Abstract
Cardiac hypertrophy, the enlargement of myocytes in the heart, develops in response to physiological (e.g. exercise and pregnancy) or pathological (e.g. myocardial infarction and hypertension) stresses, increasing risk of heart failure and malignant arrhythmia. The cardiac hypertrophy response is managed by a dense web of signaling pathways, with many molecular species influencing cardiac myocyte growth. Little is known about the specific signaling pathways that distinguish pathological and physiological forms of hypertrophy. The complexity of this network has hindered the development of successful therapeutic strategies and indicates the need for integrative systems approaches which can provide a global view of functional relationships in the network. The overall goal of this dissertation is to integrate experimental and computational approaches to determine how the components and network topology of hypertrophy signaling lead to differential regulation of myocyte shape and gene expression.
To understand network organization in a complex process like hypertrophy, new large scale experimental approaches are required to quantitatively characterize a large number of input and output relationships at multiple time points. To address this challenge, we developed an automated image acquisition method that records 5 × 5 mosaic images of fluorescent protein-labeled cardiac myocytes within each well of a 96-well plate using an automated stage and focus. Post-processing algorithms automatically identify cell edges, quantify cell phenotypes, and track cells. We uniquely applied our imaging platform to study hypertrophy reversibility in a scalable cell model. Cell area changes after washout of a dose response to the α-adrenergic receptor (αAR) agonist phenylephrine (PE) showed that hypertrophy reverses at low but not high levels of α-adrenergic signaling: a reversibility delay. Perturbations with specialized αAR antagonists, a mathematical model, and live imaging of αAR localization identify the mechanism for this reversibility delay: ligand trapping with internalized PE acting on intracellular αAR's.
While many proteins and genes have been identified that affect hypertrophy, it is unclear how these parts work together as a coordinated system. To address this challenge, we developed a computational model of the cardiac myocyte hypertrophy signaling network to determine how the components and network topology lead to differential regulation of transcription factors, gene expression, and myocyte size. Our computational model of the hypertrophy signaling network contains 106 species and 193 reactions, integrating 14 established pathways regulating cardiac myocyte growth. 109 of 114 model predictions were validated using published experimental data testing the effects of receptor activation on transcription factors and myocyte phenotypic outputs. Network motif analysis revealed an enrichment of bifan and biparallel cross-talk motifs. Sensitivity analysis was used to inform clustering of the network into modules and to identify species with the greatest effects on cell growth. Many species influenced hypertrophy, but only a few nodes had large positive or negative influences. Ras, a network hub, had the greatest effect on cell area and influenced more species than any other protein in the network. We validated this model prediction in cultured cardiac myocytes. With this integrative computational model, we identified the most influential species in the cardiac hypertrophy signaling network and demonstrate how different levels of network organization affect myocyte size, transcription factors, and gene expression.
Finally, while different presentations of hypertrophy are seen in vivo (ex: physiological vs. pathological, eccentric vs. concentric), it is unclear how such a cross-talk dense network could manage these distinct responses. Moreover, little differential regulation is seen among hypertrophy agonists in commonly measured hypertrophy features such as cell size and fetal gene expression. We hypothesized that increasing hypertrophy measurements to include more shape features (ex: elongation and form factor) and expression of other genes relevant in cardiac remodeling (Ex: cell death, fibrosis, proliferation, and inflammation) would allow us to observe more diverse responses among the hypertrophic agonists.
To test this hypothesis, we stimulated cardiac myocytes with 15 hypertrophic agonists and quantitatively characterized differential regulation of 5 shape features using high-throughput microscopy and transcript levels of 12 genes using qPCR. Transcripts measured were associated with phenotypes including fibrosis, cell death, contractility, proliferation, angiogenesis, inflammation, and the fetal cardiac gene program. While hypertrophy pathways are highly connected, the agonist screen revealed distinct hypertrophy phenotypic signatures for the 15 receptor agonists. We then used k-means clustering of inputs and outputs to identify a network map linking input modules to output modules. Five modules were identified within inputs and outputs with many maladaptive outputs grouping together in one module: Bax, C/EBPβ, Serca2a, TNFα, and CTGF. Subsequently, we identified mechanisms underlying two correlations revealed in the agonist screen: correlation between regulators of fibrosis and cell death signaling (CTGF and Bax mRNA); and myocyte proliferation (CITED4 mRNA) and elongation. Follow-up experiments revealed positive regulation of Bax mRNA level by CTGF and an incoherent feed forward loop linking Nrg1, CITED4 and elongation. With this agonist screen, we identified the most influential inputs in the cardiac hypertrophy signaling network for a variety of features related to pathological and protective hypertrophy signaling and shared regulation among cardiac myocyte phenotypes.
Together, this body of work identified influential network hubs and shared regulation of maladaptive and adaptive hypertrophy features. While this systems approach revealed insights into network organization, it also allowed us to prioritize experiments to reveal new mechanistic insights into hypertrophy, such as the discovery of Ras as the most influential species, CTGF regulation of Bax, and CITED4 regulation of myocyte elongation. The quantitative network understanding gained in this work will be helpful in planning therapeutic interventions for heart failure that enhance adaptive responses and suppresses maladaptive responses.
