Predicted B Cell Epitopes Highlight the Potential for COVID-19 to Drive Self-Reactive Immunity

Abstract

COVID-19, caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), whilst commonly characterised as a respiratory disease, is reported to have extrapulmonary manifestations in multiple organs. Extrapulmonary involvement in COVID-19 includes autoimmune-like diseases such as Guillain-Barré syndrome and Kawasaki disease, as well as the presence of various autoantibodies including those associated with autoimmune diseases such a systemic lupus erythematosus (e.g. ANA, anti-La). Multiple strains of SARS-CoV-2 have emerged globally, some of which are found to be associated with increased transmissibility and severe disease. We performed an unbiased comprehensive mapping of the potential for cross-reactivity with self-antigens across multiple SARS-CoV-2 proteins and compared identified immunogenic regions across multiples strains. Using the Immune Epitope Database (IEDB) B cell epitope prediction tool, regions predicted as antibody epitopes with high prediction scores were selected. Epitope sequences were then blasted to eight other global strains to identify mutations within these regions. Of the 15 sequences compared, eight had a mutation in at least one other global strain. Predicted epitopes were then compared to human proteins using the NCBI blast tool. In contrast to studies focusing on short sequences of peptide identity, we have taken an immunological approach to selection criteria for further analysis and have identified 136 alignments of 6-23 amino acids (aa) in 129 human proteins that are immunologically likely to be cross-reactive with SARS-CoV-2. Additionally, to identify regions with significant potential to interfere with host cell function-or promote immunopathology, we identified epitope regions more likely to be accessible to pathogenic autoantibodies in the host, selected using a novel combination of sequence similarity, and modelling protein and alignment localization with a focus on extracellular regions. Our analysis identified 11 new predicted B-cell epitopes in host proteins, potentially capable of explaining key aspects of COVID-19 extrapulmonary pathology, and which were missed in other in silico studies which used direct identity rather than immunologically related functional criteria.

Keywords: COVID-19; autoimmunity; epitope mapping; molecular-mimicry; peptides; self-reactivity.

FIGURE 1

Research pipeline of the project to explore the potential of immune cross-reactivity. Protein sequences for select SARS-CoV-2 proteins were obtained for epitope predictions (highlighted green. Spike: Surface glycoprotein. E, envelope; M, membrane; NP, nucleoprotein). Using the Immune Epitope Database (IEDB) epitope prediction tool, B cell epitopes were predicted and those selected were screened against human proteins using the NCBI Blastp tool. Human proteins with sequence similarities to the SARS-CoV-2 epitopes were investigated for their function, disease association and alignment localization. Potential immune cross-reactivity between SARS-CoV-2 and human alignments was then explored by applying specific selection criteria. Alignments of interest from the human proteins were explored as potential epitopes within the human protein sequences.

FIGURE 2

Linear schematic of selected B cell epitopes in SARS-CoV-2 Structural proteins. B cell epitopes within the SARS-CoV-2 spike (SP), membrane (M) and nucleoprotein (NP) were mapped and selected based as having ≥6aa length and a predicted epitope score ≥1. (A) Five epitopes in the spike protein, beginning at positions aa249, 597, 675, 805 and 1256 were selected for downstream analysis. (B) One epitope in the membrane protein at position aa205 was identified. (C) Nine epitopes in the nucleoprotein, aa1, 58, 115, 164, 232, 273, 338, 361, 408 were identified. All associated sequences can be found in Supplementary Table S3.

FIGURE 3

Predicted epitopes mapped to protein structure of Surface Glycoprotein and Nucleoprotein. (A) X-ray crystal structure of the Surface Glycoprotein (Cai et al., 2020) coloured via spectrum (N to C terminal) with the structured B cell epitopes highlighted via spheres; aa294-261 (cyan), aa597-606 (green), aa675-687 (yellow) and aa805-816 (orange). Also labelled are the mutation sites; L249, D253, Q677, P681 and R682. (B) Structure of the nucleoprotein coloured via spectrum (N to C terminal). The N-terminal regions are from the X-ray crystal structure (PDB code: 6vyo) followed by an unstructured linear (yellow line) to a C-terminal homology model. The structural B cell epitopes are highlighted via spheres; aa58-85 (cyan), aa115-127 (green), aa164-216 (yellow), aa232-269 (orange) and aa273-287 (salmon). Also labelled is the location of known mutation sites; P80, P199, R201, S202, R203, T205, G214, N234 and S235.

FIGURE 4

Linear schematic of selected B cell epitopes in SARS-CoV-2 Orf proteins. B cell epitopes within the SARS-CoV-2 Orf3a, Orf3b, Orf7a and Orf8 and selected based as having ≥6aa length and a predicted epitope score ≥0.55. (A) Three epitopes in Orf3a (aa172-197, 216-225 and 238-272) and one epitope in Orf3b (aa9-28) were identified. (B) Three epitopes in Orf7a (aa17-25, 33-51 and 71-96) were identified. (C) Three epitopes in Orf8 (aa23-45, 48-56 and 63-78). All associated sequences can be found in Supplementary Table S3.

FIGURE 5

Predicted epitopes mapped to the structures of SARS-CoV-2 Orf proteins. (A) Cryo-EM structure of the Orb3a (Kern et al., 2021) coloured via spectrum (N to C terminal) with the structured B cell epitopes highlighted via spheres; aa172-197 (orange), aa216-225 (light red), aa238-272 (red). (B) X-ray crystal structure of Orb7a (Zhou et al., 2021) with the structured B cell epitopes highlighted via spheres; aa17-25 (blue), 33-51 (green) and aa71-96 (red). (C) X-ray crystal structure of Orb7a (Flower et al., 2021) with the structured B cell epitopes highlighted via spheres; aa 23-45 (blue), 1148-56 (cyan) and aa63-78 (light green).

FIGURE 6

Examples of sequence comparison between predicted B cell epitopes and key SARS-CoV-2 variants. Structural protein sequences for a strain from Wuhan (WIV04/2019), VOCs (Alpha, Beta, Gamma and Delta) and VOIs (Eta, Iota, Kappa and Lambda) were obtained from the GISAID database and blast aligned with predicted B cell epitopes. Quality refers to the alignment quality based on blosum2 algorithm scores, Consensus indicates the abundance of the amino acids present in a particular position and Occupancy is the number of aligned positions. (A) SP VOC sequence alignment at position aa675-687. (B) VOC sequence alignment of membrane aa206-215. (C) NP VOC sequence alignment at position aa164-216. (D) SP VOI sequence alignment at position aa675-687. (E) VOI sequence alignment of membrane aa206-215. (F) NP VOI sequence alignment at position aa164-216.

FIGURE 7

Workflow to identify human proteins that share sequence similarities with SARS-CoV-2 immunogenic regions. The predicted SARS-CoV-2 B cell epitopes were compared to the human proteome using the NCBI protein BLAST tool. Two epitopes, NP1-51 and NP164-216 had no sequence similarities to human proteins. The top 100 sequence alignments from the remaining 23 epitopes were narrowed by removing duplicates (alternative nomenclature/sequence IDs), uncharacterized/hypothetical proteins and the variable regions of lymphocyte receptors. This resulted in a final list of 281 sequence alignments that was comprised of 256 human proteins.

FIGURE 8

Investigation of human proteins and association with diseases. Using the UniProtKB, GeneCards and Pubmed online resources, research into expression location, protein function and association with diseases was performed on all proteins listed in Supplementary Table S5 (n = 256). This resulted in a further exclusion of proteins (listed in Supplementary Table S6), which were computationally predicted or unreviewed protein sequences, leaving 246 sequence alignments comprising of 223 proteins (Supplementary Table S7). Two proteins have been reported to be associated with COVID-19, 144 have an association with diseases (Supplementary Table S8) and 50 with autoimmunity (Supplementary Table S10).

FIGURE 9

Overlap of proteins between body systems. Proteins found to be associated with diseases were grouped based on body system location of the diseases. Key systems with known complications in COVID-19 disease were found to have overlapping protein associations.

FIGURE 10

Proteins found to share alignments with SARS-CoV-2 epitopes have associations with various autoimmune diseases. Proteins associated with autoimmune diseases were grouped based on the specific disease or diseases (human and animal model combined) they are found to be associated with Systemic lupus erythematosus (SLE), multiple sclerosis (MS), rheumatoid arthritis (RA) and diabetes had the most protein associations. Proteins were also found to be associated with other autoimmune diseases to a lesser extent.

FIGURE 11

Alignments of extracellular proteins of interest mapped to protein structures. Homology models of human proteins of interest were downloaded from the SWISS-MODEL Repository and colored rainbow from N to C termini. The alignment motifs similar to SARS-CoV-2 B cell epitopes are highlighted via spheres (Bienert et al., 2017) (A) C-C motif chemokine 22 (CCL22). Model coverage aa25-91, alignment location aa23-30. (B) Lysozyme-like protein-1 and -2. Model coverage aa20-146, alignment location aa 16-24. (C) Reelin. Model coverage aa1251-1947, alignment location 1719-1724. (D) Alpha-internexin. Model coverage aa90-302, alignment location aa226-238.