Integrated Immunopeptidomic and Proteomic Analysis of COVID-19 lung biopsies

Abstract

Introduction: Severe respiratory illness is the most prominent manifestation of patients infected with SARS-CoV-2, and yet the molecular mechanisms underlying severe lung disease in COVID-19 affected patients still require elucidation. Human leukocyte antigen class I (HLA-I) expression is crucial for antigen presentation and the host's response to SARS-CoV-2.

Methods: To gain insights into the immune response and molecular pathways involved in severe lung disease, we performed immunopeptidomic and proteomic analyses of lung tissues recovered at four COVID-19 autopsy and six non-COVID-19 transplants.

Results: We found signals of tissue injury and regeneration in lung fibroblast and alveolar type I/II cells, resulting in the production of highly immunogenic self-antigens within the lungs of COVID-19 patients. We also identified immune activation of the M2c macrophage as the primary source of HLA-I presentation and immunogenicity in this context. Additionally, we identified 28 lung signatures that can serve as early plasma markers for predicting infection and severe COVID-19 disease. These protein signatures were predominantly expressed in macrophages and epithelial cells and were associated with complement and coagulation cascades.

Discussion: Our findings emphasize the significant role of macrophage-mediated immunity in the development of severe lung disease in COVID-19 patients.

Keywords: COVID-19 -; HLA-I; immunopeptidome; macrophage; proteomics.

Figure 1

HLA-I immunopeptidome in lethal COVID-19 lung and controls. (A) Schematic of the experimental workflow. HLA-I proteins were immunoprecipitated (IP) from lung tissue lysates and HLA-bound peptides were identified by liquid chromatography-tandem MS (LC-MS/MS). Subsequently, the IP flowthrough was further analyzed by tandem mass tags (TMT) quantitative MS to measure global protein expression. HLA-I binding peptides and total proteins were identified and quantified in COVID-19 lung autopsy specimens (n=4) and Control lungs (n=6). Controls were lung tissues from patients undergoing transplantation for end stage lung disease. (B) The number of HLA-I peptides identified in each sample. (C) Fraction of observed peptides assigned to different HLA-I alleles using HLAthena prediction. (D) Venn diagram showing the common and unique peptides predicted to be associated with the same HLA allele in different samples. The Motifs of 9-mer peptide sequences were also shown.

Figure 2

Proteins with differential HLA-I presentation in lethal COVID-19 lung versus Controls. (A) Volcano plot comparing HLA-I peptide presentation between lethal COVID-19 lung and Controls. The most abundant peptide was used to represent HLA presentation of the corresponding protein in each sample. MS intensity of HLA-I peptides was log-transformed and differentially presented proteins were determined by MaxQuant (p<0.05, Fold-change >2). (B) Schematic of the rank product-based method to evaluate HLA presentation. Proteins had been divided into 10 bins in each sample based on their rank in the abundance of the HLA peptide (e.g., top 10%, top 20%). A rank product was then generated including all samples in the group (e.g., COVID-19). (C) Mapping of the significantly up- or downregulated proteins identified in (A) into the plot of rank products. (D) Heatmap of HLA peptide abundance of proteins with differential HLA-I presentation in lethal COVID-19 lung versus Controls. (E) The enrichment of proteins with differential HLA-I presentation in KEGG pathways, MsigDB gene sets and Cell Atlas markers. The top five categories were shown, also see Tables S5-7 .

Figure 3

Differential protein expression in lethal COVID-19 lung versus Controls. (A) Histogram showing distribution of protein expression levels across different samples. (B) Box plot showing distribution of protein expression levels across different samples. The box shows the quartiles, the bar indicates median, and the whiskers show the distribution. (C) Volcano plot comparing protein expression between lethal COVID-19 lung and Controls. Foldchange>2 and P<0.05 was considered significant. (D) Heatmap of differentially expressed proteins in lethal COVID-19 lung versus Controls. (E) The enrichment of proteins with differential HLA-I presentation in KEGG pathways, MsigDB gene sets and Cell Atlas markers. The top five categories were shown, also see Tables S9-11 .

Figure 4

Cell subpopulations associated with HLA-I and protein signatures. (A) UMAP projection highlighting major parenchymal, endothelial and immune cell subsets in Broad lung atlas and (B) Columbia lung atlas. (C) Intensity map of mean expression of genes with significantly upregulated HLA-I presentation or protein expression in Broad lung atlas or (D) Columbia lung atlas. Gene expression was calculated by log(1+TP10K). TP10K, transcript per 10000 reads.

Figure 5

Lung proteins as early plasma markers for viral infection. (A) Volcano plot of differentially expressed plasma proteins based on mean normalized protein expression (NPX) values between COVID-19(+) (n=206) and COVID-19(−) patients (n=78). Only signature proteins with consistent changes in COVID-19 lung are shown. Δ NPX: COVID19(+) – COVID19(−). (B) Heatmap of plasma expression of upregulated protein signatures between COVID-19(+) versus COVID-19(−) patients in (A). (C) ROC curve analysis of protein signatures predicting viral infection. Proteins with area under the ROC Curve (AUC)>0.8 are shown. (D) Prediction powers of different markers and their combination. (E) Intensity map of signature genes expression in the COVID-19 lung UMAP.