Learning How Immune Cells of the Gut Contribute to Autoimmune Disease through Reverse-Engineered Transcriptional Regulatory Networks

The National Institute of Health estimates that at least 23.5 million Americans suffer from autoimmune disease, yet even the most prevalent autoimmune diseases have no cure, requiring costly lifelong immuno-suppressive therapies. Genome-wide association studies (GWAS) provide lists of genes associated with autoimmune diseases, and many of these genes are specifically expressed by diverse immune cell populations, including arms of both the adaptive and innate immune system. Many of these genes are involved in immune-cell sensing and response to the microbiota. As the gut contains the highest concentration of host-associated microbiota, we reasoned that characterization of gene expression by RNA-seq in gut-associated immune cell populations would implicate specific populations in disease etiology. In this study, we focus on innate lymphoid cells (ILCs) under homeostatic conditions and in response to microbial perturbations in mice. Furthermore, we wanted to design perturbations (e.g., gene knock down) for specific populations to change phenotypes that might contribute to autoimmunity. Thus, we not only set out to characterize transcriptional responses but also to infer mechanisms of transcriptional control, to describe gene expression as multivariate functions of transcription factor (TF) activities. TF gene expression measurements are often a poor proxy for protein TF activities, so we integrated chromatin accessibility (ATAC-seq) measurements to get noisy sets of signed, TF-gene interactions and then use penalized matrix decomposition methods to estimate TF activities and reverse-engineer transcriptional regulatory networks. We connected specific TFs (on the basis of predicted targets), in specific ILC populations, with autoimmune and other diseases. Predicted TF-gene regulatory hypotheses are currently being tested experimentally.