Single-cell Network Inference as a new approach to better characterize autoinflammatory diseases
It is generally thought that an inflammatory response is beneficial for the host to clear infections and repair damaged tissues. However, if uncontrolled, it can become detrimental, harmful, and even life threatening when leading to organ failure. Non-infectious fever and systemic and/or disease specific organ inflammation characterize autoinflammatory diseases. Type I interferonopathies represent discrete examples of disturbances of the homeostatic control of this system, caused by Mendelian mutations. An increasing number of mutations in genes involved in the sensing and clearance of nucleic acids (SAMHD1, TREX1, MDA5 (IFIH1), …) have been characterized and linked to an excessive type I IFN response in Aicardi-Goutières syndrome (AGS), resulting in neurological disorders and skin lesions. Other gain-of-function mutations leading to constitutive activation of cytosolic interferon signaling pathways/increased sensitivity to cytosolic nucleic acid ligands have been detected such as in STING (TMEM173), causing a STING-associated vasculopathy with onset in infancy (SAVI). Major symptoms include early-onset systemic inflammation with fever, severe skin vasculopathy, interstitial lung disease, and some lupic features. Systemic lupus erythematosus (SLE) is a more complex multisystemic autoimmune disorder (autoantibodies against DNA and ribonucleoproteins) thought to involve both genetics and environmental factors, with a postulated predominant role of type I IFN in the pathogenesis.
To investigate in an unbiased way the complexity and the heterogeneity among inflammatory responses in general and interferonopathies in particular, our ATIP-avenir/INSERM group at the Imagine Institute in Paris, propose a new and unbiased approach, combining two emerging and exciting fields (machine learning algorithm and “big Data” generation at the single-cell level), to better understand the links between mutations in genes and clinical symptoms. Our first objective will be to better characterize and stratify patients in more homogeneous groups, by performing state of the art single-cell transcriptomic analysis and clustering of peripheral blood mononuclear cells (PBMC) from different patients suffering from interferonopathies. Then based on artificial intelligence we are aiming towards inferring networks of genes interactions to get a deeper and unbiased understanding of the diversity of the molecular mechanisms behind inflammation. Once established, this combined approach of single-cell coupled with Network inference, could have a tremendous impact on patients at various different levels (better screening, stratification, diagnosis of patients and discovery of new therapeutics) and will be moving the medical research field toward a more personalized medicine.
We already used the power of machine learning algorithms (Inferelator) combined with the generation of “Big Data” (Whole transcriptomic (bulk RNA-seq) and chromatin accessibility changes (ATAC-seq) to infer a dynamic computational model that describes transcriptomic pathways involved in type I IFN production in the context of HIV-1 sensing by human dendritic cells (DCs), induced by inhibition of SAMHD1 function. This network will be used as a starting point for reproducing this work in rare genetic diseases linked to autoinflammation.
Inflammatory Responses and transcriptomic Networks in diseases, Institut Imagine UMR 1163
Domain 3 - U932 - Immunity and Cancer