect the Wnt11 response seen at Stage I of treatment (Figure 3C). Dkk1 addition was able to reduce endogenous levels of β-catenin (Figure 3E), indirectly suggesting canonical Wnts are produced by hEB cells, but have no biological affect on hematopoietic development (Figure 3F, or at any other dose of Dkk-1 tested, Figure S2D). Together, these observations suggest that inhibition of endogenous canonical Wnt signaling cascade is not sufficient to affect hematopoietic differentiation, and that Wnt11 signaling functionally enhances and temporally regulates hematopoiesis during Stage I (Days 0–7) of hEB development.
To further evaluate the early effects of Wnt11, hEBs were treated with Wnt11 for 1 day and analyzed at Day 15. This 24hr Wnt11 exposure was sufficient to promote blood development of both hemogenic precursors (Figure 3G) and hematopoietic progenitors (Figures 3I). Total CFU production, as a functional measure of hematopoietic progenitor capacity, was determined for CD31+CD45− cells derived from Day 10 and Day 15 hEBs treated with Wnt11 (Figure 3H) and demonstrated that Wnt11 increases CFU output from this population. Interestingly, unlike the modest affects seen during continued treatment through Stage I and II on hemogenic precursor output (Suppl. Figures S4E and F), a single exposure of Wnt11 in the first 24hrs upon hEB formation was sufficient to increase the hemogenic precursor frequencies by 2.4-fold compared to control (Figure 3I). This 24hr effect was also demonstrated under serum free conditions indicating that serum does not play a role in Wnt11 effects (Suppl. Figures S2B and C). Similar to effects on hemogenic precursors, 24hrs of Wnt11 treatment resulted in a 2.2-fold increase in CFU production arising from CD34+CD45+ cells at Day 15 of hEB development (Figure 3J). These data suggest that 24 hr of Wnt11 treatment (Day 1 of Stage I) equally augments both the hemogenic and hematopoietic progenitor phenotype and function progenitor capacity during hEB hematopoietic development.
Based on the requirements of Wnt11 during the early phase (Stage I), but not the later phase (Stage II) of blood differentiation, and the immediacy of Wnt11 effects (24hrs), we hypothesized that Wnt11 may promote the progression of unknown early precursors of the blood lineage and may direct control genes associated with the ground state of pluripotency (Boyer et al., 2005; Chambers et al., 2003; Niwa et al., 2000; Zeineddine et al., 2006).
Given the immediate effects of Wnt11 in induction of hematopoiesis and sustained expression of both Nanog and Oct4 within the first 2 days of hEB formation (data not shown), we examined the potential association of Wnt11 with Oct4 and Nanog transcript regulation during the 24 hours of Wnt11 treatment. Changes in Oct4 and Nanog expression were monitored over time, from the undifferentiated hESC state to 24hrs post hEB formation and post treatment (Figure 4A). Oct4 transcript was increased upon hESC aggregation (Figure 4B), and rapid up and down regulation of Oct4 was consistently observed upon Wnt11 treatment compared to control treated cells in the absence of Wnt11 (Figure 4C). While levels of Oct4 transcript were lower under control conditions by 24hrs, Oct4 expression was maintained in response to Wnt11 (Figure 4C). Furthermore, the frequency of Oct4 positive cells measured by intracellular staining for Oct4 protein, as well as western blot analysis of Oct4 protein levels, demonstrated Oct4 was regulated in response to Wnt11 stimulation compared to control conditions (Suppl. Figures S5A and B). Given that the stability and turnover of the Oct4 protein is not well defined, the differences between the protein level and transcript expression are likely to differ (Wei et al., 2007), however Wnt11 equally effected Oct4 protein and transcript, albeit with different kinetics due to the nature of transcript vs. protein regulation. Similar to Oct4, Nanog transcript levels were found to be upregulated 24hrs post hEB formation (Figure 4D), and responded to Wnt11 treatment (Figure 4E). Our data suggests that Wnt11 regulates factors associated with pluripotent state, in addition to effecting lineage specification genes toward mesodermal development.
To test this hypothesis, we monitored changes in gene expression for factors associated with mesoderm specification, including T-box and homeobox factors within the developing hEBs. The expression of stem cell leukemia factor (SCL/Tal1), important during hematopoietic development, and the homeobox gene MixL1 and the T-box gene Brachyury, surrogate markers of primitive streak and blood-mesoderm development during early embryogenesis (Huber et al., 2004; Ng et al., 2005; Shivdasani et al., 1995), were analyzed over time with Wnt11 treatment. Interestingly, undifferentiated hESCs express a wide range of mesoderm (Eomesodermin, Brachyury, MixL1, SCL/TAL1) and endoderm genes (FoxA2, Gata5, Hnf3α). Prior to treatment with the Wnt11, ie. 24hrs post-hEB formation, endoderm genes are downregulated while mesoderm genes like Brachyury and Mixl1 are upregulated (Suppl. Figure S5C). Upon 24hrs of Wnt11 treatment, SCL/Tal1 and Brachyury expression were positively regulated by 2-fold and 2.3-fold respectively (Figures 4F and G). While Brachyury expression levels peaked in Day 1 hEBs, MixL1 remained largely unchanged upon Wnt11 stimulation (Figure 4H). Furthermore, Wnt11 induced Brachyury and SCL/Tal1 expression also suggested a previously associated role of Wnt signaling in mesendodermal transition (Hart et al., 2002; Maduro et al., 2005; McLean et al., 2007). To assess the specificity of Wnt11 stimulation, we analyzed the expression of surrogate genes associated with mesoderm and endoderm hours after Wnt11 treatment. Wnt11 stimulation preferentially promotes the early differentiation of mesoderm from hESCs (Figure 4I). These results indicate that Wnt11 represents a key signal that dually orchestrates genes involved in sustaining the pluripotent state, together with those involved in mesendodermal lineage induction, thereby revealing a novel role of non-canonical Wnt signals in human cell fate decisions.
To better understand the receptor and cellular targets for Wnt11 in hESCs, we evaluated expression of candidate Fzd receptors associated with mesoderm lineage development previously linked to Wnt ligand activity in non-human models (Toyofuku et al., 2000; Witzel et al., 2006, Medina, 2000 #170, Djiane, 2000 #165). These included the Fzd2 and 7 receptors. In comparison to Fzd2, a 2-fold upregulation in Fzd7 expression was observed with 24hrs of Wnt11 stimulation compared to control treated hEBs (Figure 4J). This was consistent with observations in the developing mouse embryo, where unique changes in the Fzd7 expression were induced upon Wnt11 exposure and were reflective of morphogenesis (Djiane et al., 2000; Winklbauer et al., 2001). To further examine the potential relationship between Wnt11 induced mesodermal specification and Fzd7 expression, we stained Wnt11 treated hEBs (Figures 5B–D) with the early mesodermal marker (Brachyury) and Fzd7, and compared them to control hEBs (Figures 5A and C). We could only observe Fzd7 and Brachyury co-stained cells in the Wnt11 treated hEBs (Figures 5B–D′). To detail this Wnt11 response, we examined cells for co-expression Fzd7 and Brachyury in control vs. Wnt11 treated hEBs via immunofluorescence staining and flow cytometry analysis. Although we observed similar proportions of Brachyury+ hEBs and Fzd7+ hEBs in both control and Wnt11 treatment, Wnt11 stimulation caused 8.3-fold or 7.9-fold increases in percentage of cells expressing Brachyury or Fzd7 proteins respectively (Figure 5E). Importantly, the identification of cells co-expressing Fzd7 and Brachyury was seen exclusively in the presence of Wnt11 treatment (Figure 5F, and Suppl. Figure S6). Based on these observations, we hypothesized that Wnt11 may act through a primitive Fzd7 receptive target population for mesoderm specification.
To understand the role of Fzd7, we used a loss of function approach by silencing Fzd7 during blood differentiation induced by Wnt11. The effect of the Fzd7 siRNAs was demonstrated via changes in the Fzd7+ population frequency (ie. cell numbers). Treatment of Wnt11-hEBs with siRNA against Fzd7 effectively decreased the percentage of Fzd7+ cells (which are only detectible under Wnt11 treatment) within the developing hEBs 48hrs post-transfection, and reduced the frequencies of Day 15 hematopoietic cells from the Wnt11 stimulated hEBs (Figures 5G and H). Silencing of other frizzled receptors, such as Fzd 2 that has been associated with the canonical Wnt response, did not affect hematopoietic differentiation induced by Wnt11, further supportive of the specific importance of non-canonical Wnt signaling during early hEB differentiation (Suppl. Figure S7). Reductions in blood development caused by these Fzd7 siRNAs were dependent on the presence of Wnt11 (Figure 5H). Interestingly, silencing of Fzd7 within the Wnt11 stimulated hEBs also resulted in significant downregulation of the pluripotent factors Oct-4 and Nanog, mesoderm gene Brachyury and MixL1, and slight downregulation in endodermal genes Hnf3α and Gata5, but had little affect on SCL/Tal1 expression levels (Figures 5I and J). These data reveal that Wnt11 stimulation induces a unique Fzd7+ population not present in the absence of Wnt11 stimulation.
The non-canonical Wnt signaling is known to play a pivotal role in cellular movements in the embryo proper that may be mediated by architectural scaffolds receptive to signaling cascades that promote lineage differentiation via critical cellular interactions (Dang et al., 2002; Skerjanc et al., 1994). To test if structure alters the receptivity of Wnt11 stimulation to augment hematopoietic development, we compared treatment of hESCs assembled into hEBs vs. hESCs in monolayers. EBs and monolayers of hESCs were treated under identical Wnt11 or control conditions, and then assayed for blood development. Hematopoietic differentiation induced by Wnt11 was only observed upon EB formation from hESCs and not monolayers cultured and treated under identical conditions (Figure 6A). Molecular analysis after 1 day of hEB formation revealed an intense upregulation of phosphorylated CaMKII activity, which was exclusive to the 30-minute Wnt11 treated hEBs and was not observed in hESCs treated in monolayer cultures (Figures 6B and C). Undifferentiated hESCs also expressed basal CaMKII activity, however 30-minute stimulation with Wnt11 only increased CaMKII activity in hESCs in EBs and not in hESCs assembled monolayers. The rapid response of CaMKII in hEBs suggests a direct activation of this signaling cascade by Wnt11. These data support a β-catenin independent mechanism for Wnt11 action via CaMKII that is interconnected and dependent on cellular interactions supported in complex EB formation and architecture not available in hESC treated under identical conditions assembled in monolayers.
Early events of mesoderm development in mammalian embryos have been associated with changes in dynamic movement and adhesion properties of nascent mesodermal progenitors (Burdsal et al., 1993; Solnica-Krezel, 2006). In addition, Wnt11 signaling has been shown to mediate E-cadherin dependent morphogenesis (Burdsal et al., 1993; Toyofuku et al., 2000; Ulrich et al., 2005). To determine if similar mechanisms were responsible for Wnt11 responsiveness in hEBs undergoing mesodermal specification, we examined the expression patterns of E-cadherin in hEBs stimulated with Wnt11 for 24hrs compared to control. Under control conditions, E-cadherin+ cells were present on the periphery of hEBs (Figure 6D, open arrow), whereas Wnt11 treatment induced clustering of E-cadherin expressing cells (Figure 6E, closed arrow). Although the percentage of E-cadherin+ hEBs between the control and Wnt11 treatments were similar (data not shown), the Wnt11 treated hEBs showed a higher frequency of these clusters as compared to controls (Figure 6J). The immediate effects of Wnt11 on cellular organization within hEBs is consistent to that recently observed in murine EB development in response to Wnts (ten Berge et al., 2008). Serial sections stained for Fzd7 and E-cadherin demonstrated co-localization exclusively in the Wnt11 treated hEBs (Figures 6H–I″) versus the control (Figures 6F–G″). Interestingly, the Wnt11 treated hEBs uniquely accumulate Fzd7+, E-cadherin+, and Brachyury+ cells (Figures 6K–M″). This unique population of Frd7+E-cadherin+Brachyury+ was never detectable in the absence of Wnt11. Taken together, these results describe unique responses to non-canonical Wnt11 induced signals in hEBs that lead to the emergence of unique mesodermal population not present in the absence of Wnt11.
Understanding the cellular and molecular processes during lineage specification in vivo forms an important basis to attempt to control in vitro differentiation of hESCs. Reciprocally, hESC differentiation has been suggested as a model to map complex cellular interactions and movement that cannot be accessed in the human embryo. However, these applications have yet to be fully demonstrated in the hESC system. Using mesodermal and subsequent hematopoietic development from hESCs as a basis, our current study identifies an unpredicted role for non-canonical Wnt signaling to induce exit and hematopoietic specification of the hESCs, whereas canonical Wnt signaling was revealed to be limited to proliferation of hemogenic precursors already committed to the blood lineage. Based on our results, we propose a model by which Wnts temporally regulate hematopoietic development from hESCs via unique mechanisms known to be important in the developing embryo (Figure 7).
In vivo generation of initial blood forming stem cells involves several developmental stages coupled to anatomical movement in the human (Tavian and Peault, 2005). Accordingly, it is likely that signaling factors governing this process are diverse, and that non-canonical and canonical Wnts would affect different stages and target populations contributing to mammalian embryonic hematopoiesis similar to their effects in other lineages (Logan and Nusse, 2004). In vitro, specification events and expansion of the hematopoietic compartment can only be modeled using pluripotent cells required to make lineage choice. This cellular process has been broadly envisioned as a loss of pluripotent state and subsequent emergence of hemogenic precursors responsible for final hematopoietic differentiation and blood cell output (Ogawa et al., 2001). Although this model has been recapitulated in several species, the earlier events from pluripotent stage to mesodermal transitions and the growth factors required have not been extensively studied. Taking the current model of hESC blood development into account (Wang et al., 2004), we reveal a unique role for non-canonical Wnt signaling mediated through Wnt11 that promotes exit from the pluripotent state to mesodermal specification. The effect of Wnt11 was dependent on hESC assembly into EB structures since no affect on hematopoietic output was induced using monolayers of hESCs (Figure 7). Wnt11 induces blood cell fate through combined CaMKII and β-catenin regulation and morphogenic organization within hEBs, and causes changes in expression of pluripotent factors Oct and Nanog as mesodermal genes are rapidly induced (Figure 7). These combined processes cause the development of a unique subset of Fzd7 expressing cells that co-express Brachyury and E-cadherin, both associated with mesodermal development and mesodermal cell movement in early mammalian development (Huber et al., 2004; Kwan and Kirschner, 2003; Winklbauer et al., 2001; Yamanaka and Nishida, 2007).
This is the first report revealing the importance of non-canonical Wnt signaling during human embryonic hematopoiesis, however Wnt11 and similar cellular signals have been associated with hematopoietic output in other species. The presence of Brachyury+, Fzd7+, E-cadherin+ mesoderm progenitors as clusters within the Wnt11 treated hEBs, together with changes in mesendoderm to mesoderm gene expression, illustrate the role of the non-canonical Wnt signaling during embryonic mesoderm induction from hESCs. Association of Wnt11 to blood development has been reported in early avian and amphibian development, where Wnt11 influences the multilineage potential of blood cells from avian mesoderm stem cells (Eisenberg and Eisenberg, 1999; Eisenberg et al., 1997). Furthermore, a recent report by Kim et al showed preferential expression of Wnt11 and Fzd7 within the Flk+ (hemogenic cell marker) population derived from mESCs (Kim et al., 2008). At the signaling level, the robust upregulation of phosphorylated CaMKII observed immediately upon Wnt11 exposure of hEBs correlates with the previous observation that CaMKII activation induced by non-canonical Wnt pathway in the xenopus embryos is able to promote hematopoietic-associated ventral cell fates (Kuhl et al., 2000; Moon et al., 1993). Importantly, the exclusive increase of phosphorylated levels of CaMKII within the hEBs and not in treated monolayers of hESCs underscores the importance of 3D structure (Duprat, 1996; Gurdon et al., 1993; Kato and Gurdon, 1993), and cellular movement required to mediate mesodermal inductive Wnt11 effects. E-cadherin has been specifically implicated in the polarized segregation of mesodermal progenitor cells undergoing EMT at the primitive streak (Burdsal et al., 1993; Ciruna and Rossant, 2001). Taken together, our observations using the hESC system complement observed effects of Wnt11 based CaMKII signaling in other species, where the earliest events of specification towards mesoderm can be moderated.
The role of Wnt3a to enhance blood production has been observed in several systems such as zebrafish specification to hemogenic mesoderm (Martin and Kimelman, 2008; Shimizu et al., 2005), early mouse embryo development (Lindsley et al., 2006; Liu et al., 1999), mESC-hematopoiesis (Lengerke et al., 2008; Nostro et al., 2008), and hESC-derived hematopoiesis (Woll et al., 2007). However the temporal nature of the signaling and cellular mechanism of Wnt3a actions was not delineated nor the potential interactions and effects of non-canonical pathways and associated ligands such as Wnt11. Similar to these previous studies, addition of Wnt3a augmented hematopoiesis derived from hESCs here, and further revealed the action of Wnt3a was mediated via induced proliferation of cells already committed to the blood lineage and not early specification events (Figure 7). The biological effect of canonical Wnt observed here in hESCs is consistent with multiple studies showing the proliferative effects of canonical Wnt activation on mouse and human somatic blood stem cells (Kirstetter et al., 2006; Reya et al., 2003; Scheller et al., 2006; Trowbridge et al., 2006; Willert et al., 2003). Accordingly, despite the previously recognized role of canonical Wnt signals in mesodermal development, the interplay of non-canonical Wnt signaling should be examined in other development systems based on our current observations in hESCs. Moreover, interplay of canonical and non-canonical Wnt pathways in the context of human mesodermal and endodermal development cannot be ruled out, and it can set the platform for further studies, utilizing loss-of function and gain-of-function genetic techniques, that could clarify these complex interactions.
Our study reveals that non-canonical vs. canonical Wnt signaling is capable of guiding differentiation of embryonic stem cells towards mesoderm and subsequent hematopoietic fate respectively, thereby establishing that members of the Wnt family are capable of controlling unique transitions of blood development in the human. These observations provide unique insights into early developmental events in the human, given the inability to access these processes in the human embryo. Since Wnt11 induces emergence of a unique population of Fzd7+/Brachury+/E-cadherin+ cells, it will be important to evaluate the lineage potential of these cells that may possess broader potent developmental capacity to other mesodermal lineages such as muscle, cardiac, and bone derivatives. These capacities are currently being explored in our lab, along with defining culture methods to sustain this subpopulation in vitro. The ability to sustain these cells in culture will further allow examination of the role of Wnts in combination with others associated with ventral mesodermal fate such as BMP-4 towards a better understanding organization of signals required for hESC hematopoietic differentiation. Collectively, these findings will need to be applied to clinical goals of generating sufficient numbers of appropriately programmed hematopoietic cells from hESCs that possess HSC properties of in vivo reconstituting function similar to HSCs obtained from human bone marrow or cord blood.
Control and Wnt expressing L-cells were grown in media used to culture the hEBs. The hEB media consisted of 80% knock-out Dulbecco modified Eagle medium (KO-DMEM; Gibco, Burlington, ON, Canada), 20% non-heat inactivated FBS (Hyclone), 1% nonessential Amino (Gibco) Acids, 1mM L-Glutamine (Gibco) and 0.1mM β-mercaptoethanol (Sigma, Oakville, ON, Canada). The CM containing active Wnt3a, Wnt11, and CM from control L-cells were collected every three days and two collections were carried out before use. Cell debris was removed from the CM by centrifugation at 250g for 10mins.
Human ESC lines H1 and H9 were cultured in MEF-CM supplemented with 8ng/ml of bFGF (Invitrogen) as a feeder-free culture on Matrigel (BD Biosciences, Mississauga, ON, Canada). When hESC cultures have reached 80–90% confluence and hESC colonies are dense, the cultures are disassociated using 200U/ml Collagenase IV (Gibco) and passaged as 1:2 ratios onto fresh Matrigel. The media was changed every day. All experiments were carried out on both the H1 and H9 cell lines. Human embryoid bodies (hEBs) were formed as previously described (Chadwick et al., 2003). The hEBs were cultured under hematopoietic conducive conditions (300ng/ml SCF, 300ng/ml Flt-3L, 10ng/ml IL3 and IL-6, 50ng/ml G-CSF, 25ng/ml BMP-4), and supplemented with 4% vol/vol of either control Wnt3a or Wnt11 CM the day after hEB formation. Conditioned media was supplemented either during the entire hEB differentiation or during 0–7 days [stage I] or 7–15 days [stage II] of the developing hEB. Time zero thus represents the day at which the cytokines, growth factor, or CM are added. Dkk1 treatment (200ng/ml) was done during Stage I or Stage II or both stages of hEB differentiation.
Hemogenic (recognized phenotypicially as CD45−CD31+) and hematopoietic cells (CD45+) were identified by staining single cells (2–5 × 105 cells/ml) isolated from hEBs disassociated with 0.4U/ml Collagenase B (Roche Diagnostics, Laval, QC, Canada) from day 15 with fluorochrome-conjugated monoclonal antibodies (mAb) CD31-PE and pan-leukocyte marker CD45–APC (Milteny Biotech, Germany). The mAb and their corresponding isotypes were used at 1–2mg/ml. Frequencies of cells possessing the hemogenic and hematopoietic phenotypes were determined on live cells by 7AAD (Immunotech) exclusion, using FACSCalibur, and analysis was performed using the FlowJo software (Tree Star). Hemogenic and hematopoietic cellularity was determined by multiplying total cellular yield by their respective frequencies. For data analysis, effect on frequencies and total hemogenic and hematopoietic output by the Wnt-CMs were normalized to the control-CM within each experiment. Data presented are mean ± s.e.m of pooled normalized values between experiments. Statistical significance was determined using the ANOVA-Tukey HSD test and is reported as p < 0.05. For E-cadherin analysis, 2mg/ml of monoclonal anti-human APC conjugated E-cadherin and APC-labeled mouse IgG2a isotype were used (R&D Systems). For Frizzled 7 and Brachyury analysis, 0.5μg/μl anti-human Frizzled 7 (R&D Systems) conjugated with FITC and anti-human PE conjugated Brachyury (R&D Systems) were used.
Clonogenic blood progenitor assays were carried out by plating 10–15 × 103 single cells from day 15 hEBs onto methylcellulose H4230 (StemCell Technologies, Vancouver, BC, Canada) supplemented with human recombinant growth factors described previously (Wang et al., 2004). Differential colonies in the methylcellulose culture were scored based on standard morphological criteria, after 14 days incubation at 37°C in a humidified chamber.
hEBs were pulsed with 10mM BrdU (Becton Dickinson), 24hrs before hEB disassociation. 1×106 single cells were stained for membrane CD proteins with fluorochrome-conjugated CD31-PE and CD45–FITC (Milteny Biotech, Germany) prior to processing for intracellular staining with BrdU-APC mAb (Becton Dickinson). The cells were fixed and permeabilized according to the manufacturer protocol (Becton Dickinson). Cellular proliferation was quantified by Flowcytometry and qualified by Spinning Disk Confocal microscopy. For image analysis the fluorochrome immunostained single cells were cytospun and excited with a 488 (FITC detection), 543 (PE) or 647 nm (APC) laser path on the Lieca DMI 6000 B microscope (Bernsheim, Germany). Images were acquired with the Volocity 4 (Improvision, Coventry, UK) and analyzed with the ImageJ v1.37 (http://rsb.info.nih.gov/ij/) and ImagePro plus v 6.1 software
For siRNA transfections, clumps of hESC colonies were obtained as previously mention and transferred to 6-well ultra low attachment plates (Corning). Cell were transfected in EB differentiation medium using lipofectamine (Ambion, Inc., USA) and 50–100nM siRNA according to manufacturer’s instructions. After 24 h, medium was changed to EB differentiation medium supplemented with hematopoietic growth factors. All siRNAs were purchased from Ambion, Inc and their effect were tested individually and in combination. Following siRNA were used: Fzd2 siRNA ID: 4057, 45981 and 3962; Fzd 7 siRNA ID: 4955 and 4861; Scrambled siRNA #1. siRNA transfection efficiency was assessed using a Silencer®Cy™ 3-labeled negative Control # 1 siRNA (Ambion, Inc., USA).
1–3 × 106 hESCs or day 1 hEB cells or 2–4 × 105 cells day 15 hEBs were lysed with buffer containing 1% Triton X-100, 150mM NaCl, 10mM Tris-HCl, 5mM EDTA, 10mg/ml protease inhibitors (Leupeptin, Aprotonin, Apeptin), 0.5mM PMSF and phophatase inhibitor cocktail (SetIV, CalBiochem). For Western blots, equal amount of proteins (20ug) were separated using 12–15% SDS-PAGE and transferred to PVDF membrane. Membranes were blocked with 5% skimmed milk for 2hrs and blotted with the primary antibody overnight at 4°C. The following primary antibody used: rabbit phospho-β-catenin (Ser33/37, Thr41) (1:1000, Cell Signaling) mouse monoclonal dephopho β-catenin (1:500, Alexis Biochemicals), mouse monoclonal total β-catenin (1:1000, Becton Dickinson), rabbit polyclonal phospho-CaMKII (1:1000, Cell Signaling) and mouse monoclonal phospho-SAPK/JNK (1:2000, Cell Signaling), mouse monoclonal Oct4 (1:1000, BD Transduction Lab). The membranes were washed and stained with HRP-conjugated goat anti-rabbit or mouse Ab (1:10 000, Santa Cruz) and signals were detected with the enhanced chemiluminescence method (Pierce), membranes were exposed to X-ray film and UVP Bioimaging system (UVP, California) and band intensities were subsequently quantified using the ImagePro software. Immunoprecipitation (IP) was performed for detection of CaMKII and JNK activity. 300–500mg of total hESCs or hEB cell lysates were allowed to interact with rabbit polyclonal total CaMKII (1:250, Cell Signaling) or total JNK (1:250, Cell Signaling) immobilized antibodies (Aminolink Plus Coupling Gel, Pierce Biotech.) according to the manufacturer protocol. Western blots were performed on the total CaMKII and JNK IP proteins and phosphorylated versions of the respective proteins were identified with the phospho-CaMKII and phospho-JNK Ab. Immunoblots were stripped and re-blotted for either mouse polyclonal β-actin (1:5000, Sigma); mouse GAPDH (1:10,000, Cell signaling) or total JNK (1:1000, Cell signaling) as loading control.
Wnt and control treated hEBs were washed 2–3 times in 1X PBS/3% FCS, fixed with 4% paraformaldehyde/PBS for 2hrs, embedded, then snap-frozen in liquid nitrogen and stored at −80°C, for staining with Fzd7, E-cadherin, Brachyury, and PCNA. 8mM cryostat serial sections were made for each treated specimen and successive serial sections were single-stained with Fzd7, E-cadherin, and Brachyury proteins for detection within the same hEB region. For each single staining, 3 sections at an interval of 7–10 (25–30 μm) serial sections were used. The sections were hydrated with 1XPBS and permeabilized with 0.3% Saponin/PBS for staining Fzd7 and E-cadherin, and 0.1% Triton X-100/PBS for Brachyury and PCNA. All Ab stained sections were washed in their respective permeabilization buffer. Sections were blocked with either 10% Normal Goat Serum (Fzd7 and E-cadherin), 10% Normal Donkey Serum (for Brachyury) or 10% Normal Rabbit Serum (PCNA) + 1% BSA at room temperature (RT) for 2hrs. The following primary antibody and dilutions were used: rabbit polyclonal Fzd7 (13μg/ml, Abcam), goat affinity purified Brachyury (10μg/ml, R&D systems), mouse monoclonal E-cadherin (10μg/ml, Alexis) and mouse monoclonal PCNA-FITC (1:250, AbCAM). Sections were incubated with primary antibodies overnight and subsequently secondary stained for 1hr with Alexa Fluor 488 goat anti-rabbit IgG (Molecular Probes) for Fzd7, Alexa Fluor 594 goat anti-mouse IgG (Molecular Probes) for E-cadherin, or Alexa Fluor 594 donkey anti-goat IgG (Molecular Probes) for Brachyury at 2μg/ml. Slides were mounted and counterstained using VECTASHIELD HardSet Mounting Medium with DAPI (Vector Labs). Sections were examined using the Olympus IX18 microscope and images were captured with a Photometrix Cool Snap HQ2 camera using In Vivo version 3.1.2 (Photometrix) software. Images were pseudo-colored and analyzed using Image-Pro software.
Total RNA was isolated from hEBs using the Qiagen Allprep RNA Mini Kit (Qiagen). Complementary DNA (cDNA) was made with 1–5μg of total RNA using the first-strand cDNA synthesis kit (Amersham Biosciences) and subsequent quantitative Real-time PCR (Q-rtPCR) were carried out in duplicate, using Platinum SYBR Green qPCR Super Mix-UDG on an Mx3000P® Q-PCR System according to manufacturer instructions (Invitrogen). Amplifications were performed using the following conditions: 95° C, 10 min and 40 cycles 95 °C, 30 s; 60° C, 1min; 72° C, 30s. All data were normalized to GAPDH and relative gene quantification for Wnt effects was calibrated against the control-CM effects: Wnt and control treatment Ct-Gapdh Ct = ΔCt then Wnt Ct − control ΔCt = ΔΔCt, and relative expression is expressed as 2(−ΔΔCt). Q-rtPCR primers are listed in Suppl. Table S1
M.B. by the Canadian Chair Program who holds the Canada Research Chair in human stem cell biology. This work was supported by a grant from CIHR and NCIC to M.B. We also are grateful to Aimee Kohn, R. Mondeh and Jiabi Yang for their technical assistance, and to Morag Stewart, Tamra Werbowetski-Ogilvie, Jennifer Trowbridge and Chantal Cerdan for useful suggestions.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.