Meta-analysis of Immunochip data of four autoimmune diseases reveals novel single-disease and cross-phenotype associations

Abstract

Background: In recent years, research has consistently proven the occurrence of genetic overlap across autoimmune diseases, which supports the existence of common pathogenic mechanisms in autoimmunity. The objective of this study was to further investigate this shared genetic component.

Methods: For this purpose, we performed a cross-disease meta-analysis of Immunochip data from 37,159 patients diagnosed with a seropositive autoimmune disease (11,489 celiac disease (CeD), 15,523 rheumatoid arthritis (RA), 3477 systemic sclerosis (SSc), and 6670 type 1 diabetes (T1D)) and 22,308 healthy controls of European origin using the R package ASSET.

Results: We identified 38 risk variants shared by at least two of the conditions analyzed, five of which represent new pleiotropic loci in autoimmunity. We also identified six novel genome-wide associations for the diseases studied. Cell-specific functional annotations and biological pathway enrichment analyses suggested that pleiotropic variants may act by deregulating gene expression in different subsets of T cells, especially Th17 and regulatory T cells. Finally, drug repositioning analysis evidenced several drugs that could represent promising candidates for CeD, RA, SSc, and T1D treatment.

Conclusions: In this study, we have been able to advance in the knowledge of the genetic overlap existing in autoimmunity, thus shedding light on common molecular mechanisms of disease and suggesting novel drug targets that could be explored for the treatment of the autoimmune diseases studied.

Keywords: Autoimmune disease, functional enrichment analysis; Celiac disease; Cross-disease meta-analysis, Immunochip; Rheumatoid arthritis; Systemic sclerosis; Type 1 diabetes.

Fig. 1

Novel genome-wide associated loci for celiac disease, rheumatoid arthritis, systemic sclerosis and type 1 diabetes. Pleiotropic SNPs reaching genome-wide significance level and SNPs associated with a single disease and reaching p values lower than 5 × 10^− 6 in the subset-based meta-analysis were checked for genome-wide association in each of the diseases included in the best subset. Negative log₁₀-tranformed p value (disease-specific p values) (upper plot) and odds ratio (lower plot) for the new genome-wide signals are shown. The six loci are annotated with the candidate gene symbol. Circles represent the analyzed diseases (red: celiac disease; yellow: rheumatoid arthritis; green: systemic sclerosis; blue: type 1 diabetes). The red line represents genome-wide level of significance (p = 5 × 10^− 8)

Fig. 2

Functional annotation of 38 pleiotropic polymorphisms (p < 5 × 10^–8 in the subset-based meta-analysis) and four single-disease associated variants (p < 5 × 10^–6 in the subset-based meta-analysis and p < 5 × 10^–8 in disease-specific meta-analyses). Haploreg v4.1 was used to explore whether lead SNPs, and their proxies (r² ≥ 0.8), overlapped with different regulatory datasets from the Roadmap Epigenomics project, the ENCODE Consortium and more than ten eQTL studies in immune cell lines, cell types relevant for each specific disorder and/or whole blood. Colors denote both lead and proxy SNPs overlapping with the different regulatory elements analyzed: G (red): conserved positions (Genomic Evolutionary Rate Profiling, GERP); P (orange): promoter histone marks; E (yellow): enhancer histone marks; D (green): DNase I hypersensitive sites (DHS); T (blue): transcription factor binding sites (TFBSs); eQ (purple): expression quantitative trait loci (eQTL). Functional annotations overlapping with proxy SNPs are marked with an asterisk. N proxy, number of proxy SNPs for each lead variant. The different loci are annotated with the candidate gene symbol

Fig. 3

Functional regulatory elements and PPI enrichment analysis. a Heat map showing DNase 1 hypersensitive sites (DHSs) and histone marks enrichment analysis of the set of pleiotropic variants. GenomeRunner web server was used to determine whether the set of pleiotropic SNPs significantly co-localize with regulatory genome annotation data in 127 cell types from the Roadmap Epigenomics project. First column shows cell types grouped and colored by tissue type (color-coded as indicated in the legend). Tissues relevant for the autoimmune diseases studied as well as other tissues for which any of the analyzed functional annotations showed a significant enrichment p value (p < 0.05 after FDR correction) are shown. The remaining four columns denote the analyzed functional annotations, DHSs, H3K27ac, H3K4me1, and H3K4me3. Results of the enrichment analysis are represented in a scale-based color gradient depending on the p value. Blue indicates enrichment and white indicates no statistical significance after FDR adjustment. b Interaction network formed for the set of common genes. Direct and indirect interactions among genes shared by different disease subgroups were assessed using STRING. Plot shows results of the “molecular action” view such that each line shape indicates the predicted mode of action (see legend). Genes involved in the biological pathways enriched among the set of pleiotropic loci (Additional file 2: Table S10) are shown in color: red: Th1 and Th2 cell differentiation; green: Th17 cell differentiation; yellow: Jak-STAT signaling pathway; blue: T cell receptor signaling pathway