Human leukocyte antigen (HLA) is a key genetic factor conferring risk of
systemic lupus erythematosus (SLE), but precise independent localization of HLA effects is extremely challenging. As a result, the contribution of specific HLA alleles and
amino-acid residues to the overall risk of SLE and to risk of specific
autoantibodies are far from completely understood. Here, we dissected (a) overall SLE association signals across
HLA, (b) HLA-
peptide interaction, and (c) residue-
autoantibody association. Classical alleles, SNPs, and
amino-acid residues of eight HLA genes were imputed across 4,915 SLE cases and 13,513 controls from Eastern Asia. We performed association followed by conditional analysis across HLA, assessing both overall SLE risk and risk of
autoantibody production. DR15 alleles
HLA-DRB1*15:01 (P = 1.4x10-27, odds ratio (OR) = 1.57) and
HLA-DQB1*06:02 (P = 7.4x10-23, OR = 1.55) formed the most significant haplotype (OR = 2.33). Conditioned
protein-residue signals were stronger than allele signals and mapped predominantly to
HLA-DRB1 residue 13 (
P = 2.2x10-75) and its proxy position 11 (P = 1.1x10-67), followed by HLA-DRB1-37 (P = 4.5x10-24). After conditioning on
HLA-DRB1, novel associations at HLA-A-70 (P = 1.4x10-8), HLA-DPB1-35 (P = 9.0x10-16), HLA-DQB1-37 (
P = 2.7x10-14), and HLA-B-9 (P = 6.5x10-15) emerged. Together, these seven residues increased the proportion of explained heritability due to HLA to 2.6%. Risk residues for both overall disease and hallmark
autoantibodies (i.e., nRNP: DRB1-11,
P = 2.0x10-14; DRB1-13,
P = 2.9x10-13; DRB1-30, P = 3.9x10-14) localized to the
peptide-binding groove of
HLA-DRB1. Enrichment for specific
amino-acid characteristics in the
peptide-binding groove correlated with overall SLE risk and with
autoantibody presence. Risk residues were in primarily negatively charged side-chains, in contrast with
rheumatoid arthritis. We identified novel SLE signals in HLA Class I loci (
HLA-A,
HLA-B), and localized primary Class II signals to five residues in
HLA-DRB1,
HLA-DPB1, and
HLA-DQB1. These findings provide insights about the mechanisms by which the risk residues interact with each other to produce
autoantibodies and are involved in SLE pathophysiology.