Applying a proteomic approach to embryonic vascular tissue generated from an ex vivo model of vasculopathy, we have identified proteins and pathways that are disrupted during cardiovascular development and lead to the observed pathological phenotype. Dysregulation of this set of proteins in a “normal” fetus may signal that the fetus is congenitally at risk or encountering an environmental stressor. Several of these protein targets (WNT16, Pcsk1 and ST14) provided the basis for a novel prenatal screen for the detection of CHDs, suggesting dysregulation of WNT16, ST14 and Pcsk1 may play a role in the etiology of human CHDs. Several forms of CHDs are known to have a genetic basis such as microdeletions (22q11) in DiGeorge Syndrome and single gene mutations (JAG1) in Alagille Syndrome. It would be of interest to determine if individuals with CHDs carry genetic mutations in WNT16, ST14 or Pcsk1. At the protein level, these markers are overexpressed therefore if sequence variants are found mutations would likely affect protein function and/or activity. In the case of ST14 and Pcsk1, both of which are enzymes, a hypomorphic or hypermorphic mutation in these genes has the potential to affect multiple substrates and hence several cellular processes.
Changes in cellular processes not only reflect the molecular origins of disease but also provide avenues for molecular therapies to ameliorate the defect. To this end, we identified several proteins, previously unknown to have a role in cardiovascular development, with diverse functions that cluster to adhesion/migration, differentiation, transport, IGF signaling pathways (). Although we discovered and characterized several proteins with currently unknown roles in cardiovascular development, we were able to place them in their relevant biological context by relating them to existing functional cascades (signaling, enzymatic, polarity, migration, and differentiation). Protein networking analysis revealed that several of the identified proteins interact upstream or downstream of each other and with established factors in cardiovascular development (). The presence of proteins known to have roles in cardiac development within our dataset further validates our approach. For example, our screen detected the transcriptional repressor Jumonji whose deletion is known to result in myocardial defects and ventricular septal defects 
. However our data demonstrating decreased Jumonji during vasculopathy extends the role of Jumonji to vascular development.
Integration of Proteins Identified by Proteomic Profiling with Known Pathways and Biological Processes Involved in Cardiovascular Development.
Previously, we demonstrated that yolk sac vasculopathy is associated with increased nitric oxide, however the nitric oxide targets remained elusive 
. Recently nitrosylation sites on the cytoplasmic domain of TRPC5 were identified and in response to nitric oxide, Ca(2+) entered bovine aortic endothelial cells 
. In this study we observed increased TRPC5, raising the possibility that the observed yolk sac vasculopathy is mediated by nitric oxide driven over-activation of TRPC5. We speculate that the subsequent abnormal influx of calcium through TRPC5 activates the calcineurin/NFAT pathway 
. In the developing heart, the NFAT pathway is tightly regulated and has complex interactions with DSCR (another protein identified in our screen) and VEGF to drive differentiation; low levels of VEGF are required for EMT while high levels terminate it 
. Therefore, it is tempting to speculate that TRPC5 is a novel player in cardiac development, acting upstream of NFAT, leading to dysregulation of VEGF expression and ultimately termination of heart development.
On the other hand, positive regulation of cardiac EMT relies on Wnt/β-catenin signaling however the specific Wnts involved are largely unknown with the exception of Wnt9a in the avian heart 
. Variations in Wnt signaling are expected to have direct effects on valve plasticity as canonical Wnt/β-catenin signaling induces mesenchymal differentiation and modulates the amount of mesenchyme produced 
. One defect associated with vasculopathy is increased endothelial cell number per vessel, leading to a hypertrophic, nonfunctional capillary plexus. Thus the observed upregulation of WNT16 may lead to unregulated endothelial cell differentiation or proliferation, ultimately affecting the organization of vascular structures. Concomitantly, antibody neutralization of WNT16 blunts cardiac EMT in the atrioventricular canal assay. Though the WNT16 isoform is known to be expressed in the embryonic heart 
, a functional role for WNT16 has not been previously reported, thus our atrioventricular canal assay results describes a new function for a novel WNT member in the cardiac developmental cascade. Additionally, we observed increased WNT16 in human amniotic fluid (AF) of women carrying fetuses with CHDs and demonstrated that this is predictive of CHDs. Monitoring WNT16 levels and targeting WNT16 signaling may provide novel avenues for future diagnostic and pharmacological research.
Proteoglycans and extracellular matrix are important players in cell adhesion, migration, proliferation, and angiogenesis, processes important in both development and disease. CHST3 catalyzes the transfer of sulfate to position 6 of the N-acetylgalactosamine (GalNAc) residue of chondroitin. Human endothelial cells express CHST3 mRNA while endothelial cells overlying the porcine mitral valve produce chondroitin sulphate 
. Interestingly, gain of function experiments with the heart development regulatory gene Tbx20
in the avian endocardial cushion resulted in decreased chondroitin sulfate proteoglycans and increased mesenchyme 
. Our data demonstrating CHST3 expression in the murine endocardial cushion suggests that this may be a conserved mechanism to regulate mesenchyme production, although further studies are required to confirm this hypothesis. Though not predictive of CHDs, CHST3 was dysregulated in the AF of woman carrying fetuses with CHDs, suggesting a role in cardiac development.
An additional factor involved in adhesion/migration identified in our study is the protease ADAM15. Multiple individual ADAM isoform KO mice display cardiac defects, including Adam 9, 17, and 19 
. ADAM15 is expressed in blood vessels and endocardium, and although ADAM15 mice are viable studies using retinopathy and tumor angiogenesis models demonstrate an inducible vascular phenotype (failure of remodeling of the primary capillary plexus) 
. In conjunction, our data indicates that decreased ADAM15 enzymatic activity in the hyperglycemic yolk sacs disrupts growth factor/cytokine signaling of a yet unknown pathway, resulting in an inability of the endothelial cells to initiate remodeling, consistent with the previous report on an inducible vascular phenotype. Our data from the atrioventricular canal assay extends the known functions of ADAM15 to include cardiac EMT. Speculating on the substrates involved in its effects on differentiation to mesenchymal cells, ADAM15 proteolytic activity on cadherins is one plausible target 
. Further, the strong immunofluorescent staining of ADAM15 in the cellular projections of cardiac mesenchymal cells suggests ADAM15 may be necessary for cell interaction with the matrix potentially via integrin binding. These findings warrant future studies to determine if ADAM15 is predicative of CHDs. Interestingly, ADAM12 is a known biomarker for Down syndrome, demonstrating the applicability of this family to biochemical assays for diagnosis 
. The novel finding of a functional role for ADAM15 in the embryonic heart combined with its strong expression in the cardiovascular system make it an appealing target for future studies that isolate its downstream targets in the heart.
In addition to proteins involved in differentiation and adhesion/migration, several proteins involved in the IGF/insulin pathway were identified in our study. Early in embryogenesis the embryo does not produce insulin nor does maternal insulin cross the placental barrier. However, embryos express high levels of IGF receptors and early growth is mediated via IGF signaling through Akt. Factors that regulate IGF receptor engagement and its downstream signaling cascades were identified both on the extracellular face, including the proteases Pcsk1 and ST14, and the intracellular face, including ZFP106 (or son of insulin receptor mutant) and HES6. Pcsk1 is a secreted protease that cleaves precursors including neuropeptides, renin, somatostatin, and insulin/IGF 
. Additionally substrates include adhesion molecules, proteases and receptors. Pcsk1 KO mice are viable, but growth retarded. Two strains of proprotein convertase KO mice, Furin and PACE4, exhibit defects in the vasculature and heart, while the other 8 proprotein convertase KO mice exhibit defects in hormone production, fertility, and lipid metabolism 
. Our data demonstrating decreased Pcsk1 levels in the yolk sac after hyperglycemia insult suggests that Pcsk1 KO mice may have an unappreciated inducible vascular phenotype. We also demonstrated that Pcsk1 is predictive for CHDs, expressed in the embryonic myocardium, and increased in the AF of fetuses with CHDs. Pcsk1 downregulation in the yolk sac and upregulation in AF may reflect differences in developmental age, cell differentiation processes or clearance/accumulation of proteins in the AF versus tissue. Although proprotein convertases have been linked to Alzheimer's disease, tumorigenesis, and infections, they have not been previously reported to have a role in CHDs. Further given that gene knock out experiments have determined that individual PCs do not seem to be redundant in embryonic development, understanding the role of Pcsk1 in metabolism, growth, and development of cardiovascular system is an intriguing avenue of future investigation.
Another molecule in the IGF signaling pathway, ST14, a transmembrane serine protease, has been implicated in metastatic cancer, which utilizes a molecular program reminiscent of embryonic EMT 
. Substrates for ST14 include hepatocyte growth factor, urokinase-type plasminogen activator, protease-activated receptor 2 (PAR-2), IGF binding protein-related protein-1, trask, laminin and fibronectin 
. Recently, VEGFR-2 was reported to be a putative substrate of ST14 in HUVECs, resulting in ectodomain cleavage and inactivation of cell signaling 
. Here, we demonstrated for the first time that ST14 is present in the embryonic cardiovascular system and is predictive of CHDs. Given the known substrates of this protease, ST14 induction during vasculopathy and in AF of fetuses with CHDs may affect differentiation, adhesion/migration and growth.
Cell movements, growth and differentiation depend on environmental homeostasis; of particular importance during development is redox balance. According to current knowledge, SVCT transports ascorbate into cells and thereby participates in metabolic processes, matrix synthesis and antioxidant defense 
. Interestingly, dehydroascorbate utilizes glucose transporters to enter the cell prior to conversion to ascorbate, and during a hyperglycemic episode glucose competitively inhibits the transport of dehydroascorbate, potentially compromising antioxidant defense and leading to reactive oxygen species induced damage 
. Consistent with the theory of ROS induced vasculopathy; our data demonstrating upregulation of SVCT may signify that the embryo is “at risk” and employing a defense to combat oxidative stress. Although SVCT was not predictive of CHDs, the fetal cardiovascular system is sensitive to oxidative stress and benefits from antioxidant supplements, such as folic acid 
. Therefore understanding the upstream regulation, downstream targets, and/or activity of SVCT may shed light onto novel antioxidant therapies for fetuses “at risk”.
Molecular-based diagnostic tests that detect fetuses “at risk” provide clinicians with information to guide clinical interventions aimed at improving outcome. First-trimester biochemical screening is becoming a powerful tool, allowing clinicians to efficiently triage patients 
. For example, increased pregnancy-associated plasma protein A (PAPP-A) and decreased free β-human chorionic gonatrophin (β-hCG) in maternal serum indicate increased risk of trisomy 21, while reduced PAPP-A in maternal serum is associated with an increased risk of trisomy 18 
. An obvious advantage of serum screening is that it is non-invasive and serum screening may be performed earlier than amniocentesis. In the present study, we utilized AF rather than maternal serum because of its low complexity and similarity to the fetus due to its origin. Cells in AF are fetal rather than maternal. Future studies involving early maternal serum samples are required to determine the merit of adding our biomarkers to existing screening panels. To this end, using an in vivo
streptozocin-induced diabetic model in mice, we detected differences in specific Laminin chains, ADAM15 ectodomain, and MMP2 activity in maternal sera from mice carrying a high proportion of fetuses with heart defects compared to those with a high proportion of fetuses with normal hearts (unpublished data). These data suggest that protein fragments are released into the maternal circulation in sufficient quantities that allow detection early in pregnancy (E10.5 which is approximately equivalent to human developmental week 6). Experiments are currently underway to evaluate the levels of WNT16, ST14, and Pcsk1 in maternal sera from murine models and humans.
The creation of new biochemical diagnostic tools to assay proteins such as Pcsk1, ST14, and WNT16 may give insights into the well-being of the fetus by sensing an inability to combat an environmental insult, disruption in growth, or damage due to reactive oxygen species. Affordable, rapid, and accurate molecular-based diagnosis of potential problems in fetal cardiovascular development early in pregnancy will provide clinicians with information to guide clinical interventions and pave the way toward eradicating one of the major global causes of infant mortality in the world. In the future, dietary supplementation or molecular therapies based on Pcsk1, ST14 and WNT16 or their targets may ameliorate the deficiencies that ultimately lead to the formation of a CHD in the first trimester. Therefore isolating the specific substrate targets of ST14 and Pcsk1 that disrupt cardiac development and investigating the transcriptional or post-transcriptional regulatory mechanisms that lead to overexpression of ST14 and Pcsk1 will aid in deciphering the role of these proteins in the etiology of CHDs. Further, continued mining of the proteomic dataset coupled with analysis of novel factors in physiologically relevant models will continue to yield fruitful avenues for investigation into the biology of cardiovascular development and pathogenic mechanisms of CHDs in the hope of advancing prevention, diagnosis and treatment.