Constructing Causal Cytokine Networks for Autoimmune and Inflammatory Diseases Target Identification

Abstract

Cytokines are cell signaling proteins that act as inducers of physiological responses including the activation and differentiation of innate and adaptive immune cell types. In some cases, this mediates the initiation or progression of inflammatory and auto-immune diseases such as septic shock, rheumatoid arthritis, psoriasis, and inflammatory bowel disease. Many cytokines are pleiotropic, meaning that they can have effects on multiple cell types and play several distinct functions. On the other hand, multiple cytokines might have similar function and be redundant or degenerate. Some cytokines also regulate the expression and functions of other cytokines. Such complex cause-effect relationships can be modeled computationally by building "cytokine networks". Those networks describe intricate interactions between different cytokines in their associated tissue environment and cell types. Here we propose a computational approach to find sensitive and "causal" nodes within a cytokine interaction network by identifying whether persistent positive feedback (self-loop) exists on that cytokine in its mRNA transcription network. Based upon two microarray gene expression datasets from immune stimulation experiments using up to 15 different cytokines, we identified TNF as a "causal" cytokine in the colonic epithelial cell line HCT8, and IL-23, IL-1{beta} and TGF{beta} as "causal" cytokines in primary myofibroblasts. We hypothesize that a cytokine with a persistent positive feedback loop that amplifies its transcription and secretion could be identified as "causal" for the transcriptional regulation of other cytokines and potentially, downstream disease phenotypes. The existence of such positive feedback loop may determine whether targeting suppression of such cytokine will result in clinical efficacy.