Expression signatures with specificity for type I and II IFN response and relevance for autoimmune diseases and cancer.

backgroundAberrant interferon signaling is a key element of various diseases, but resolving gene expression signatures that stem from different types of IFNs in tissue samples is still a challenge. Most published IFN signatures comprise genes that are activated by different IFNs: they cannot discriminate type-I (IFN-I) and type-II (IFN-II) IFN stimulation. Most often such signatures were obtained from a single expression dataset that had been obtained in a specific cellular context, and their translatability to other experimental contexts has not been demonstrated.

resultsWe leveraged multiple RNA-seq datasets of IFN stimulation in a network meta-analysis workflow to obtain IFN gene signatures separating IFN-I and IFN-II. We validated our signatures in bulk and single cell RNA-seq datasets of various cellular contexts demonstrating similar or higher coherence than previously published signatures. Our IFN-II signature is broader applicable than other published signatures as it demonstrates strong performance in detecting IFN-II response in more cell types. In three SLE microarray datasets our IFN-I signature was highly coherent and correlated with disease severity better than most published signatures. In TCGA, our IFN-II signature produced distinct profiles compared to published IFN-I signatures and correlated strongly with published CD8

conclusionsOur IFN-I and IFN-II response-specific gene expression signatures can inform on complex IFN responses in a more fine-grained way than previously possible. They can be used to assess type I versus II IFN response in gene expression data produced by different technologies, for different diseases and even different cell types in single cell studies. The association of high scores of our IFN-II signature with anti-tumor response to ICIs suggests a role as a biomarker to predict ICI response.