Introduction
Psoriasis vulgaris (PsV [MIM 177900]) is a common immune-mediated skin disease characterized by epidermal hyperplasia, inflammatory cell infiltration, and vascular remodeling.1,2 Approximately one-third of PsV-affected individuals develop a related inflammatory musculoskeletal condition called psoriatic arthritis (PsA), which is considered more severe than the other more common PsV subtype, cutaneous psoriasis (PsC).3 Previous linkage and association studies have mapped PsV risk to a critical region spanning ∼300 kb within the major histocompatibility complex (MHC) class I region on 6p21 (this region is termed PSORS1).4 Subsequent analyses have identified HLA-Cw6 (MIM 142840) as the risk allele of PSORS1.5,6 Although recent large-scale genome-wide association studies (GWASs) have identified more than 35 genetic risk loci for PsV outside the MHC region, HLA-C∗06:02 (the most frequent four-digit allele equivalent to HLA-Cw6) has consistently demonstrated the strongest association with PsV risk.7–12
However, the effects of the genetic architecture of the MHC region on PsV risk have yet to be fully elucidated. Previous studies have suggested the existence of other risk variants in addition to HLA-C∗06:02 in class I human leukocyte antigen (HLA) genes (HLA-A [MIM 142800] or HLA-B [MIM 142830]) and class II HLA genes (HLA-DRB1 [MIM 142857], HLA-DQA1 [MIM 146880], HLA-DQB1 [MIM 604305], HLA-DPA1 [MIM 142880], or HLA-DPB1 [MIM 142858]).13–18 Investigators have also studied polymorphisms of MHC class I polypeptide-related sequence A (MICA [MIM 600169]), an HLA-like gene that does not present antigen.17,19,20 However, strong and complex linkage-disequilibrium (LD) patterns in the MHC region21,22 have thus far challenged the identification of independent risk signals. Moreover, analyses focusing on the two major subsets of PsV have identified different effect sizes of associated risk alleles (including HLA-C∗06:02) between PsA and PsC, suggesting heterogeneous effects of HLA alleles on the two major subphenotypes.13–16,23
Recently, we devised an analytical approach to fine map risk of the variants in the MHC region by imputing genotypes of amino acid polymorphisms in the classical HLA genes and classical two- and four-digit alleles.24–26 Each classical four-digit HLA allele corresponds to a unique amino acid sequence encoded by the HLA gene, and thus polymorphic residues at each amino acid position could also be targets of disease-risk association studies. This approach has been used for successfully fine mapping HLA alleles of several autoimmune or immune-mediated diseases, including durable host control of HIV infection (MIM 609423), seropositive and seronegative rheumatoid arthritis (MIM 180300), myasthenia gravis (MIM 254200), and follicular lymphoma (MIM 613024).25–29 In certain instances, our approach has been able to pinpoint individual amino acid sites that might account for disease risk within HLA molecules.25–27
In this study, we aimed to fully characterize the genetic architecture of the MHC region for PsV. Our goals were (1) to define the set of risk alleles for PsV at the four-digit HLA allele and amino acid resolutions, (2) to examine how the role of MICA in PsV compares to that of other HLA genes, and (3) to identify a genetic marker that distinguishes the risk of two subtypes, PsA and PsC. To this end, we applied our HLA-variant imputation approach to large-scale PsV GWASs and Immunochip studies comprising 9,247 affected individuals and 13,589 control individuals of European ancestry. We also expanded our approach to impute MICA alleles and MICA amino acid polymorphisms by constructing a MICA imputation reference panel. With the imputed MHC sequence variations, including classical HLA genes and MICA, we fine mapped the MHC associations with overall PsV risk and specifically focused on risk comparisons between the PsA and PsC subphenotypes.