VEGF-mediated PI3K class IA and PKC signaling in cardiomyogenesis and vasculogenesis of mouse embryonic stem cells

The dissection of signaling pathways regulating cardiac versus vascular differentiation is of particular impact because recent evidence has shown that, during embryogenesis, cardiomyogenesis and vasculogenesis occur in a strictly time- and space-dependent manner. Using mouse embryonic stem (ES) cells, which mimic early embryonic development, it was discovered that cardiac-independent vasculogenesis arises from Flk-1⁺ (vascular endothelial growth factor receptor 2-positive, VEGFR2⁺) hemangioblasts, whereas at cardiogenic day 3.25 a ‘second’ Flk-1⁺ cell population with enriched cardiogenic as well as vasculogenic potential was identified (Yang et al., 2008). This observation raised the intriguing possibility that the cardiac lineage develops from a progenitor cell that also displays vascular potential and thus might represent the cardiac equivalent to the hemangioblast (Bu et al., 2009; Garry and Olson, 2006; Kattman et al., 2007). As a consequence, Flk-1⁺ cardiac as well as endothelial progenitor cells should be responsive towards VEGF stimulation, and might activate distinct signaling pathways that variegate cardiac-associated vasculogenesis versus cardiac-independent (yolk sac hemangioblast-derived) vascular differentiation (Iida et al., 2005; Kattman et al., 2006).

The signal transduction pathways that are activated in different cell types in response to VEGF are well established (Cross et al., 2003; Wang et al., 2008). However, relatively little is known about the involvement of distinct phosphoinositide 3-kinase (PI3K) catalytic subunits and protein kinase C (PKC) isoforms in VEGF action (Hamada et al., 2005; Gliki et al., 2002; Gerber et al., 1998). VEGF receptors (VEGFR1 and VEGFR2) are present on vascular endothelial cells and their signaling is mediated by receptor dimerization leading to autophosphorylation of the cytosolic domains of the receptors. Phosphorylated VEGF receptors serve as docking sites for adapter molecules or signaling enzymes such as PI3K. VEGF has been shown to activate PI3K generating phosphatidylinositol (3,4,5)-trisphosphates (Bos 1995; Gerber et al., 1998). Moreover, numerous studies demonstrated that PI3K plays an important role in regulating endothelial proliferation, migration and survival (Gerber et al., 1998; Thakker et al., 1999; Jiang et al., 2000).

PI3Ks are classified into classes I, II and III (Vanhaesebroeck et al., 1997). Class I PI3Ks, which include class IA and class IB (consisting of PI3Kγ only), are a family of dual-specificity lipid and protein kinases that control many cellular functions, such as growth and proliferation, survival and apoptosis, as well as adhesion and migration of a wide range of cell types (Katso et al., 2001; Wymann and Pirola, 1998). All four class I PI3Ks are heterodimers composed of a catalytic subunit with a molecular weight of 110 kDa and a tightly associated regulatory subunit that controls activation and subcellular localization (Whitman et al., 1988; Stephens et al., 1991). The importance of PKC regarding the VEGF–PI3K pathway was suggested by several authors (Chou et al., 1998; Dutil et al., 1998; Gliki et al., 2002; Le Good et al., 1998), although a detailed VEGF-related mechanism involving PKC has not been determined until now in differentiating ES cells. PKC is a family of protein kinases and phospholipid-dependent enzymes that consists of at least of 11 isozymes (Teicher, 2006). They are classified as conventional (α, β1, β2, γ), novel (δ, ε, η, θ, μ) and atypical (ζ, λ) isozymes.

The aim of this study was to identify the role of PI3K class IA (consisting of PI3Kα, PI3Kβ and PI3Kδ isoforms containing p110α, p110β and p110δ catalytic subunits, respectively) and their downstream effector pathways, i.e. PKC (PKCα/βII, PKCδ and PKCζ), in signaling pathways that diversify cardiomyogenesis versus vasculogenesis, and elucidate the impact of VEGF in these processes. Our data demonstrate that PI3K is a key regulator of cardiovascular differentiation of ES cells, and outline a specific role of the PI3K–Akt pathway in cardiomyogenesis and vasculogenesis versus a PI3K–PDK1–PKC pathway regulating vasculogenesis independently of cardiomyogenesis.

In the present study, the necessity of VEGF-regulated PI3K- and PKC-signaling pathways for the process of cardiomyogenesis as well as vasculogenesis was investigated in embryoid bodies (EBs) derived from CGR8 ES cells. These pathways and subsequent catalytic subunits are expressed in parallel to the first steps of cardiovascular differentiation (supplementary material Fig. S1A,B; n=3). To examine the effect of PI3K inhibition on cardiovascular differentiation, EBs were incubated with the general PI3K inhibitors wortmannin (1 μM) and LY294002 (50 μM) in the presence or absence of VEGF from day 4 to day 10 of cell culture. In parallel, specific inhibitors of distinct class 1A PI3K catalytic subunits, i.e. the p110α inhibitor compound 15e (0.5 μM), the p110β inhibitor TGX-221 (1 μM) and the p110δ inhibitor IC-87114 (1 μM), were applied in the presence or absence of VEGF (Fig. 1A–E; n=5). None of the pharmacological inhibitors significantly activated cleaved caspase 3 as a signature of apoptosis (supplementary material Fig. S2A,B; n=3). However, protein expression of the proliferation marker Ki-67 was significantly decreased in EBs treated with wortmannin, LY297002, compound 15e and IC-87114, but not with TGX-221 (supplementary material Fig. S2C; n=3), which resulted in decreased EB growth. Treatment of EBs with VEGF significantly increased CD31- and α-actinin-positive cell areas (Fig. 1A–C). Incubation of EBs with wortmannin and LY294002 significantly reduced the size of cardiac areas (Fig. 1A,C) and CD31-positive vascular areas (Fig. 1A,B) as well as the protein expression of CD31 and VE-Cadherin in the treated EBs (Fig. 1E). Strongest effects were achieved upon treatment with compound 15e (0.5 μM), which is a specific PI3K p110α inhibitor. Moreover, the percentage of beating areas per EB (Fig. 1D) and the protein expression of typical cardiac genes (α-actinin, cardiac MHC, cardiac MLC) (Fig. 1E) were significantly decreased upon incubation with wortmannin, LY294002 and compound 15e. Treatment with IC-87114 (1 μM), an inhibitor of the PI3K p110δ subunit, only reduced the size of cardiac and vascular areas (see Fig. 1A–C) as well as protein expression (see Fig. 1E), whereas the percentage of beating EBs remained unaffected (see Fig. 1D). Finally, treatment with the p110β inhibitor TGX-221 (1 μM) did not exert significant effects on cardiovascular cell area, typical cardiovascular proteins and the percentage of spontaneously contracting EBs (see Fig. 1A–E). A potential lineage shift towards the neuronal lineage caused by an inhibition of PI3K isoforms was excluded because mRNA expression of Nestin and neuron-specific enolase (NSE) was not changed upon treatment with PI3K inhibitors (data not shown). It was apparent that all inhibitors of PI3K downregulated the expression of cardiovascular genes, with most pronounced effects observed upon inhibition with the PI3K p110α inhibitor compound 15e and weakest effects with the p110β inhibitor TGX-221. None of the described effects was abolished through parallel application of VEGF.

To verify the results obtained from the pharmacological inhibition we used stable short hairpin RNA (shRNA) transduction targeting PI3K p110 catalytic subunits. A specific and significant inhibition of class IA PI3K isoforms was detected in clone V for P110α subunit (supplementary material Fig. S3A), in clone I for p110β subunit (supplementary material Fig. S3B), and in clone IV for p110δ subunit (supplementary material Fig. S3C). Interestingly, silencing of the PI3K p110β subunit resulted in complementary increase of p110α subunit expression in shRNA p110β (see supplementary material Fig. S3B). In corroboration with the results achieved upon pharmacological inhibition of p110α with compound 15e, mRNA silencing of p110α resulted in the most pronounced inhibition of cardiovascular differentiation. Downregulation of p110δ resulted in significant decrease in the size of cardiac and vascular cell areas, whereas the percentage of spontaneously contracting EBs was not affected (supplementary material Fig. S4A,B). By contrast, a non-significant decrease in vascular areas and significantly increased cardiac areas were observed after silencing p110β (see supplementary material Fig. S4A), whereas the number of spontaneously contracting EBs was not affected (supplementary material Fig. S4B). This might be due to the upregulation of p110α achieved upon downregulation of p110β (see supplementary material Fig. S3B).

As previously described, VEGF plays a crucial role in cardiac and vascular differentiation, and acts on Flk-1⁺ cardiovascular progenitor cells (Chen et al., 2006; Cheung, 1997; Gerber et al., 1998; Gliki et al., 2002; Song et al., 2007; Yang et al., 2002; Zisa et al., 2009; Lange et al., 2009). To investigate whether PI3K class IA isoforms are essential for VEGF signaling pathways leading to cardiac and vascular differentiation of ES cells, p110α, p110β and p110δ gene-inactivated EBs were treated from day 4 to day 10 of cell culture with VEGF (500 pM). This treatment resulted in a significant increase in cardiovascular differentiation in the control pLKO.1 EBs and in p110β knockout EBs, as evaluated by quantification of cardiac and vascular cell areas (Fig. 2A–C; n=3) and western blot analysis (Fig. 2D; n=3). By contrast, stimulation of cardiovascular differentiation upon treatment with VEGF was nearly absent in EBs lacking p110α and p110δ subunits (see Fig. 2A–C). When the percentage of spontaneously contracting EBs was assessed, it was apparent that silencing of PI3K p110α subunit resulted in significant reduction of beating EBs down to 40% in comparison with control EBs, which could not be restored upon treatment with VEGF (Fig. 2E; n=3). Taken together, our experiments demonstrate that the PI3K catalytic subunits p110α and p110δ regulate cardiac and vascular differentiation of EBs derived from murine ES cells and are involved in VEGF-mediated cardiovascular signaling pathways.

The importance of PKC in the VEGF–PI3K pathway was suggested by several authors (Chou et al., 1998; Dutil et al., 1998; Gliki et al., 2002; Le Good et al., 1998), but the actual mechanisms of PKC stimulation by VEGF are not known. To evaluate the role of PKC during cardiovascular differentiation of ES cells, EBs were treated from day 4 to day 10 of cell culture with inhibitors of the PKC family, i.e. bisindolylmaleimide-1 (BIM-1), GÖ 6976 (inhibitor of PKC α/βII), rottlerin (inhibitor of PKCδ) and myristoylated PKCζ peptide (myr-PKCζ peptide). Immunohistochemistry (Fig. 3A,B; n=4) and western blot (Fig. 3C; n=4) showed that vasculogenesis was decreased in EBs preincubated with PKC inhibitor BIM-1 (10 nM and 1 μM), the PKCα/βII inhibitor GÖ 6076 (2.3 nM), and the PKC δ inhibitor rottlerin (0.5 μM) (Fig. 3A–C). Inhibition of PKCδ showed the most efficient downregulation of vascular areas. By contrast, myr-PKCζ peptide (50 μM) did not influence vascular differentiation in EBs, thus excluding a role of PKCζ in vasculogenesis of ES cells (see Fig. 3A–C). Above all, pharmacological inhibition with PKC inhibitors did not affect cardiac marker protein expression (see Fig. 3C; n=3) and the percentage of spontaneously contracting EBs (Fig. 3D; n=3), thus excluding a role of PKC for commitment of the cardiac cell lineage and pointing towards the notion that PKC regulates a cardiomyogenesis-independent signaling pathway of vasculogenesis.

To elucidate a potential crosslink between the VEGF-PI3K pathway and activation of PKC, EBs were incubated during the time of VEGF treatment with wortmannin (1 μM), LY294002 (50 μM), Akt inhibitor (50 μM) and VEGF receptor inhibitor SU5614 (1 μM) at day 5 of cell culture (Fig. 4A–D; n=3). It was apparent that PI3K inhibition significantly abolished the VEGF activation of PKCα/βII (Fig. 4A,B) and PKCδ (Fig. 4A,C) but not of PKCζ (Fig. 4A,D). Treatment the EBs with SU5614 totally blunted the activation of PKC (Fig. 4A–D), suggesting a significant role of Flk-1 in activation of all PKC isoforms. By contrast, treatment the EBs with Akt inhibitor did not affect PKC activation by VEGF (see Fig. 4A–D), thus indicating that activation of PKC by VEGF is mediated via PI3K but not by Akt downstream of PI3K. Furthermore our data strongly support a role of PKCα/βII and PKCδ downstream of PI3K for cardiac-independent vasculogenesis of ES cells.

It is well established that Akt and phosphoinositide-dependent kinase-1 (PDK1) are downstream targets of PI3K and both become active through phosphorylation (Cantley, 2002; Gerasimovskaya et al., 2005). To investigate the involvement of Akt and PDK1 in VEGF-mediated regulation of the PI3K–PKC pathway, in a first step the phosphorylation of Akt and PDK1 was investigated using pharmacological inhibition of PI3K and PKC. After VEGF stimulation of 5-day-old EBs, Akt (Fig. 5A) and PDK1 (Fig. 5B) became phosphorylated within 10 minutes, which was blunted after preincubation with the PI3K inhibitors wortmannin or LY294002. By contrast, preincubation with the general PKC inhibitor BIM-1 did not reduce Akt or PDK1 phosphorylation (Fig. 5A,B). These findings indicate that stimulation of Akt and PDK1 phosphorylation upon VEGF treatment was mediated through a PI3K pathway but did not depend on PKC.

To further specify the PI3K IA catalytic subunits involved in Akt and PDK1 activation upon treatment of EBs with VEGF, p110α, p110β and p110δ gene-inactivated EBs (5-day-old) were incubated with VEGF and phospho-Akt or phospho-PDK1 and analyzed. VEGF-mediated Akt and PDK1 phosphorylation was absent in p110α and p110δ gene-inactivated cells, whereas stimulation of Akt and PDK1 phosphorylation was still present in p110β gene-inactivated cells (Fig. 6A,B; n=3). In addition, it was found that in the presence of Akt inhibitor VEGF still activated PDK1, thus suggesting that PDK1 activation upon VEGF treatment occurs via a signaling cascade that is independent of Akt and might be located upstream of PKC (Fig. 6B). To further substantiate our observation of a relation between VEGF and the PI3K–Akt–PDK1 signaling pathway, PI3K catalytic subunit gene-inactivated EBs were treated with the Flk-1 inhibitor SU5614 (1 μM) during VEGF treatment. This treatment totally inhibited Akt and PDK1 phosphorylation (Fig. 6C,D; n=3) which clearly shows that Akt and PDK1 are downstream of Flk-1. Taken together, our data demonstrate that Akt and PDK1 act downstream of the Flk-1–PI3K pathway and might regulate independent signaling pathways of cardiac-associated versus cardiac-independent vasculogenesis.

The data from the present study demonstrate that VEGF stimulation of differentiating ES cells resulted in cardiac and vascular differentiation. We also demonstrated that inhibition of PKC isoforms abolished vasculogenesis, whereas cardiomyogenesis was maintained. Because PKC inhibition did not exert effects on VEGF-mediated Akt phosphorylation, we hypothesized that cardiomyogenesis and cardiomyogenesis-associated vasculogenesis might be regulated by a PI3K–Akt-dependent but PKC-independent pathway. To determine whether Akt was involved in cardiomyogenesis-associated vasculogenesis, EBs were incubated with Akt inhibitor in the presence or absence of VEGF. This treatment resulted in a significant decrease in cardiac as well as vascular differentiation in EBs preincubated with Akt inhibitor, as evidenced by immunohistochemistry (Fig. 7A,B; n=3) and western blot analysis (Fig. 7C; n=3). When the percentage of spontaneously contracting EBs was assessed, it was apparent that inhibition of Akt resulted in significant reduction of beating EBs down to 7±7% and 10±5% in comparison with control EBs and EBs treated with 500 pM VEGF, respectively. Moreover, the reduction in the percentage of beating EBs could not be restored upon VEGF treatment in the EBs preincubated with Akt inhibitor (Fig. 7D; n=3). Taken together, our experiments demonstrate that the VEGF–PI3K–Akt signaling pathway is essential for cardiomyogenesis and vasculogenesis, whereas PCK activated via the VEGF–PI3K pathway is only involved in vasculogenesis.

The effects of pharmacological inhibition of PI3K and PKC signaling on cardiovascular differentiation of pluripotent ES cells might result from unspecific effects exerted on different cell types present in EBs and regulating cardiovascular commitment. We therefore isolated Flk-1⁺ cardiovascular progenitor cells from 4-day-old EBs by magnetic cell sorting (MACS), i.e. during the culture time when Flk-1 protein expression was maximal (supplementary material Fig. S5A; n=3). Fluorescence-activated cell sorting (FACS) resulted in a purity of approximately 91% Flk-1⁺ cells that expressed the vascular marker CD31 and the cardiac marker Nkx2.5 (supplementary material Fig. S5B; n=3). Flk-1⁺ cells were incubated with the PI3K inhibitor wortmannin (1 μM); the class 1A PI3K catalytic subunits inhibitors compound 15e (0.5 μM), TGX-221 (1 μM) and IC-87114 (1 μM); the PKC inhibitors BIM-1 (1 μM), GÖ 6976 (2.3 nM), rottlerin (0.5 μM) and myr-PKCζ peptide (50 μM); and Akt inhibitor (50 μM) in the presence or absence of VEGF (500 pM). Subsequently, mRNA expression of Flk-1, the vascular markers CD31 and VE-Cadherin, the cardiac transcription factors Nkx2.5 and GATA4, and the cardiac genes encoding MLC2v, cTNT, α-MHC and β-MHC were analyzed (Fig. 8A–F; n=3). Additionally, EBs grown from Flk⁺ cells were cultivated from day 4 until day 10 on a matrigel-coated surface, and were exposed to the PI3K and PKC inhibitors (Fig. 9A,B; n=3). A clear reduction in vascular sprout formation and reduction in mRNA expression, analyzing typical endothelial as well as early and late cardiac genes, were found after exposure to wortmannin (1 μM), 15e (0.5 μM) and IC-87114 (1 μM) either in the presence or absence of VEGF. In corroboration of our results achieved with unselected ES cell inhibition of PKC using BIM-1, GÖ 6976 and rottlerin as well as Akt inhibitor abolished the effects of VEGF on vascular marker expression, whereas the myr-PKCζ peptide was without effect. By contrast, cardiac transcription factor and cardiac marker expression was only downregulated upon treatment with PI3K inhibitors (with the exception of TGX-221) and Akt inhibitor, whereas PKC inhibitors were without effect. Notably, Flk-1 expression was downregulated by PI3K inhibitors and Akt inhibitor but not by PKC inhibitors in Flk-1⁺ cells isolated from EBs. These data clearly support the notion that PI3K catalytic subunits p110α and p110δ are central to cardiovascular differentiation of Flk-1⁺ progenitor cells, and that Akt downstream of PI3K is involved in both cardiomyogenesis and vasculogenesis, whereas PKC is involved only in vasculogenesis.

Recent research has shown that VEGF-regulated signaling pathways are not only involved in vasculogenesis but also in cardiomyogenesis by acting on VEGFR2⁺ cardiovascular progenitor cells in vitro in ES cells and in vivo in the embryonic heart. These progenitor cells are distinct from VEGFR2⁺ vasculogenic progenitor cells present in hemangioblasts from the yolk sac (Iida et al., 2005; Kattman et al., 2006), thus raising the issue of how cardiomyogenesis and vasculogenesis are controlled during embryogenesis. PI3K signaling pathways are known to be regulated by VEGF (Hamada et al., 2005; Gliki et al., 2002; Gerber et al., 1998), suggesting the possibility that signaling pathways involved in embryonic vasculogenesis are diversifying downstream of PI3K. Hence, the present study was undertaken to evaluate the role of PI3K class IA (consisting of PI3Kα, PI3Kβ and PI3Kδ isoforms), their downstream effector pathways and their interrelation with the PKC family to dissect the mechanisms of cardiomyogenesis versus vasculogenesis in differentiating ES cells.

Treatment of differentiating ES cells cultures with the PI3K inhibitors wortmannin and LY294002 reduced not only the area of α-actinin-positive cardiomyocytes and the number of EBs containing beating foci, but also cell areas of blood vessel-like structures. These results were consistent with the previously demonstrated link between PI3K and cardiovascular differentiation during embryogenesis and ES cell differentiation (Shiojima and Walsh, 2002; Klinz et al., 1999; Jiang et al., 2000; Sauer et al., 2000).

PI3Ks probably stimulate cardiovascular differentiation through either one or several isoforms. In this context, we demonstrated that targeting class IA PI3K catalytic subunit p110 using specific pharmacological inhibitors for p110α (compound 15e) and p110δ (IC-87114) impaired cardiovascular differentiation, whereas inhibition of p110β by TGX-221 was without significant effects. We additionally applied the shRNA technique to knockdown class IA PI3K isoforms in ES cells. Unexpectedly, it was observed that silencing the p110β subunit led to upregulation of p110α, which might reflect a potential cross-talk of the p110β and p110α isoforms. It has been previously reported that, in some cases, deletion of the PI3K regulatory subunits also impaired the expression level of the corresponding catalytic subunits. Similarly, it has been shown that knockout of genes encoding the class IA p110 catalytic subunits leads to an overexpression of the regulatory subunits, which might block or activate other signaling pathways (Vanhaesebroeck et al., 1997).

Previous studies showed that VEGF is essential for cardiomyogenesis and/or vascular differentiation (Chen et al., 2006; Cheung, 1997; Gerber et al., 1998; Gliki et al., 2002; Song et al., 2007; Yang et al., 2002; Zisa et al., 2009; Lange et al., 2009; Bekhite et al., 2010) and might activate PI3K (Bos 1995; Gerber et al., 1998). Moreover, it has been suggested that Flk-1⁺ plays a crucial role in regulating PI3K activity in endothelial cells (Gerber et al., 1998; Ferrara, 1999; Thomas and Owen, 2008). To investigate whether the PI3K class IA isoforms are essential for VEGF–Flk-1⁺ signaling pathways leading to cardiovascular differentiation, PI3K p110 isoform gene-inactivated ES cells were treated with VEGF. A significant increase in cardiovascular differentiation upon treatment of EBs with VEGF was observed in the pLKO.1 EBs (negative control) and P110β knockdown cell lines. By contrast, the increase of cardiomyogenesis and vascular differentiation by VEGF was totally blunted after silencing the p110α and p110δ subunits, thus corroborating our experiments with specific pharmacological inhibitors of p110α and p110δ. These data were consistent with previously published data demonstrating that PI3K determines heart size in mice (Shioi et al., 2000). Furthermore, the data corroborated studies of others demonstrating that PI3K class IA was required for cardiovascular differentiation in mouse embryos in the presence of VEGF (Luo et al., 2005; Yuan et al., 2008), and in differentiating ES cells where VEGF was found to significantly enhance α-MHC, cardiac troponin I and Nkx2.5 expression (Chen et al., 2006).

To investigate the interrelation between the PI3K signaling pathway and PKC activation and its impact for cardiomyogenesis versus vasculogenesis, ES cells were treated with inhibitors of the PKC family. It was evident that BIM-1, GÖ 6976 and rottlerin significantly inhibited vasculogenesis. Notably, cardiomyogenesis remained unimpaired after treatment of EBs with BIM-1, GÖ 6976, rottlerin or myr-PKCζ peptide inhibitor, which indicates that the signaling pathways regulating cardiac and vascular differentiation of ES cells diversify downstream of PI3K. It furthermore demonstrated that PKC downstream of PI3K regulates vasculogenesis but is not involved in cardiomyogenesis. The importance of PKC regarding the VEGF-mediated PI3K pathway was previously suggested by several authors (Chou et al., 1998; Dutil et al., 1998; Gliki et al., 2002; Le Good et al., 1998). Others provided direct evidence that the PI3K–Akt pathway is required for VEGF to induce endothelial cells, but a possible involvement of PKC was not specified (Shiojima and Walsh, 2002; Abid et al., 2004; Qi and Claesson-Welsh, 2001).

To elucidate a potential crosslink between the VEGF–PI3K pathway and activation of PKC, EBs were incubated during the time of VEGF treatment with general PI3K inhibitors, which significantly blunted the activation of PKCα/βII and PKCδ but not of PKCζ. Treatment of EBs with SU5614 totally abolished the activation of all PKC isoforms, thus confirming that VEGF-mediated PKC activation occurs via Flk-1. By contrast, treatment of EBs with Akt inhibitor did not affect PKC activation by VEGF, which indicates that the PI3K–Akt pathway is apparently distinct from the PI3K–PKC axis. Furthermore, PDK1 was activated in response to VEGF treatment independently of Akt, as indicated by the fact that PDK1 phosphorylation upon VEGF treatment still occurred in the presence of Akt inhibitor. This finding suggests that the VEGF–PI3K–PDK1 signaling cascade is acting upstream of PKCα/βII and PKCδ in differentiating EBs. Comparable data have been previously obtained in human embryonic kidney cells, where PKC isotypes were controlled by PI3K through PDK1 (Le Good et al., 1998).

Because inhibition of PKC abolished vasculogenesis but not cardiomyogenesis of ES cells we hypothesized that cardiomyogenesis-associated vasculogenesis might utilize a PKC-independent signaling pathway, which could be regulated by Akt downstream of PI3K. Akt-dependent signaling pathways are known to be involved in the regulation of cardiac growth, contractile function, and coronary angiogenesis (Shiojima and Walsh, 2006; Fujio et al., 2000; Matsui et al., 1999), suggesting a possible role in pathways regulating cardiovascular differentiation from cardiovascular progenitor cells in the heart. Our assumption was firstly supported by experiments demonstrating that silencing of the PI3K catalytic subunits p110α and p110δ abolished VEGF-mediated Akt and PDK1 phosphorylation and likewise blunted PKC activation, which is presumably mediated via the PDK1 pathway. Secondly, we observed that treatment of VEGF-stimulated EBs cells with Akt inhibitor not only inhibited vasculogenesis but also cardiomyogenesis, despite a persistent PKC activation under these experimental conditions. Hence, our findings show that Akt regulates cardiovascular differentiation downstream of the PI3K in VEGF-stimulated ES cells.

The biological activities of VEGF are mainly mediated by Flk-1 (Ferrara et al., 2003), which is one of the early lateral mesoderm markers, when cardiomyogenesis occurs. It has been shown that Flk-1⁺ cells could act as cardio-hemangioblasts to form cardiomyocytes as well as endothelial cells as constituents of the coronary and endocardial vasculature (Iida et al., 2005; Kattman et al., 2006). Our results demonstrate that VEGF via Flk-1 stimulates the PI3K–Akt axis to initiate cardiomyogenesis as well as cardiac-associated vasculogenesis, and stimulates PI3K–PDK1–PKC signaling to promote vasculogenesis independently of cardiomyogenesis. These effects are not only present in whole mount EBs, which differentiate various potentially interacting cell types of all three germ layers, but also in isolated Flk-1⁺ cardiovascular progenitor cells. More precisely, the data from our study demonstrate that PI3K class IA isoforms PI3Kα and PI3Kδ control cardiomyogenesis and vascular differentiation, whereas PKCα/βII and PKCδ are involved in vasculogenesis upon VEGF treatment but are dispensable for cardiomyogenesis (see Fig. 10). Apparently, signaling pathways in cardiomyogenesis and vasculogenesis are closely associated to synchronize and phase-lock the development of the cardiovascular system.

The ES cell line CGR8 was obtained from the European Collection of Cell Cultures (ECACC) Wiltshire, UK and cultured on gelatine-coated cell culture flasks in Glasgow minimal essential medium (GMEM; Sigma-Aldrich, Taufkirchen, Germany). Medium was supplemented with 10% inactivated fetal bovine serum (FBS) (Sigma-Aldrich), 2 mM L-glutamine (Biochrom, Berlin, Germany), 45 μM 2-mercaptoethanol (Sigma-Aldrich) and 103 U/ml leukaemia inhibitory factor (LIF) (Chemicon, Hampshire, UK) in a humidified environment containing 5% CO₂ at 37°C. At day 0 of differentiation, adherent cells were enzymatically dissociated using 0.25% trypsin with 1 mM ethylenediaminetetraacetic acid (EDTA) in Hanks' balanced salt solution (HBSS) (Invitrogen, Karlsruhe, Germany). A total of 1×10⁷ cells was seeded into spinner flasks (Integra Biosciences, Fernwald, Germany) containing 125 ml Iscove's medium supplemented with the same additives as described above but devoid of LIF. Following 24 hours, 125 ml medium was added to give a final volume of 250 ml. The spinner flask medium was stirred at 20 r.p.m. using a stirrer system (Integra Biosciences), and 125 ml cell culture medium was exchanged every day.

The growing EBs were removed from spinner flasks and plated at day four of culture either on tissue culture dishes (Greiner Bio-One, Frickenhausen, Germany) or for enhanced cardiac differentiation on ‘lumox’ cell culture dishes (Sigma-Aldrich).

pLKO.1-puro (Sigma-Aldrich) derivative plasmid carrying shRNA sequence targeting PI3K p110α catalytic subunit, shRNA targeting PI3K p110β catalytic subunit, and shRNA targeting PI3K p110δ catalytic subunit were separately introduced into CGR8 cells by lentiviral particles. Transduced cells with pLKO.1 vector (containing non-hairpin insert) were used as negative control. Lentiviral particles were generated using human embryonic kidney (HEK) 293FT-based amphotropic Phoenix packaging cells (Phoenix-Ampho, Invitrogen, Karlsruhe, Germany) cultured in DMEM, 10% fetal calf serum, 1% sodium pyruvate plus penicillin/streptomycin. For a 10-cm dish, lentiviral vector plasmids (10 μg) were co-transfected with plasmids encoding the HIV-Rev (5 μg), HIV-MDL (10 μg) and the ecotropic envelope (2 μg) in the presence of polyethylenimine (70 μg, Sigma-Aldrich). Supernatants were harvested after 24 hours and 48 hours and filtered through 0.22-mm filters (Schambach et al., 2006). Supernatants were added to CGR8 cells, which were plated in a well of a 6-well plate, centrifuged at 400 g in a cell culture centrifuge for 1 hour and replaced with fresh medium on the following day. A total of three infection rounds were carried out within 48 hours. On the following day, the cells were passaged and selected with 2 μg/ml puromycin (Sigma-Aldrich) for 10–14 days. Transduction effectiveness was assessed via GFP control vector (Sigma-Aldrich) as previous described (Sauer et al., 2008). Downregulation of PI3K p110 catalytic subunit p110α, p110β and p110δ was analyzed by RT-PCR and western blot.

EBs differentiated from ES cells were incubated in the presence or absence of VEGF-165 (500 pM) (Sigma-Aldrich) with wortmannin (1 μM), LY294002 (50 μM) (both from Cell Signaling Technology, Frankfurt, Germany), bisindolylmaleimide I (BIM-1) (10 nM/1 μM) (Calbiochem, Bad Soden, Germany), or with inhibitors of p110 catalytic subunit by using compound 15e (0.5 μM), TGX-221 (1 μM) (Alexis Biochemicals, Lörrach, Germany) and IC87114 (1 μM), which have been shown to inhibit PI3K p110α, PI3K p110β and PI3K p110δ, respectively (Hayakawa et al., 2006; Jackson et al., 2005; Lee et al., 2006). PKC was inhibited by Gö6976 (2.3 nM), which is specific for PKCα/βII (Sigma-Aldrich), myristoylated PKCζ peptide (myr-PKCζ peptide) inhibitor (50 μM) (Biomol, Lörrach, Germany), and rottlerin (0.5 μM), which is a specific inhibitor of PKCδ (Biomol). Furthermore, Akt inhibitor (50 μM) (Calbiochem) and SU5614 (1 μM) inhibitor for VEGF receptor (Calbiochem) were used in the study.

Total RNA isolation from ES cells was performed using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to manufacturer's recommended procedures, followed by genomic DNA digestion using DNaseI (Invitrogen, Karlsruhe, Germany). cDNA was synthesized from 2 μg total RNA using a superscript II reverse transcriptase (Invitrogen) according to manufacturer's instructions and using random primer. The reverse transcription product was diluted 1:10, and PCR was performed with following primers (Sigma, Geonysis, Germany): CD31: sense 5′-GAGCCCAATCACGTTTCAGTT-3′, antisense 5′-TCCTTCCTGCTTCTTGCTAGCT-3′; VE-Cadherin: sense 5′-GTCAGCTATAGGGACCTCTGT-3′, antisense 5′-TCATTTCCTTTCACGATTTGG-3′; Nkx2.5: sense 5′-CCACTCTCGCTACCCACCT-3′, antisense 5′-CCAGGTTCAGGATGTCTTTGA-3′; Flk-1: sense 5′-GTGGACCAAATGCCTGACTC-3′, antisense 5′-TTCTGTTCTGTTGGCCCTTT-3′; α-MHC: sense 5′-TGAAAACGGAAAGACGGTGA-3′, antisense 5′-TCCTTGAGGTTGTACAGCACA-3′; β-MHC: sense 5′-CTACAGGCCTGGGCTTACCT-3′, antisense 5′-TCTCCTTCTCAGACTTCCGC-3′; MLC2v: sense 5′-AAAGAGGCTCCAGGTCCAAT-3′, antisense 5′-CCTCTCTGCTTGTGTGGTCA-3′; GATA4: sense 5′-ACTCTGGAGGCGAGATGGG-3′, antisense 5′-GACACCGCAGCATTACGGCTC-3′; PI3k p110α: sense 5′-ACTGTTCAGAGAGGCCAGGA-3′, antisense 5′-CGGTTGCCTACTGGTTCAAT-3′; PI3k p110β: sense 5′-AGCTGGTCTTCGTTTCCTGA-3′, antisense 5′-TCCACCACGACTTGACACAT-3′; PI3k p110δ: sense 5′-CTGACCCCTCATCTGACCAT-3′, antisense 5′-TCGTCAGCATTCACTTTTCG-3′; polymerase II: sense 5′-GACAAAACTGGCTCCTCTGC-3′, antisense 5′-GCTTGCCCTCTACATTCTGC-3′; Nestin: sense 5′-CGGCCCACGCATCCCCCATCC-3′, antisense 5′-AGCGGCCTTCCAATCTCTGTTCC-3′; Neuron-specific enolase: sense 5′-CCAAGTCACCCAGAACACCT-3′, antisense 5′-AAACACCCCAACACACCAAT-3′.

The PCR product was amplified using 40 cycles performed at 58°C annealing temperature. Gel images were subsequently captured using a 1% agarose gel.

The western blot assays were carried out after washing the ES cells in cold PBS and lysing in lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, and 1 mM Na₃VO₄) that contained 1 mM phenylmethylsulfonyl fluoride for 30 minutes on ice. Samples were centrifuged at 13.000 g for 10 minutes to pellet the debris. After determination of protein concentrations using the method of Bradford, 40 μg of protein per sample was heated to 95°C for 10 minutes and separated in SDS polyacrylamide gels and transferred to nitrocellulose membranes at 20 V over 10 hours. Membranes were blocked with 20% (wt/vol) dry fat-free milk powder in Tris-buffered saline with 0.1% Tween (TBST) for 5 hours at 4°C. After that, primary antibodies were incubated for 10 hours at 4°C. Subsequently membranes were washed with 0.1% TBST. For the secondary antibody reaction, membranes were incubated with the adequate horseradish peroxidase (HRP)-conjugated secondary antibodies (Santa Cruz Biotechnology) for 60 minutes at room temperature. Corresponding bands were visualized using ECL detection kit (Amersham, Freiburg, Germany) and captured by a digital imaging system (LAS 3000, Fujifilm, Japan). The following primary antibodies were used for western blot analyses: polyclonal goat-anti mouse phospho-PKCδ (Ser643) and polyclonal rabbit anti mouse phospho-PKCζ (Thr410), PI3K p110α, PI3K p110β and PI3K p110δ were all obtained from Santa Cruz Biotechnology. The monoclonal rabbit anti-mouse antibodies directed against phospho-PKCα/β (Thr638/641), phospho-PDK1 (Ser241) and phospho-Akt (Ser473) as well as the polyclonal rabbit anti mouse PDK1 and Akt antibodies were all obtained from Cell Signaling Technologies (Frankfurt, Germany). For cardiomyocyte differentiation, a monoclonal mouse anti-mouse sarcomeric α-actinin (Sigma-Aldrich), a monoclonal mouse anti-mouse myosin heavy chain (MHC), and a monoclonal anti-myosin light chain (MLC) antibody were used (both from Abcam, Cambridge, UK). For assessing endothelial differentiation, a monoclonal rabbit anti mouse Flk-1 antibody from Cell Signaling Technologies, a monoclonal mouse anti-mouse CD31 antibody and a polyclonal rabbit anti-mouse VE-Cadherin antibody (both from Abcam) were used.

For immunofluorescence stainings, the ES cells were fixed in methanol–acetone (7:3) for 1 hour at –20°C. Rat monoclonal anti-CD31 (Chemicon, Schwalbach, Germany) and monoclonal mouse anti-sarcomeric α-actinin (Sigma-Aldrich) were used. Fluorescence recordings were performed using a confocal laser scanning microscope (LSM 510; Carl Zeiss, Germany) connected to an inverted microscope (Axiovert 135; Carl Zeiss). Quantitative IHC was analyzed as previously described (Sauer et al., 2005) using an Axiovision software tool (Carl Zeiss). Briefly, images (512×512 pixels) were acquired either from CD31-stained or α-actinin-stained EBs corrected for background fluorescence using the extended depth of focus algorithm of the confocal setup. Generally, four full-frame images separated by a distance of 8 μm in the z-direction were recorded that included information on the capillary or cardiac area and spatial organization in a tissue slice 32 μm thick. From the acquired images, an overlay image giving a three-dimensional projection of the vascular and cardiac structures in the scanned tissue slice was generated. By use of the image analysis facilities of the confocal setup, the antigen-positive vascular and cardiac cell areas within the three-dimensional projection of vascular and cardiac structures were identified and related to the cross-section of the respective EB.

CGR8 EBs were generated and cultivated as described above. The 4-day-old EBs were dissociated by incubation with IK-buffer containing collagenase type II (4 mg/ml, PAA, Coelbe, Germany) at 37°C for 5–10 minutes. For cell separation, MACS columns (Miltenyi Biotec, Bergisch-Gladbach, Germany) were used. The separation procedure was carried out using a PE-conjugated rat anti-mouse Flk-1 antibody (dilution 1:10; BD, Franklin Lakes, NJ). For labeling, anti-PE MicroBeads (all from Miltenyi Biotec) were used. Positive selected cells were placed on a matrigel surface and further cultured. Purity analysis was carried out using a FacsCalibur (Becton Dickinson, Heidelberg, Germany) with a scanning wave length of 488 nm (PE-labeled cells) and a detection wave length of 575 nm.

Data in the figures are given as mean values + s.d., with n denoting the number of experiments unless otherwise indicated. In each experiment at least 25 EBs were analyzed. GraphPad InStat-3 software (GraphPad Software, San Diego, CA) was applied for One-way ANOVA or Student's t-test for unpaired data as appropriate. A value of P<0.05 was considered statistically significant. Data are presented as percentage of expression relative to control values, which were set to 100%.

This study was supported by the Interdisciplinary Center for Clinical Research (IZKF) of the Medical Faculty University Jena, the German Foundation for Heart Research and the Excellence Cluster Cardiopulmonary System (ECCPS) of the German Research Foundation (DFG).