Association of a functional variant downstream of TNFAIP3 with systemic lupus erythematosus

Abstract

Systemic lupus erythematosus (SLE, MIM152700) is an autoimmune disease characterized by self-reactive antibodies resulting in systemic inflammation and organ failure. TNFAIP3, encoding the ubiquitin-modifying enzyme A20, is an established susceptibility locus for SLE. By fine mapping and genomic re-sequencing in ethnically diverse populations, we fully characterized the TNFAIP3 risk haplotype and identified a TT>A polymorphic dinucleotide (deletion T followed by a T to A transversion) associated with SLE in subjects of European (P = 1.58 × 10(-8), odds ratio = 1.70) and Korean (P = 8.33 × 10(-10), odds ratio = 2.54) ancestry. This variant, located in a region of high conservation and regulatory potential, bound a nuclear protein complex composed of NF-κB subunits with reduced avidity. Further, compared with the non-risk haplotype, the haplotype carrying this variant resulted in reduced TNFAIP3 mRNA and A20 protein expression. These results establish this TT>A variant as the most likely functional polymorphism responsible for the association between TNFAIP3 and SLE.

Figure 1

SNPs in and around the TNFAIP3 gene associated with SLE in European-ancestry (a.), Asian (b.) and Korean (c.) populations. Genotyped SNPs are depicted with blue diamonds and imputed SNPs are shown with red circles. An orange solid line represents recombination rates across the region. The dashed line represents a Bonferroni corrected P < 1 × 10⁻⁴. Arrows identify SNPs demonstrating the most significant association results in each population.

Figure 2

Haplotype and conditional association analysis results of the TNFAIP3 risk haplotype. Haplotypes present at a frequency > 1% were compared in the European-ancestry and Korean populations (a.). Alleles in white boxes represent the major allele and those in grey boxes represent the minor allele for each haplotype. Black bold rectangles identify minor alleles that differentiate the SLE risk haplotype from the non-risk haplotype. Conditional association analysis was performed in the European (b.) and Korean (c.) populations for each of the SNPs within the 48.5 kb segment bounded by rs5029937 and rs61117627. We assessed three models: first, conditioning the on F127C coding variant rs2230926 (white bars), then conditioning on the TT>A variant (gray bars) and finally conditioning on rs7749323 (black bars).

Figure 3

Functional characterization of the TT>A polymorphic dinucleotide and TNFAIP3 associated risk haplotype. (a.) Shown is a representative EMSA result from six independent experiments for THP1 and three for U937). The first two lanes show free probe for wild type (WT) and TT>A variant followed by increasing amounts of nuclear protein and labeled probes as indicated. A non-specific band is labeled N.S. (b.) Super shift was performed using antibodies specific to NF-κB subunits.. Complexes formed in the presence and absence of antibodies are identified by arrows on the left of the figure.. Densitometric quantification of nuclear protein binding in independent experiments was performed for THP1 cells (c.) and U937 cells (d.) using optimal concentrations of nuclear extract. Expression of TNFAIP3 transcripts were evaluated from CEU, CHB and JPT populations (AA, N=2; AG, N=24; GG, N=115) (e.) and compared to the YRI population (AG, N=6; GG, N=54) (f.) using a one-way ANOVA and unpaired t-test. A20 protein expression from cell lines of European-ancestry subjects (AA, N=2; AG, N=5; GG, N=5) (g.) were compared to African-American subjects (AG, N=10, GG, N=9) (h.) using one-way ANOVA and unpaired t-test, respectively.