Genomic approaches for the identi cation of risk loci for Rheumatoid Arthritis

Abstract

Rheumatoid Arthritis (RA) is one of the most prevalent autoimmune diseases in the world and is characterized by the chronic in ammation of the synovial joints. The origin of the disease is unknown but it is actually accepted that it is caused by the complex interaction of a genetic susceptibility background and environmental factors. To date, the characterization of the genetic architecture of RA is far from complete. In the present work we will use the power of two distinct genomic approaches to identify new candidate genes for the susceptibility to RA. In the rst genomic approach, we have used gene expression microarrays to characterize the in vitro transcriptional response of the synovial broblast (SF) to the stimulation with RA synovial uid. Using a reverse engineering approach, we have inferred the main transcriptional regulatory network that governs the response to this complex proin ammatory stimulus. We have then studied the genes in this regulatory network as risk factors for RA susceptibility using a casecontrol approach. We have analyzed the association of each gene with disease independently, but we have also analyzed the presence of higher order interactions associated with disease risk (i.e. epistasis) using the Multifactor Dimensionality Reduction method. In the second genomic approach, we have used whole genome genotyping microarrays targeting more than 300,000 SNPs (Single Nucleotide Polymorphisms) markers to perform a Genome-wide Association Study (GWAS) in RA. In order to increase the statistical power of our study we have implemented a liability-based design. We have subsequently validated those loci showing highest evidence of association using an independent replication cohort. Also, in order to integrate our ndings with the evidence of previous GWAS in RA, we have determined those genomic loci showing increased clustering of signals between studies. Finally, we have performed an exhaustive genome-wide analysis of the two-way epistatic interactions associated with RA applying parallel computation. Using the SF in vitro stimulation model we have identi ed n = 157 genes signi cantly associated with the response to RA proin ammatory stimulus. Within this set of di erentially expressed genes there are genes that have been clearly associated to RA pathophisiology but also new genes not previously linked to this disease. From the di erential expression data we have been able to identify a 13 gene Nuclear Factor kappa-Beta (NF-kB) transcriptional regulatory network, as the key transcriptional regulatory force in this RA SF model. Whilst several of the genes in the network showed nominal association to disease, we have identi ed a signi cant epistatic interaction between interleukin 6 (IL6 ) and interleukin 4 induced 1 (IL4I1 ) genes. In the GWAS approach we have identi ed several candidate genes for RA, advanced RA and chronic arthritis risk. Using an independent replication dataset we have found an intronic SNP in Kruppel-Like Factor 12 (KLF12) gene as the most strongly associated SNP with RA. The meta-analysis with previous GWAS results has also identi- ed several genomic regions -including KLF12 locus- that are likely to harbour new risk variants for RA. In the genome-wide epistasis analysis we have found a number of SNP pairs associated with RA with a signi cance close to the conservative multiple test correction threshold. Also, we have found that two-way interactions including the HLA region, the strongest main e ect in RA, are ranked secondarily to many other potentially interacting loci, thus suggesting a minor role for this locus in the epistatic susceptibility to disease. The two alternative genomic approaches we present in this work have identi ed a group of new loci which are likely to be associated with the risk to RA. This group of candidate loci should be now validated in independent populations to con rm their implication in RA susceptibility.