Induction of EndMT results from integration of multiple extr

Induction of EndMT results from integration of multiple extrinsic inputs within the dynamically shifting cellular environment of cardiogenesis. Owing to the membrane bound presentation of ligands, Notch signaling status is dependent on the organization and distribution of signal-presenting and signal-receiving cells. Conversely, the TGFβ pathway acts through a nuclear receptor of secreted morphogens, but has also been shown to play a pivotal role in EndMT (Gonzalez and Medici, 2014). Knockout of the type I receptor, ALK5, which responds to TGFβ1 ligand by activation of SMAD2/3, results in failure of EndMT induction from endocardium (Sridurongrit et al., 2008); and in hESC-derived ECs activation of SMAD2/3 induces phenotypic transition that can be mitigated by co-culture with the ALK4/5/7 inhibitor SB431542 (James et al., 2010). In this study, we found increased expression of BMPER, and decreased expression of endoglin in conditions that promote EndMT of hESC-derived ECs (Fig. 4f). BMPER antagonizes BMP signaling (Moser et al., 2003) and endoglin augments TGFβ signal activation by enabling ECs to respond to TGFβ1 ligand via activation of an ALK1:SMAD1/5/8 axis instead of a ALK5:SMAD2/3 axis (Lebrin et al., 2004). Hence, the shift from endoglin to BMPER expression in confluent hESC-ECs denotes a change from SMAD1/5/8 activation to SMAD2/3 activation, and may account, in part, for reduced EC number and increased EndMT. Indeed, ALK1 (Urness et al., 2000) or endoglin (Sorensen et al., 2003) loss of function in mice results in aberrant cardiac cushion formation and, in vitro, induces a shift in cultured ECs from proliferation to growth arrest (Lebrin et al., 2004). Our data revealed an increase in BMPER expression even in control conditions, but endoglin was uniquely reduced upon JAG1 knockdown. Although TGFβ signaling functions independently to promote EndMT, these results may reflect crosstalk between Notch and TGFβ signaling to enhance SMAD2/3 activation at the expense of SMAD1/5/8 activation. Another pathway that has been linked to Notch is Nitric oxide (NO) signaling. NO signaling is a critical modulator of cardiac function (Rastaldo et al., 2007) and deficiency in NO generation in embryonic mice (Aicher et al., 2007; Feng et al., 2002; Lee et al., 2000) results in major cardiac anomalies that are similar to those displayed by human patients carrying Notch1 mutations (Garg et al., 2005). Chang et al. have linked Notch to activation of to an autocrine loop consisting of GUCY1A3, GUCY1B3, ACTIVINA and IGF2, which enables propagation of nitric oxide signaling, and ultimately EndMT in embryonic mouse heart (Chang et al., 2011). All of these factors were elevated in hESC-derived ECs under conditions that promoted EndMT (Fig. 4f), affirming that multiple signaling elements present during embryonic EndMT are recapitulated in our in vitro platform. Despite being linked to multiple downstream effectors of EndMT, including NO signaling (Chang et al., 2011) and SLUG (Niessen et al., 2008), the molecular mechanisms that tip the balance of Notch signal activation leading up to EC trans-differentiation are not well defined. The increased surface expression of JAG1 observed in cells undergoing EndMT and coincident reduction in DLL4 expression (Fig. 3) suggested a role for JAG1 in attenuating Notch activation. Benedito et al. have demonstrated the role of JAG1 in designating tip versus stalk phenotype of ECs during retinal neo-vascularization (Benedito et al., 2009). In mice with EC-specific loss of JAG1, progression of the retinal vascular plexus is marked by reduced EC proliferation, elevated DLL4 expression and increased stalk to tip cell ratio. This phenotype arises from loss of JAG1-mediated Notch inhibition in ECs expressing Fringe genes. Similarly, our knockdown of JAG1 in hESC-derived ECs resulted in decreased cell counts and increased Notch activation/DLL4 expression, as well as increased EndMT. Although not emphasized in their study, Benedito et al. noted increased pericyte coverage and ectopic localization of vascular smooth muscle cells to venules in JAG1 loss-of-function mice. Because EC-specific loss of JAG1 in these mice may be partially rescued by surrounding non-endothelial cells, this phenotype may be a milder correlate to the effect observed in our in vitro platform, in which JAG1 loss of function was uniformly applied across a purified population of ECs.