Abstract
BACKGROUND: Systemic lupus erythematosus (SLE) is a polygenic autoimmune disease characterized by the production of autoantibodies leading to widespread inflammation, whereas primary immunodeficiency (PID) disease is a group of specific genetic defects in molecular pathways required for host defense to infectious agents. PID genes are found to play critical roles in development and function of the immune system, and decreased function results in dramatically increased susceptibility to infection. We hypothesized that PID genes would be genetically over-represented and over-expressed in SLE, facilitating the identification of novel risk genes and molecular pathways involved in lupus pathogenesis. METHODS: A comprehensive database of 453 PID genes was developed. PID genes were clustered into protein-protein interaction (PPI) networks and compared to known SLE risk loci. Over-expression of PID genes was examined in datasets from multiple sample types. Machine learning (ML) classifiers using PID genes as input were employed to predict SLE disease status and severity. RESULTS: PPI clustering of PID genes revealed 18 distinct cellular and functional groups. PID genes overlapped with SLE risk loci more than expected by random chance and in greater magnitude than PID gene overlap with risk alleles of other autoimmune conditions. In both whole blood and immune cell-specific samples, PID genes were significantly differentially expressed (DE) more than expected by random chance, with most of these genes being over-expressed in SLE. Specific patterns of DE genes, according to the cellular and functional groups, were mapped to expression by specific immune cell subsets, with many PID genes upregulated in T cells, classical monocytes, and non-classical monocytes. Over-expression of and enrichment of PID genes and functional clusters were also positively associated with increased measures of lupus disease activity. Finally, functionally different groupings of PID genes were useful in classifying lupus from healthy controls and active lupus from inactive lupus by ML with accuracies of 0.80 and 0.74, respectively. CONCLUSIONS: PID genes are not only genetically associated with SLE but are also broadly over-expressed, particularly in active disease. These results highlight the profound link between pathways governing host defense and SLE immunopathogenesis, and PID genes may be considered further for therapeutic targeting.