Susceptibility and Severity of COVID-19 Are Both Associated With Lower Overall Viral-Peptide Binding Repertoire of HLA Class I Molecules, Especially in Younger People

Abstract

An important number of studies have been conducted on the potential association between human leukocyte antigen (HLA) genes and COVID-19 susceptibility and severity since the beginning of the pandemic. However, case-control and peptide-binding prediction methods tended to provide inconsistent conclusions on risk and protective HLA alleles, whereas some researchers suggested the importance of considering the overall capacity of an individual's HLA Class I molecules to present SARS-CoV-2-derived peptides. To close the gap between these approaches, we explored the distributions of HLA-A, -B, -C, and -DRB1 1st-field alleles in 142 Iranian patients with COVID-19 and 143 ethnically matched healthy controls, and applied in silico predictions of bound viral peptides for each individual's HLA molecules. Frequency comparison revealed the possible predisposing roles of HLA-A*03, B*35, and DRB1*16 alleles and the protective effect of HLA-A*32, B*58, B*55, and DRB1*14 alleles in the viral infection. None of these results remained significant after multiple testing corrections, except HLA-A*03, and no allele was associated with severity, either. Compared to peptide repertoires of individual HLA molecules that are more likely population-specific, the overall coverage of virus-derived peptides by one's HLA Class I molecules seemed to be a more prominent factor associated with both COVID-19 susceptibility and severity, which was independent of affinity index and threshold chosen, especially for people under 60 years old. Our results highlight the effect of the binding capacity of different HLA Class I molecules as a whole, and the more essential role of HLA-A compared to HLA-B and -C genes in immune responses against SARS-CoV-2 infection.

Keywords: COVID-19; HLA; HLA binding prediction; SARS-CoV-2-derived peptides; overall binding repertoire.

Figure 1

Distribution comparison of allele frequencies for HLA-A (A), -B (B), and -DRB1 (C) loci between patients and controls. Noting that HLA alleles for each group are ranked by frequencies and represented by bars of different lengths with their names on top, each bar pair does not necessarily correspond to the same allele pair.

Figure 2

Odds ratio with 95% confident intervals for HLA alleles and HLA two-locus haplotypes. For haplotypes, only those with significant results are shown. Significance is represented by different symbols: crosses for non-significant results; circles for significant results before correction and dots for significant results after correction.

Figure 3

Density distributions of the numbers of SARS-CoV-2-derived peptides predicted by %Rank as weak binders to HLA-A (n_A), HLA-B (n_B), HLA-A and -B (n_AB), and HLA-DR (n_DR) molecules in patients (solid curves) and controls (dashed curves) of all ages and in those under 60 years old, respectively.

Figure 4

Density distributions of the numbers of SARS-CoV-2-derived peptides predicted by %Rank as as weak binders to HLA-A (n_A), HLA-B (n_B), HLA-A and -B (n_AB), HLA-C (n_C), HLA-A, -B and -C (n_ABC), and HLA-DR (n_DR) molecules (A) in three subgroups of patients (Mild/Moderate: solid curves in yellow; Severe: solid curves in red; Critical: solid curves in purple) and controls (dashed curves in black) and (B) in two subgroups of patients (Mild/Moderate: solid curves in yellow; Severe/Critical: solid curves in rose) and controls (dashed curves in black) of all ages and in those under 60 years old, respectively. Note that n_C and n_ABC data were not available for controls.