Constructing Tissue-Tissue Networks to Highlight Novel Causal
Patterns of Association with Core Disease Processes
Eric Schadt
Rosetta Inpharmatics/Merck Research Labs
Seattle
Friday, April 25, 2008, 12:30–1:30 pm
GEMS classroom, 3rd Floor in
Shriner's Building
Coffee, tea, and cookies will be provided
Abstract
We have previously detailed an alternative to the classic forward genetics approach for dissecting complex disease traits where, instead of identifying susceptibility genes directly affected by variations in DNA, gene networks that are perturbed by susceptibility loci and that in turn lead to disease are identified. This approach has been applied to individual tissues in human and mouse populations, leading to the identification of highly interconnected subnetworks predicted and experimentally validated as causal for disease. However, this and other network approaches to understanding complex system behaviors have largely ignored tissue-to-tissue communications that are critical to living systems manifesting complex behaviors. For example, the central nervous system (CNS) receives information regarding the status of peripheral metabolic processes via hormonal signaling, direct macromolecular sensing, and through a complex neuronal network that connects the CNS with the periphery. At the center of these CNS networks is the hypothalamus, which serves as the target for a plethora of signals such as insulin, leptin, and a diverse set of macromolecules including glucose and long chain fatty acids. These signals in turn serve to modulate the hypothalamic response through the autonomic neuronal pathways, where disruption of these pathways connecting the periphery and hypothalamus partially explains obesity. Beyond these known interactions between tissues are a number of unknown interactions that have the potential to define much of the complex behavior that emerges from living systems.
To decipher the communication between tissues at the molecular level, we examine interactions among gene expression traits in blood, adipose, muscle, pancreas, liver, and brain tissues from human and experimental mouse cross populations using integrative genomics approaches previously applied to single tissues. Gene-gene relationships specific to interactions between two given tissues are observed to give rise to coherent subnetworks involved in important functions like circadian rhythm and energy balance that are independent of subnetworks detected from single tissue analyses, highlighting novel networks associated with disease that have previously escaped notice. Further, not only do the tissue-tissue networks highlight genes in one tissue that respond to changes in genes in a second tissue, but they elucidate entire subnetworks in one tissue that influence subnetworks in a second tissue. Our modeling approach provides direct support for cross-tissue processes influencing a diversity of disease traits related to obesity, diabetes, atheroscelrosis, and Alzheimer's, and suggests hypotheses on how biological processes observed in one tissue as driving a given disease (e.g., obesity) may influence processes observed in a different tissue as driving a related disease (e.g., diabetes). Many of the specific causal relationships we detect via the tissue-tissue networks would be difficult or impossible to detect via single-tissue analyses. Our analyses provide further support that complex traits like obesity, diabetes, and atheroscelrosis are emergent properties of complex interactions among molecular networks in different tissues that are modulated by complex genetic loci and environmental factors.