VEGF modulates early heart valve formation



Although hypoxic and/or nutritional insults during gestation are believed to contribute to congenital heart defects, the mechanisms responsible for these anomalies are not understood. Given the role vascular endothelial growth factor (VEGF) plays in response to hypoxia, it is a likely candidate for mediating deleterious effects of embryonic hypoxia. The ectopic or overproduction of endogenous factors such as VEGF may contribute to specific heart defects. Here we compared hypoxia-induced precocious production of VEGF during early heart valve development to normal VEGF production. Mouse prevalvular cardiac endocardial cushions were explanted onto hydrated type I collagen gels under normoxic or hypoxic conditions. The extent of transformation of cardiac endothelium into mesenchyme was inversely correlated with the levels of VEGF during the various culture conditions. A soluble VEGF antagonist confirmed that endogenous production of VEGF was specific for blocking normal cushion mesenchyme formation. We further demonstrated that E10.5 endocardium retains the ability to transform into cardiac mesenchyme in the absence of endogenous VEGF. Anat Rec Part A 271A:202–208, 2003. © 2003 Wiley-Liss, Inc.

The angiogenic response is essential for a variety of physiological and pathological processes. The importance of angiogenesis during heart development is demonstrated by a clinical association between congenital heart defects and gestational hypoxia (DeSesso, 1987; Lueder et al., 1995). VEGF is known to act in restoring hypoxic mature tissue to normal oxygen homeostasis by inducing the generation of new blood vessels (Dor and Keshet, 1997). Strict control of VEGF levels is required to control the angiogenic response and prevent deleterious effects (Lee et al., 2000; Isner, 2001). In embryonic mouse development, a two- to threefold increase in endogenous VEGF production results in midgestation lethality (Miquerol et al., 2000). These embryos exhibit cardiac defects, such as overdeveloped trebeculae and coronary vessels, as well as septation abnormalities. Conversely, the loss of one VEGF allele results in early embryonic lethality due to cardiovascular defects (Carmeliet et al., 1996; Ferrara et al., 1996). This haploinsufficient phenotype exhibits underdeveloped endocardial cushions and chamber malformations in addition to impaired vascular development. In a controlled transgenic system, we demonstrated that production of VEGF a full day earlier (E9.5 vs. E10.5) results in septal and valve defects arising from malformed endocardial cushion tissues (Dor et al., 2001). These findings, which demonstrate a dependence on appropriate VEGF production for normal valve formation and chamber septation, appear to correlate with septal defects observed from hypoxia or exposure to air pollutants during early gestation (Miao et al., 1988; Ritz et al., 2002). These results suggest that precocious increases in VEGF may prematurely terminate epithelial–mesenchymal transformation (EMT) in the endocardial cushions, and contribute to chamber septation and valvular defects.

At present, it is not clear whether hypoxia induces the production of physiologic levels of VEGF during cardiac development, or excess VEGF production prior to endocardial cushion morphogenesis leads to septation defects. In addition, it is unknown whether cessation of cardiac endothelial transformation is regulated exclusively by VEGF, or there is a requirement for additional mediators. At a specific time in development (E9.0–E10.0), inductive signals derived from the cardiac myocardium activate endothelial cells in these regions to undergo EMT. Cell transformation within the endocardial cushions occurs as a subset of endothelial cell hypertrophy, disengage cell–cell adhesions, extend filopodia, and invade the extracellular matrix (Markwald et al., 1979; Eisenberg and Markwald, 1995). In the current study we used an in vitro collagen gel invasion assay to reproduce the morphogenetic events of atrioventricular canal (AVC) morphogenesis (Runyan and Markwald, 1983; Camenisch et al., 2002). In this context, endocardial cushion tissue from E9.5 and E10.5 mouse embryos were explanted and examined in normoxic and hypoxic conditions. Levels of VEGF were correlated with mesenchymal formation and/or expansion of the AVC endocardium. The results indicate that precocious production of physiologic levels of VEGF can be induced by hypoxia during this critical stage of heart development. These observations are the first to define the normal events initiated by VEGF immediately following EMT (∼E10.5). VEGF acts to terminate the formation of mesenchymal progenitors that are required for chamber septation and cardiac valve formation.


AVC Explants

Mouse AVC endocardial cushion “explants” from wild-type “FVB” mouse embryos were isolated from timed fertilized female mice, and embryonic age was verified by somite number. Experimental conditions were optimized for mouse endocardial cushion cultures to obtain morphologic events similar to those in the established avian system (Camenisch et al., 2002). Briefly, M199 media was used for gel casting in four-well microculture dishes (Nalge Nunc, Naperville, IL). OPTI-MEM medium (Gibco BRL, Rockville, MD) plus 0.01% ITS (Gibco-BRL) was used to hydrate the polymerized type I collagen gels before placement of the mouse AVC endocardial cushion tissue. Mouse explants required 12 hr for attachment to the collagen gel surface at 37°C, 5% CO2 prior to the application of 0.1 ml of M199 supplemented with 1% fetal calf serum (Hyclone, Logan, UT), 50 U/ml penicillin, 50 μg/ml streptomycin, and 0.01% ITS (Gibco-BRL). Explants were cultured for 48 hr at 37°C, 5% CO2 under normoxic (21%), hypoxic (10%), or anoxia conditions produced by a BBL GasPak anaerobic system (Becton Dickinson, Sparks, MD). For explants treated with sFlt (25 ng/ml), the recombinant chimeric receptor was immediately added at the time the AVC was placed onto the gel surface. The extent of endocardial migration or spread, and mesenchymal cell invasion was assessed using differential interference contrast (DIC) optics, as previously described (Camenisch et al., 2002; Dor et al., 2001). Briefly, the morphologic criterion used for endocardial cell classification was the appearance of rounded, polygonal cells on the collagen gel surface with intact cell–cell junctions. The primary morphologic criterion for mesenchymal cell invasion was the appearance of cells with characteristic stellate appearance within the gel matrix (Runyan and Markwald, 1983). Images were captured using a Nikon microscope with Spot image software (Diagnostic Instruments, Sterling Heights, MI).

Analysis of VEGF

VEGF levels were determined from culture supernatants collected following 48 hr of incubation using a specific mouse VEGF assay (Oncogene Research Products Inc., Boston, MA). Briefly, harvested supernatants from three AVC explants per well were diluted 1:2 in buffer and treated according to the manufacturer's instructions for the detection of VEGF, with levels calculated to pg/ml per AVC explant. A minimum of six wells was used to calculate VEGF levels for each condition. A paired-sample t-test was used to compare VEGF concentrations between experimental groups. A P-value < 0.05 between experimental groups and normoxic controls was considered significant.


Hypoxia and Post-transformation Isolated AVC Explants Exhibit Reduced Mesenchymal Cell Formation and Expansion of the Cardiac Endothelium

Embryonic endocardial cushions isolated from the AVC were cultured on type I collagen gels to recapitulate the events of EMT during AVC morphogenesis (Runyan and Markwald, 1983). Mouse AVC explants isolated prior to cell transformation in vivo (E9.5) demonstrated EMT with little to no endothelial cell expansion in the collagen gel invasion assay (Fig. 1A) (Lakkis and Epstein, 1998; Camenisch et al., 2000). These mesenchymal cells extended filopodia, and invaded the collagen matrix. In contrast, AVC explants at the post-transformation stage (E10.5) showed a substantial expansion of a cardiac endothelial cell monolayer, and minimal transformation (Fig. 1B). The cells typically maintained cell–cell borders and did not invade the collagen gel. When E9.5 AVC explants were exposed to hypoxic culture conditions, they exhibited a phenotype similar to that of post-transformation E10.5 explants, with an expanded sheet of endothelial cells and reduced mesenchymal cell formation (Fig. 1C). Anoxia resulted in a similar post-EMT phenotype, but with increased cell toxicity (Semenza, 1999) and patches of endocardial cells (Fig. 1D) (Dor et al., 2001). Thus, hypoxia at E9.5 appears to circumvent endocardial cushion EMT, resulting in a “post-EMT” phenotype observed normally with E10.5 AVC endocardial cushion tissue (Fig. 1C). This response to hypoxia appears to override the normal transformation events required for endocardial cushion mesenchyme formation.

Figure 1.

AVC in vitro morphogenesis. AVC explants cultured for 48 hr on hydrated type I collagen gels under the indicated conditions: (A) normoxic E9.5 AVC, (B) normoxic E10.5 AVC, (C) E9.5 AVC under 10% hypoxia, and (D) E9.5 AVC under anoxic conditions. There is a normal EMT in part A, but EMT is drastically reduced or ablated in B–D. A representative image is shown from a minimum of 10 explants for each condition. M, myocardium. Scale bar in A = 100 μm, and scale bar in D (for B–D) = 100 μm.

Post-EMT Phenotype Correlates With Increased VEGF Production

VEGF, which is produced in response to hypoxia, antagonizes EMT. VEGF protein levels produced by E9.5 and E10.5 AVC explants cultured under normoxic conditions were compared to E9.5 AVC explants subjected to hypoxic and anoxic environments. An elevation in VEGF concentration for endocardial cushions cultured under normoxic conditions was observed between stages E9.5 and E10.5 (Fig. 2), supporting earlier in vivo findings of an increase in VEGF mRNA during this period (Dor et al., 2001). Both hypoxia and anoxia induced precocious VEGF production in E9.5 AVC explants to levels significantly greater than the endogenous levels normally produced at the E9.5 stage under normoxic conditions. The 10% O2 environment with E9.5 explants produced VEGF amounts comparable to naïve E10.5 explants that were increased approximately 12-fold over control E9.5 AVC explants (43.12 and 45.1 vs. 3.35 pg/ml/AVC, respectively) (Fig. 2). These elevated doses are within the in vivo range for VEGF observed in embryonic hearts post-EMT (∼E12.5) (Miquerol et al., 2000). The elevated VEGF concentrations detected from both E10.5 and hypoxic culture conditions with E9.5 AVC explants correlates with the observed endothelial cell phenotype and decreased EMT (Fig. 1).

Figure 2.

VEGF production during in vitro AVC morphogenesis. VEGF was detected in each experimental group, as described in the Methods section. Reduced levels in anoxia and 5% O2 are speculated to reflect reduced overall cellularity due to toxicity under these conditions (Semenza, 1999). Concentration of VEGF shown as pg/ml per AVC explant. *P < 0.001 and **P < 0.01. Values were determined from independent experiments as described in the Methods section.

VEGF Is Sufficient for Inducing the Post-EMT Phenotype

The remodeling events following EMT probably depend on multiple factors, but VEGF may be a central switch that demarcates this transition step away from EMT and toward remodeling of the rudimentary valve tissue (Camenisch et al., 2002). In this regard, a soluble VEGF receptor 1 chimeric protein (sFlt), which functions as a soluble antagonist of VEGF signaling (Gerber et al., 1999), was used to determine whether the hypoxia-induced cellular responses at E9.5 are specific to VEGF. When cultured in the presence of the inhibitor sFlt, hypoxic E9.5 AVC cultures reverted to the normal E9.5 phenotype with mesenchymal cell formation and invasion (Fig. 3D vs. 3C), similarly to anoxic E9.5 cultures with sFlt (Dor et al., 2001). We further examined the capacity of sFlt to attenuate the post-EMT endocardial expansion observed in normal E10.5 AVC explants (Fig. 3A). E10.5 AVC cultures treated with sFlt demonstrated extensive mesenchymal cell formation and reduced endocardial outgrowth (Fig. 3B). This phenotype is most similar to normoxic E9.5 AVC explants (Fig. 1A), and shows that neutralizing VEGF promotes EMT. These results define a critical temporal requirement for the production of VEGF during early cardiac valve formation. These observations also indicate that E10.5 AVC explants maintain the capacity to form cardiac mesenchyme. Furthermore, our results indicate that VEGF is a predominant mediator produced during embryonic hypoxia, and a key negative regulator of EMT within the AVC endocardial cushions.

Figure 3.

Neutralization of VEGF restores EMT in hypoxic and E10.5 AVC explants: (A) normoxic E10.5 AVC (control), (B) E10.5 with sFlt inhibitor, (C) E9.5 AVC under 10% hypoxia, and (D) E9.5 AVC under 10% hypoxia with sFlt inhibitor. Dashed line indicates footprint of removed myocardium for ease of visualization. A representative image is shown for each condition from observations of a minimum of eight AVC explants. Scale bar = 100 μm.


VEGF is a critical factor during embryonic vasculogenesis and angiogenesis, much of which is thought to be stimulated by temporal hypoxic environments. A recent survey of embryonic tissues revealed that most organs, including the heart, experience hypoxia during development (Lee et al., 2001). Cardiac chamber and vascular endothelium malformations observed in HIF-1α-deficient mice also suggest that localized hypoxia most likely occurs in embryos to induce vessel development related to organogenesis (Iyer et al., 1998; Ryan et al., 1998; Semenza et al., 1999). However, no direct evidence has linked hypoxia with induction of cardiac VEGF production during wild-type embryonic development. In this study, we defined the developmentally programmed production of VEGF during in vitro AVC endocardial cushion morphogenesis, as well as the induction of VEGF by hypoxia. We detected a significant increase in the amount of VEGF protein produced by the AVC between stages E9.5 and E10.5. During this period, EMT produces mesenchymal cells that populate the prevalvular endocardial cushions. Our current findings support a previous report (Dor et al., 2001) which showed that VEGF mRNA is normally expressed in the AVC following transformation events in vivo. Following cardiac AVC mesenchyme formation, VEGF appears to down-regulate EMT, and may stimulate the endocardium of the AVC to proliferate. This capacity may serve to maintain the integrity of endothelial cell junctions following EMT, and antagonize subsequent cell invasion into the cardiac jelly (see Fig. 4).

Figure 4.

Schematic model comparing events during normal conditions and hypoxic conditions for AVC endocardial cushion morphogenesis. Proceeding from top to bottom, events are depicted chronologically from E9.5 to E11.5. Precocious production of VEGF during hypoxia decreases EMT, resulting in decreased remodeling and maintenance of cushion volume. Text in bold during hypoxia conditions highlights differences compared to the normoxia or normal events for EMT related to endocardial cushion morphogenesis. M, myocardium; E, endocardium; Ec, endocardial cells; ECM, extracellular matrix; Cm, cushion mesenchyme. Up arrows and down arrows denote increase and decrease, respectively, and + indicates production or activity.

The strict spatiotemporal production of VEGF during heart morphogenesis emphasizes the necessity of regulating the concentrations of this growth factor during development. The loss of one VEGF allele results in early embryonic lethality due to cardiovascular defects (Carmeliet et al., 1996). Likewise, a premature increase in endogenous VEGF production results in midgestation lethality (Miquerol et al., 2000). These transgenic embryos exhibit vasculature and cardiac abnormalities, such as septal defects, due to precocious induction of endocardial development. In this regard, we have demonstrated that a normal increase in endogenous VEGF produced by myocardium of the AVC in vitro terminates EMT. This level of production is in accordance with a previous study by Miquerol et al. (2000), who reported that a two- to threefold increase in VEGF can elicit endothelial changes, and if mistimed can have deleterious developmental impact. Our data show that a premature elevation in VEGF, such as that induced by hypoxia, results in decreased mesenchyme formation along with an early angiogenic response by the endocardium of the developing AVC (Fig. 4). Furthermore, the addition of exogenous VEGF165 (10–100 ng/ml; Pepro Tech, Rocky Hill, NJ) to normal E9.5 AVC explants reproduces the post-EMT phenotype (data not shown). Collectively, these findings indicate that strict spatiotemporal regulation of VEGF production is required to establish sufficient cardiac mesenchyme within the early valve tissue.

The current study emphasizes the importance of the precise regulation of VEGF production during cardiovascular development, particularly in AVC morphogenesis. VEGF is a well characterized mediator of responses to hypoxia and other environmental insults (Carmeliet and Jain, 2000). VEGF recruits new vessels to hypoxic tissues to restore oxygen homeostasis in a variety of pathological states and model systems (Dor and Keshet, 1997). In this regard, congenital heart defects are more prevalent at high altitudes (Miao et al., 1988), and increased cardiac anomalies resulting from experimental hypoxia have been shown in several animal models (Ingalls et al., 1952; Clemmer and Telford, 1966; Jaffee, 1974). Although the mechanisms mediating these effects have not been elucidated, congenital heart defects occurring from gestational hypoxia likely depend on the onset and duration of the hypoxic stress. Insufficient endocardial cushion mesenchyme formation resulting from premature VEGF activity could contribute to structural heart anomalies, such as atrial septal defects, which can arise from valvuloseptal tissue deficiencies (Olson and Srivastava, 1996; Eisenberg and Markwald, 1995).

Progress in understanding the molecular regulation of endocardial cushion morphogenesis has been facilitated by use of an in vitro collagen gel assay (Markwald et al., 1981; Runyan and Markwald, 1983). In this regard, the peptide growth factors TGFβ2, signaling through the type III TGFβ receptor, and TGFβ3, through the type II TGFβ receptor, direct nonredundant signaling cascades to initiate EMT in the endocardial cushions of the AVC (Brown et al., 1996, 1999; Boyer et al., 1999; Boyer and Runyan, 2001). TGFβ2 appears to mediate initial cell–cell separation of activated canal endocardium derived from chick embryos, while TGFβ3 is essential for subsequent mesenchymal cell formation and invasion into the underlying matrix. TGFβ3 is detected in mouse endocardial cushions participating in coronary vasculogenesis and cardiac valve remodeling after transformation events (∼E11.5) (Baldwin, 1996; Camenisch et al., 2002). TGFβ3 increases VEGF protein production in a dose-dependent manner (Saadeh et al., 2000). In the current study, this timing of TGFβ3 activity coincided with the production of VEGF, which suggests that TGFβ3 may be one factor upstream of AVC production of VEGF, or, conversely, VEGF may trigger TGFβ3 activity demarcating a shift in the morphogenetic events of EMT. The ability of E10.5 explants to retain the capacity to undergo EMT emphasizes the induction power of VEGF to circumvent the signals that promote mesenchyme formation. We have demonstrated that VEGF mediates post-transformation responses by endocardial cells within the embryonic AVC, and that these events can be induced prematurely by exposure to hypoxic conditions. Further investigations are warranted to determine whether TGFβ3 and VEGF production is functionally linked in the context of remodeling the cardiac cushions into heart valves.


We thank Dr. N. Ferrara (Genentech) for the generous gift of the sFlt-Ig chimeric protein. We appreciate the critical review of the manuscript by Dr. J. Schroeder, and the assistance of Ms. Sharon Fleck with the manuscript preparation.