MICBG406A polymorphism reduces risk of mechanical ventilation and death during viral acute lung injury

Abstract

MHC class I polypeptide-related sequence B (MICB) is a ligand for NKG2D. We have shown NK cells are central to lung transplant acute lung injury (ALI) via NKG2D activation, and increased MICB in bronchoalveolar lavage predicts ALI severity. Separately, we found a MICB polymorphism (MICBG406A) is associated with decreased ALI risk. We hypothesized this polymorphism would protect against severe SARS-CoV-2 respiratory disease. We analyzed 1,036 patients hospitalized with SARS-CoV-2 infection from IMPACC. Associations between MICBG406A and outcomes were determined by linear regression or Cox proportional hazards models. We also measured immune profiles of peripheral blood and the upper and lower airway. We identified 560 major allele homozygous patients, and 426 and 50 with 1 or 2 copies of the variant allele, respectively. MICBG406A conferred reduced odds of severe COVID-19. MICBG406A homozygous participants demonstrated 34% reduced cumulative odds for mechanical ventilation or death and 43% reduced risk for mortality. Patients with MICBG406A variant alleles had reduced soluble inflammatory mediators and differential regulation of multiple immune pathways. These findings demonstrate a potentially novel association between increasing MICBG406A variant allele copies and reduced COVID-19 severity, independent of SARS-CoV-2 viral burden and humoral immunity, suggesting the NKG2D-ligand pathway as an intervention target.

Keywords: COVID-19; Immunology; Innate immunity; NK cells; Pulmonology.

Figure 1. Schematic for the study and the nested subsets of assays available for immunophenotyping.

Figure 2. COVID-19 morbidity and mortality by MICB^G406A polymorphism.

(A–D) Frequency of severe COVID-19 (trajectory group 4 or 5) (A), need for mechanical ventilation or death by 28-days post-admission (ordinal respiratory score ≥ 6) (B), death by 28-days post-admission (C), and death (D) at any point during the study by number of copies of the MICB^G406A variant allele were compared by binomial generalized logistic regression. (E and F) Survival analysis of death by 28 days after admission (E) or ever during the study (F) and number of copies of the MICB^G406A variant allele were compared by Cox proportional hazards models. Results for patients with no copies of the variant allele (GG) are shown in blue, 1 copy of the variant allele (GA) in orange, and 2 copies of the variant allele (AA) in magenta.

Figure 3. Nasal SARS-CoV-2 viral load and serum anti–Spike IgG by MICB^G406A polymorphism.

(A and B) Visit 1 (A) nasal SARS-CoV-2 viral load (N1 CT value) (A) and serum anti–spike IgG levels (AUC) (B) by number of copies of the MICB^G406A variant allele were compared by linear regression. Results for patients with no copies of the variant allele (GG) are shown in blue, 1 copy of the variant allele (GA) in orange, and 2 copies of the variant allele (AA) in magenta.

Figure 4. MICB gene expression in PBMC and the upper airway by MICB^G406A polymorphism.

(A and B) Visit 1 MICB gene expression in PBMC (A) and nasal transcriptomics (B) by number of copies of the MICB^G406A variant allele was compared by linear regression. (C) Blood NK cell frequency at visit 1 by number of copies of the MICB^G406A variant allele were compared by linear regression. Results for patients with no copies of the variant allele (GG) are shown in blue, 1 copy of the variant allele (GA) in orange, and 2 copies of the variant allele (AA) in magenta.

Figure 5. Serum proteins by MICB^G406Apolymorphism.

Visit 1 serum protein levels by number of copies of the MICB^G406A variant allele were compared by linear regression. (A and B) Z-scaled serum levels by MICB^G406A genotype for all proteins with FDR ≤ 0.05 (A) and unscaled serum levels for key proteins (B) are shown. Analytes significantly associated with MICB^G406A genotype after adjusting for baseline disease severity are indicated by asterisks in A. Results for patients with no copies of the variant allele (GG) are shown in blue, 1 copy of the variant allele (GA) in orange, and 2 copies of the variant allele (AA) in magenta.

Figure 6. PBMC and upper airway gene expression by MICB^G406 polymorphism.

Visit 1 PBMC gene expression by number of copies of the MICB^G406A variant allele were compared by linear regression with the limma package. (A) Log-transformed fold-change with increasing copies of the variant allele against P values corrected for multiple comparisons is shown. (B and C) Z-scaled expression by MICB^G406A genotype for the top downregulated (B) and upregulated (C) genes is shown. (D) Log-transformed fold-change for all complement factors is shown with FDR-corrected P values. Visit 1 upper airway gene expression by number of copies of the MICB^G406A variant allele were compared by linear regression with the limma package. (E) Log-transformed fold-change with increasing copies of the variant allele against P value corrected for multiple comparisons is shown. (F and G) Z-scaled expression by MICB^G406A genotype for the top downregulated (F) and upregulated (G) genes is shown. (H) Log-transformed fold-change for all complement factors is shown with FDR-corrected P-values. Results for patients with no copies of the variant allele (GG) are shown in blue, 1 copy of the variant allele (GA) in orange, and 2 copies of the variant allele (AA) in magenta. In A, D, E, and H, blue and magenta indicate genes downregulated and upregulated with increasing copies of the MICB^G406A variant allele, respectively.

Figure 7. Hyperinflammatory subtype is associated with severe disease and decreasing copies of the MICB^G406A variant allele.

(A) Odds of hyperinflamm atory subtype (Class B) by trajectory group was compared by binomial generalized logistic regression. (B) Survival analysis of death by 28 days after admission and inflammatory subtype was compared by Cox proportional hazards models. (C) Odds of Class B hyperinflammatory subtype by number of copies of the MICB^G406A variant allele was compared by binomial generalized logistic regression. (D) Survival analysis of death by 28 days after admission and number of copies of the MICB^G406A variant allele in patients not displaying the Class B hyperinflammatory subtype was compared by Cox proportional hazards models. (E) Results for patients with no copies of the variant allele (GG) are shown in blue, 1 copy of the variant allele (GA) in orange, and 2 copies of the variant allele (AA) in magenta.