Purpose Systemic lupus erythematosus (SLE) is an autoimmune, chronic and multisystemic autoimmune disorder, which typically affects women and presents high severity and mortality. It is characterized by the loss of self-tolerance, and both innate and adaptive immune systems contribute to its progression. SLE is considered a multifactorial disease as genetic, hormonal and environmental factors have been shown to contribute to its pathogenesis.

Mouse models have helped to understand the molecular mechanisms behind SLE and are commonly used to test new pharmaceutical compounds. However, the phenotype and molecular heterogeneity of SLE in humans limits its mimicking in animal models. Thus, we hypothesized that different mouse models might mimic better than others specific subgroups of SLE patients at the molecular level. The aim of this study is to characterize and compare different spontaneous mouse models at the molecular level and compare them with subgroups of human SLE patients.

Methods Four mouse models (MRL/lpr, NZB/W, BXSB.Yaa, and TLR7.Tg) were analyzed along four time points, including their respective genetic controls. The transcriptome (RNA-Seq) and cell proportions (flow cytometry) of the spleen, as well as numerous cytokines and autoantibodies in the serum were profiled. In addition, the molecular landscape of the mouse models was compared to SLE patients from the PRECISESADS project cohort.

Results Specific and shared molecular pathways were identified across the four mouse models. For example, upregulation of TNF signaling was shared by all models. On the other hand, BXSB.Yaa and TLR7.Tg shared the upregulation of leukocyte migration related genes, while the interferon-gamma response was only observed in the BXSB.Yaa and MRL/lpr mouse models. In addition, shared molecular pathways were dysregulated at different time points during the development of the phenotype. And as hypothesized, specific subgroups of SLE patients showed much closer molecular similarities to specific mouse models than others, thus allowing to stratify SLE patients based on the mouse models’ dysregulated pathways.

Conclusion The results from this study constitute a molecular benchmark for future SLE mouse model studies. Relevant information is given about the choice of the model (out of the four included in the study), regarding the molecular pathway to be studied, and also about the time point when samples should be collected. In addition, we showed that different animal models at different stages present higher functional similarities with some subgroups of SLE patients than with others, which might improve the future translation of animal model results to the human disease.