Germline variants and mosaic chromosomal alterations affect COVID-19 vaccine immunogenicity

Abstract

Vaccine immunogenicity is influenced by the vaccinee's genetic background. Here, we perform a genome-wide association study of vaccine-induced SARS-CoV-2-specific immunoglobulin G (IgG) antibody titers and T cell immune responses in 1,559 mRNA-1273 and 537 BNT162b2 vaccinees of Japanese ancestry. SARS-CoV-2-specific antibody titers are associated with the immunoglobulin heavy chain (IGH) and major histocompatibility complex (MHC) locus, and T cell responses are associated with MHC. The lead variants at IGH contain a population-specific missense variant (rs1043109-C; p.Leu192Val) in the immunoglobulin heavy constant gamma 1 gene (IGHG1), with a strong decreasing effect (β = -0.54). Antibody-titer-associated variants modulate circulating immune regulatory proteins (e.g., LILRB4 and FCRL6). Age-related hematopoietic expanded mosaic chromosomal alterations (mCAs) affecting MHC and IGH also impair antibody production. MHC-/IGH-affecting mCAs confer infectious and immune disease risk, including sepsis and Graves' disease. Impacts of expanded mosaic loss of chromosomes X/Y on these phenotypes were examined. Altogether, both germline and somatic mutations contribute to adaptive immunity functions.

Keywords: COVID-19 vaccine; Graves’ disease; clonal hematopoiesis; genome-wide association study; immunogenicity; immunoglobulin heavy chain; major histocompatibility complex; mosaic chromosomal alteration; mosaic loss of chromosome X; mosaic loss of chromosome Y; sepsis.

Graphical abstract

Figure 1

Overview of the study We enrolled 1,673 individuals receiving the mRNA-1273 COVID-19 mRNA vaccine (Moderna) and 572 individuals receiving the BNT162b2 COVID-19 mRNA vaccine (Pfizer), both of Japanese ancestry. By measuring COVID-19 vaccine-induced IgG antibody titers (humoral immunity) and T cell response levels (cellular immunity) using multiple assay kits at multiple time points, we performed genome-wide association studies (GWASs) and HLA fine-mapping to identify germline mutations affecting vaccine immunogenicity. We compared associated loci in the Japanese cohorts with those in the UK Biobank dataset. Finally, we demonstrated that hematopoietic mosaic chromosomal alterations affecting the genetic loci implicated in GWASs contribute to aging-related vaccine efficacy impairment.

Figure 2

Genome-wide association study of COVID-19 vaccine immunogenicity (A) Manhattan plot for the GWAS meta-analysis of the IgG antibody titer measured by assay kit 1. (B) Manhattan plot for the GWAS of the IgG antibody titer measured by assay kit 2. (C) The lead variants in IGHG1 were validated by PacBio HiFi long-read sequencing on the three samples selected from the imputation reference panel. A 64 bp window around the three lead variants (14:106208306:A:G, 14:106208326:G:C, and 14:106208327:G:A) is highlighted as the pink dashed box and enlarged. (D) IGHG1 amino acid position p.Leu192Val is highlighted as orange-colored spheres. The protein structure of the human IgG is based on Protein Data Bank entry PDB: 1HZH and was prepared using UCSF Chimera (version 1.16). (E) Forest plots of the lead variant associations with the IgG antibody titer measured by assay kit 1. Blue markers indicate p < 0.05. (F) Forest plots of the lead variant associations with the influenza antibody titers. Blue markers indicate p < 0.05. A/H1, influenza A virus subtype H1N1; A/H3, influenza A virus subtype H3N2; B/Vi, influenza virus subtype B/Victoria; B/Ya, influenza virus subtype B/Yamagata. (G) Manhattan plot for the GWAS of the IgG serostatus in the UKB. (H) Manhattan plot for the GWAS meta-analysis of the T cell response level against antigen 1. (I) Manhattan plot for the GWAS meta-analysis of the T cell response level against antigen 2. The pink horizontal dashed lines in the Manhattan plots denote the genome-wide significance threshold of p = 5.0 × 10⁻⁸. Error bars in the forest plots denote 95% confidence intervals. Quantile-quantile plots for the GWASs are shown in Figure S13. vac., vaccination.

Figure 3

Regional associations of HLA variants with COVID-19 vaccine immunogenicity (A) Stepwise association analysis of the imputed HLA variants in the MHC region with the IgG antibody titer. (B) Stepwise association analysis of the imputed amino acid polymorphisms in HLA-DRB1 with the IgG antibody titer. (C) Stepwise association analysis of the imputed HLA variants in the MHC region with the T cell response level against antigen 1. (D and E) Stepwise association analysis of the imputed amino acid polymorphisms in HLA-DRB1 (D) and HLA-DPB1 (E) with the T cell response level against antigen 1. Each diamond represents the −log₁₀(p) of the variants, including the single-nucleotide variants; two-, four-, and six-digit HLA alleles; and amino acid polymorphisms of the HLA genes in (A) and (C). Each diamond represents the −log₁₀(p_omnibus) of the amino acid polymorphisms of the HLA genes in (B), (D), and (E). The pink horizontal dashed lines denote the genome-wide significance threshold of p = 5.0 × 10⁻⁸. The most strongly associated amino acid polymorphisms and HLA classical alleles are labeled when their associations showed p < 5.0 × 10⁻⁸. Detailed association statistics are available in Table S6.

Figure 4

Vaccine immunogenicity-associated alleles modulate blood proteome (A) Schematic view of the proteogenomic analysis. (B–F) Volcano plots representing the associated protein expression in the blood with IgG antibody titer HLA-GRS (B), T cell response HLA-GRS (C), IgG antibody titer-associated variant at IGH found in the Japanese cohorts (D), and IgG serostatus-associated variant at IGH (E) and 16p11 (F) found in the UKB. The x axis and y axis denote the effect sizes and −log₁₀(p) values in the meta-analysis, respectively. The horizontal dashed lines denote the Bonferroni-corrected significance threshold of p = 0.05/2,943 = 1.7 × 10⁻⁵. The proteins satisfying the Bonferroni-corrected significance threshold are labeled. trans-associated proteins are highlighted in pink, defined as those encoded outside MHC (B) and (C) or outside the 2 Mb window centered around the tested variant (D–F).

Figure 5

Associations of hematopoietic expanded mCAs at MHC/IGH with immune dysregulation (A) Quantile-quantile plot of the p values of the autosomal mCA class associations with the COVID-19 IgG antibody serostatus in the UKB. The x axis denotes −log₁₀(p) expected from a uniform distribution. The y axis denotes the observed −log₁₀(p). The top two strongly associated mCA classes are labeled. (B) Forest plot of the MHC/IGH-affecting mCA associations with the COVID-19 IgG antibody serostatus in the UKB. Blue markers indicate p < 0.05. Daggers (†) indicate that there were no seropositive mCA carriers. (C) Forest plots of the MHC/IGH-affecting mCA associations with the infectious or immune diseases in the BBJ and UKB in all age groups or the elderly group (≥65 years of age). Association test results are shown where at least two cases carrying the mCA class were found. Error bars in the forest plots denote 95% confidence intervals. Blue markers indicate p < 0.05. +, gain; −, loss; =, copy-neutral loss of heterozygosity.

Figure 6

Associations of hematopoietic expanded mLOX/mLOY with immune dysregulation (A) Forest plot of the expanded mLOX/mLOY associations with the COVID-19 IgG antibody serostatus in the UKB. Blue markers indicate p < 0.05. (B) Forest plots of the expanded mLOX/mLOY associations with the infectious or immune diseases in the BBJ and UKB in all age groups or the elderly group (≥65 years of age). Association test results are shown where at least two cases carrying the mCA class were found. Error bars in the forest plots denote 95% confidence intervals. Blue markers indicate p < 0.05. cf, cell fraction.