Study of susceptibility for systemic lupus erythematosus : from sex differences to the polymorphic human leukocyte antigen region

Abstract

Systemic lupus erythematosus (SLE) is a prototype autoimmune disease with a strong sex bias in disease prevalence and a strong genetic component in its pathogenesis. To date, genome-wide association studies (GWAS) have identified more than 80 SLE-associated loci, with the strongest signal identified in the human leukocyte antigen (HLA) region. However, these susceptibility loci taken together only explain less than 30% of disease heritability, suggesting that there are many more loci to be identified. In this study, I investigated sexual dimorphism at the gene regulation level and investigated the genetic signals in the HLA region in order to better understand the genetic basis of this disease. In order to study sexual dimorphism, I studied genes with a sex-biased gene expression in lymphoblastoid cell lines, along with its evolutionary and regulatory signatures. I also looked for genotypes that are differentially associated with gene expression between males and females, which I termed sex interacting expression quantitative trait loci (ss-eQTLs). I characterized differential gene expression between sexes in 462 RNA-seq samples and identified 587 sex-biased genes (sDEGs), or 3.2% of the genes surveyed. Using the same dataset, I also uncovered 21 mRNA and 3 miRNA ss-eQTLs. Some of these ss-eQTLs are in high LD with traits that have a sex bias, showing a potential link between variants, genes and sexually dimorphic traits. Genome-wide, nine transcription factors were found enriched in binding to the regions surrounding the transcription start sites of female-biased genes. Many pathways and protein domains were enriched in sex-biased genes, some of which hint at sex-biased physiological processes. To date, the most prominent association signal in SLE has been found in the HLA region. In order to understand HLA association with SLE, investigation at both the SNP and HLA allele level can lend further insight. Using three Han Chinese cohorts from Hong Kong, Anhui and Guangzhou, totaling 3911 case and 4247 controls, I performed genome-wide association analysis with SLE on the HLA region. At the SNP level, I found 614 signals that passed the genome wide threshold for significance (p_value < 1e-8). A third of the top signals (241 SNPs) reside in the DRB1, DQB1, DQA1 region. Other genic region that was strongly implicated was LOC101929163 and ATF6B. At the HLA allele level, the accuracy of HLA allele imputation from SNP genotypes is heavily dependent on the reference panels used. I used the machine learning algorithm HIBAG to generate a new Han Chinese reference panel derived from 10,689 samples. Using this Han Chinese reference panel, I then imputed HLA alleles across 8 loci (A, B, C, DQA1, DPA1, DQB1, DRB1) for the three SLE GWAS cohorts and investigated their associations to SLE. After meta-analysis, 11 HLA alleles were found significantly associated with disease status at genome-wide significance. The topmost signals reside in the DQB1 and DRB1 loci. Interestingly, class I loci, HLA-B and HLA-C, both showed signals of association with SLE at B*15:02 and C*07:02, which was not known in previous studies.