Computational modeling of inflammatory mediators in acute illness: From networks to mechanisms.

摘要

The acute inflammatory response is a complex defense mechanism that has evolved to respond rapidly to injury, infection, and other disruptions in homeostasis. The complex role of inflammation in health and disease has made it difficult to understand comprehensively. With the advent of high throughput technologies and the growth of systems biology, there has been an unprecedented amount of data and -omics analysis aimed at uncovering this complexity. However, there still remains a shortage of translational insights for acute inflammatory diseases from these studies. In this dissertation, we employ a comprehensive systems approach in order to study the coordination of inflammation and identify key control mechanisms, and how these map onto clinical outcomes. This process begins with collection of high-dimensional time course data of inflammatory mediators, followed by data-driven modeling and network inference that finally informs mechanistic computational models for prediction and analysis. In patients with pediatric acute liver failure (PALF), we inferred inflammatory networks and identified key differences between patients that were survivors versus non-survivors when other analyses proved inconclusive. We showed that inflammatory networks can be used both as biomarkers and to generate mechanistic hypotheses for this poorly understood disease. In experimental models of trauma as well as in human trauma patients, we identify a conserved central network motif of cross-regulating chemokines. We develop a logical model based on this hypothesized network, which is able to capture both inflammatory trajectory and clinical outcome differences among patients with differing injury severity. These studies suggest that the hypothesized cross-regulatory interactions among chemokines MIG, IP-10 and MCP-1 represents an important point of control regulating the progression of acute inflammation. We propose that further analysis and validation of this hypothesis will require targeted perturbation studies in cells and animals with iterative rounds of mechanistic model refinement. We explore an example of such a study focused on the anti-inflammatory effects of NAD +, wherein we characterize a signaling pathway that gives rise to a complex dose and time dependent induction of TGF-beta1.